Dave Jago, Mark Sandler, and Alicia Wise
Release 5 of the COUNTER Code of Practice became effective in January 2019 and aimed to provide consistency, simplicity, flexibility, and clarity. COUNTER commissioned us to undertake an independent review to inform the COUNTER Executive and Technical Advisory Group (TAG) by answering the following questions:
We talked with Board and Technical Advisory Group members, analysed query logs and audit reports, and conducted an online survey of content providers, intermediaries, and libraries along with follow-up with interviews and focus groups across these stakeholder groups. We engaged both with organisations who have and have not—or have not yet—implemented COUNTER and/or Release 5. We worked in a focused and systematic way to try to solicit a broad global understanding. This was made especially challenging due to the global pandemic, and responses from places at or nearing their peak during the time the data gathering took place – for example India and South America – are limited.
We discovered that the COUNTER community – both providers and users of reports – are engaged, thoughtful, helpful, and generous with their time. It was clear that there is a great deal of goodwill and support for COUNTER, and that the need for it is fully understood. It was also clear that COUNTER’s Board, Executive Committee and Technical Advisory Board are well-informed and fully engaged and represent their various constituencies well. The principal issues and concerns that have come up in the study were raised both by the COUNTER insiders we spoke with and by the community. We therefore don’t expect that much of the report will be found surprising, but we hope it will be helpful to have the issues laid out clearly and that the recommendations will be helpful.
There is detailed feedback and numerous explicit or implied recommendations woven throughout this report and it is not our intention to capture all of them in this executive summary. Instead, we’ll try here to briefly address the research questions that guided our work and to outline some of the larger themes or directions that emerged and that might become conversation starters for COUNTER.
Release 5 has been generally well received and is regarded to be a welcome improvement over Release 4. The main problems we surfaced have been related to the roll-out. For publishers, changes to the CoP during the implementation phase were particularly problematic and this in turn created confusion and challenges for intermediaries and libraries For libraries and intermediaries, the wide variations in publisher implementation dates necessitated – indeed still necessitate – dealing with a mix of R4 and R5 data from different sources, and this, combined with a lack of understanding of how to compare the two, created a number of downstream problems.
We found that all the metrics and all of the Standard Views are widely used, and that most of the libraries not currently using specific metrics or SVs suggested that they might well do so in future. There was an expectation that multimedia information resources will continue to grow in importance and that data requirements with respect to these will need fuller development in future releases. Most providers and users offered that they are still getting used to R5 and learning how to use it effectively, and that there is no great appetite for further amendments or improvements in the immediate future, although the CoP will need to continue to evolve in due course.
Our respondents acknowledged that there is now sufficient training and guidance available for producers and consumers, although some further efforts on producing translations would be welcomed. For the future, though, COUNTER should consider how best to ensure that training materials are stable and available from the outset and how best to coordinate with other organisations to minimise duplication of effort.
There are suggestions for how best to ensure that effective feedback loops are in place and some additional specific recommendations are included for consideration, for example with respect to audits. While a robust audit process is essential if the Code of Practice is to be relied upon, audit costs are a deterrent to OA publishers and repositories, and a significant expenditure for smaller subscription publishers. At the same time, the current process is producing less than satisfactory results in terms of consistency and reliability. We have suggested that COUNTER might consider taking a more hands-on role in the audit process.
Since, as we have noted, most problems associated with R5 were related to the roll-out, this is a good time to take stock of the lessons to be learned and applied to any future revisions. Our report offers recommendations in this regard.
Extensive query logs kept by COUNTER detail both those queries which could be answered straight away and those which needed referral to the TAG. The queries fell into two categories: those which were entirely customer-specific, and those which could be generalised. We noted that the latter have already been picked up and added to the website FAQs for the community’s benefit.
The main themes covered in the website FAQ can be summarised as follows:
Each of these themes appears to be now fully addressed in the documentation (although this may or may not have been the case when the questions were asked), with the possible exception of the relationships between R4 and R5 reports, which came up repeatedly during our project.
We examined the available reports for interim audit fails since February 2019, to see whether there were any discernible trends. Only one COUNTER auditor provides reports with enough detail to analyse. This auditor is responsible for approximately 50% of the total audits, so while the data is not complete, it should be enough to spot any emerging patterns.
The Project Director informs us that more interim fail reports were received than usual after the implementation of R5, which is probably to be expected. The auditors had reported that the same pattern had been seen after the implementation of R4.
In each report, the auditor lists the Standard Views that had failed one or more of the several tests specific to that view. Five possible types of failure are listed:
Of 161 test fails, only 4 were related to inclusion of OA papers and only 7 to incorrect Year of Publication.
Format problems were involved in 48 of the test fails. However, these audits were predominantly undertaken before the Validation tool was available. We note that this tool should allow such problems to be picked up and corrected ahead of any future audits.
This leaves over-reporting and under-reporting of results. Although the specific technical problems leading to these results are not reported, we hypothesised that a noticeable excess of one over the other for a given report might point to an area of difficulty. In fact, both overall and for each Standard View the patterns of over- and under-reporting are fairly randomly and evenly distributed.
PR-P1 recorded the most failures (14), but these were evenly split between over- and under-reporting.
In conclusion, the audit failure reports do not provide any evidence of a consistent and systematic difficulty in implementation of any section of R5.
We have included selected quotes from the survey responses in our summary; these are highlighted in orange. In general, the quotes have not been edited but are reproduced exactly as submitted.
There were 390 responses from libraries and consortia from the following regions:
10% of respondents were from consortia, and 90% from individual libraries. 55% get their reports directly from publishers and vendors, 1% via an intermediary such as JUSP, Ex Libris, etc., and 44% from both sources. Respondents were asked to coordinate responses from their teams, so we did not collect data on the individuals responding. Responses were received from a good range of teaching, research, and mixed libraries, and the institutions represented included community colleges, small and large, public and private universities, and commercial and governmental organisations.
Aims. R5 is generally accepted as simpler, more flexible, and clearer. However, the verdict was by no mean unanimous; 56% thought R5 was an improvement over R4, while 28% disagreed. There was good agreement regarding flexibility, but while 45% felt that R5 had succeeded in its aims of increasing simplicity, 24% disagreed.
Roll-out. Regarding the roll-out, 63% of respondents felt that the changes and the reasons for them were clear and well-explained. 53% felt that the Code of Practice itself was clear (cf.10% who disagreed or strongly disagreed). The supporting material was clearly presented (58%, cf. 15% disagreeing) and available in good time (65% cf. 6%, perhaps a little surprisingly in the light of some comments made in focus groups). Fewer respondents agreed that the documentation contained all the necessary information (39% cf. 22%), or that the relationship between R4 and R5 had been made sufficiently clear (49% cf. 25%).
Views and metrics. We asked which Standard Views were used, regularly or occasionally. All the SVs are used and needed. The least used View is Multimedia, which is used by 48% of respondents. Most (75%) people who hadn’t used a particular SV thought that they might in future.
Similarly, all the metrics are used by a reasonably large proportion of the respondents. The least-used is Searches, Federated, used by 40% of libraries. It might be sensible to revisit the search metrics during the next iteration of the CoP. Problems were noted by several respondents:
Federated and Automated Searches inflate search counts to such a degree that it becomes almost a meaningless metric. It is more useful to know that the search button was clicked by someone, than to also see that click multiplied by ten or a hundred as a result. Searches_Regular or Platform is quite sufficient to determine the necessary information. Unique Investigations metrics are of relevance for databases and platforms that may only have clicks on abstracts or news articles, but at the title level, knowing Total Investigations and then Unique Requests is normally sufficient to understand usage.
58% of those who didn’t use a particular metric thought they might in future. Overall, R5 seems to have provided reports and metrics which are useful to significant numbers of users. There are no obvious candidates for culling in any future simplification.
Ease of use. 51% found the transition for R4 to R5 difficult or very difficult. Only 16% found it easy or very easy. 65% found it time-consuming, while only 10% found it quick. However, once set up it appears to get easier. 11% found it difficult to get the reports, while 50% found it easy.
39% found the reports easy to understand and use, 29% found them complicated, while 32% were non-committal. There is clearly a trade-off between simplicity and flexibility.
Uses. The reports are widely used for acquisitions (69%), collection development (82%) or deaccessions (53%). This may be some comfort to publishers who may suspect that deaccessions are the main focus. Other reported uses fall largely into two categories: use for external reporting, including IPEDS, ARL, ACRL, NCES, SCONUL; and use for internal reporting, often including annual reports and reporting to Departments or to higher management. There were just two uses that didn’t fall into any of these categories:
User study and potential dataset for machine learning (article recommender)
Secondary indicator of high-level access and/or discoverability problems.
Data sources. For 75% of respondents, COUNTER reports were the main source of data. 30 respondents noted that some suppliers are non-compliant, and so their non-COUNTER platform reports had to be used. Other sources of data were:
JSTOR usage reports
Feedback from researchers
Of these, EZProxy and EZPaarse were mentioned most often.
65% of respondents combined the reports with other data sources, primarily KBART, ILSs, data from subscription agents or vendor dashboards. Other data sources mentioned were:
Data about faculty publishing
List prices, bibliometric and impact data
Institutional outage reports
ERM or eresources management systems
Half a dozen respondents mentioned that they planned to do this in future or would welcome information about how to do it.
16% of respondents (40/250) provide data to Unpaywall/Unsub or other services to help with collection management decisions. A further dozen indicated that they were considering doing so in future.
Support materials and tools. The Friendly Guides were well used and are well regarded. Awareness of the Manual for Librarians was high, but usage was lower. The Foundation Classes had also been found helpful (33% of respondents rated them as particularly helpful). To a slightly lesser extent the webinars were found useful – with slightly more respondents using the recordings (and finding them helpful) than attending in real time. Awareness of the Newsletter was high, but only 9 respondents categorised it as particularly helpful – this suggests that a rethink of the contents might be in order.
We asked what training was missing. Various suggestions were made, but only two came up repeatedly: (a) more was wanted on a direct comparison between R4 and R5; and (b) more use case examples would be welcomed.
52% of respondents agreed that it was easy to find what they needed in the support materials, compared with 14% who disagreed, and very similar proportions rated the materials as clear and comprehensive.
The translations were welcomed by people for whom English was not a first language, and more requested.
Overall, the view on the training and support materials was broadly positive. More translations should be commissioned, more case studies added, and a fuller explanation of the relationship between R4 and R5 added.
Direct support from COUNTER. 33 respondents reported having approached COUNTER directly for help. 26 said that the response with prompt and helpful, and 3 that it was slow but helpful. Only 4 felt that the response was not helpful. Clearly the COUNTER team are providing good customer service.
Overlap between R4 and R5. We asked whether the differences in implementation schedules from different vendors was a problem for librarians. Only 16% said that it was only a minor (or no) problem. 53% said it was a nuisance they could work around, and 28% reported that it gave them serious problems. Six respondents (2.6%) complained that it made the reports useless.
Requested amendments or improvements in the Code of Practice. The most useful suggestions are given below:
An EXACT analogue to the old JR1 would be most helpful to libraries. (That is, INCLUDING OA use.) They need an easy way to see total use, rather than buried in the new J3. It might be useful to add a total as well in such a report. Librarians really miss that. Same for books.
Make more of the critical metadata mandatory, eg. that the vendor must provide the DOI in reports if the content itself has a DOI.
There is a need for one simple report that provides a total of ALL usage (total item requests) for each category (books, journals, media). Providing that aggregated data for the annual surveys is difficult and time-consuming task that has been needlessly complicated by the structure of COUNTER 5 reports.
How frequently should future updates be made? There was no real consensus on this. 39% thought that revisions should be made as often as the need arises, while 42% would prefer changes to be grouped into evenly spaced batches. There was less support (18%) for leaving updates until absolutely unavoidable.
Trust. 68% of respondents trusted the reports. 26% weren’t sure, and 6% said they didn’t trust them. We asked for reasons, and had 66 substantive replies, falling into three main groups:
From the focus groups and interviews, we gather that there is some justification with regard to the second and third points. There were differences of interpretation, particularly about eBooks, for instance, and COUNTER should be aware that this has a negative effect on user perception as well as on vender satisfaction. Improvements in the audit process might help here. There were also issues with ongoing changes in the CoP and advice on its interpretation that caused month-on-month changes to the reports, and again this has a negative (and lasting) effect on user trust.
Open access. 77% of recipients said that R5 reports provided sufficient information for Open Access content from born-OA publishers, and 72% agreed that they provided enough information about hybrid publishers. However, the comments reveal a great deal of confusion about what such stats would mean. Library needs in this respect will continue to evolve rapidly, and we recommend that this be kept under review.
There are a few comments along these lines:
OA delayed would be nice to have (it should appear in 5.1, right?). It would be great if COUNTER report could distinguish the type of OA similarly as it is eg in UnPayWall, so we can easily distinguish pure OA, hybrid OA, free content etc.
The unknown is how many articles in an issue are OA vs paywalled, which doesn’t show up in usage reports.
We need usage data on delayed OA.
43 publishers completed the survey. The majority were based in the USA (18) and the UK (10), with a scattering from other countries: Germany (3), Switzerland (2), Brazil (1), Canada (1),Ethiopia (1), and France (1) (6 did not indicate their location).
Just over half were large (annual revenues over $14m), a few were medium-sized ($10m-$14m), and the remainder were smaller. (These size categories reflect the COUNTER membership bands).
Approximately half of respondents deliver COUNTER reports directly; the remainder via a platform provider such as Atypon or HighWire (28%) or a third-party supplier such as ScholarlyIQ or MPS (19%).
Only 10 publishers reported that they use COUNTER reports themselves. We explored this issue further in the publisher focus group and learned that while the COUNTER reports are not widely used for internal purposes the data is widely used. Publishers explore usage data to provide customer support, inform editorial practices, do sales and marketing, and provide reports to authors and Society publishing partners. See the publisher focus group section later in this document for further details.
Aims. We asked whether R5 addressed the changing needs of publishers, intermediaries, and libraries. 50% were non-committal. Of the rest, more agreed (11) than disagreed (5). Respondents generally disagreed that it had been made easier to achieve compliance, although this may reflect the fact that they needed to go through a new compliance process at all, having been previously compliant with R4.
Publishers were split on the question of whether R5 had increased clarity with clear terminology, with half agreeing and half disagreeing. As to whether it was an improvement generally over R4, 12 agreed, 4 disagreed, and 14 didn’t know.
Overall, publishers did not overwhelmingly agree that R5 met its aims, but there was a trend towards agreeing that it was an improvement but that it had been a nuisance for publishers to implement.
Roll-out. 57% of publishers felt that the changes and the reasons for them were clear and well-explained, but a significant proportion (37%) disagreed. Asked whether the technical documentation was clear, adequate, and timely, 40% agreed, 30% were noncommittal, and 30% disagreed.
55% agreed that the deadline for implementation was realistic, but 25% disagreed and a further 6% made no effort to meet the deadline.
Implementation. 23 (74%) had successfully implemented Release 5. Two more were proceeding smoothly, and 6 (19%) were still having problems. No respondents said they weren’t going to implement. Only 20% found the implementation straightforward, and 60% found it tougher than expected.
We asked what sort of issues had arisen:
Last minute changes, e.g. change of the definition of OA, still not implemented as we need to change our metadata. Discussions about the book reports, i.e. dumping full book downloads into the same aggregate number as single chapter downloads, without an option to distinguish them, are still a problem. Changes in rules which publication year to use when a book or reference work has chapters published across multiple years – we had to revise.
Complexity of our own IT landscape and the decision to build C5 as a separate application to C4. Also, the merger between Springer & Nature.
Creating new and ensuring correct data feeds to our third party report provider. Making sure all the reports were showing the correct data based on requirements. Maybe this was more an issue with our report vendor but it took many attempts and corrections and a good deal of QA testing throughout the implementation to get to where we were confident with the reports being produced. All of this delayed our implementation significantly. In addition, there was some confusion from COUNTER on data requirements in the beginning and how to process that data, but in COUNTER’s credit they did respond to questions quickly and provided clarification.
Supplier understanding was inadequate based on the new rules and what this meant specific to our content platform. Had to educate supplier as to what correct should be in accordance with COUNTER rules and even passed audit when the rules were still incorrectly being applied to our usage, which left us feeling that the supposed controls to ensure consistency across publishers simply are not there. Proactive support from COUNTER in foreseeing these issues and being more hands in guiding us throughout this journey was lacking (although one could argue had the supplier done a better job this could have mitigated this issue). Too much time spent interpreting the new rules with regards to what can and can not be measured given the lack of clarity on this front.
Many of the recommendations were ambiguous. In a few occasions the standard said once thing and then a different piece of documentation contradicted what the standard requested – we were pointed by COUNTER staff to different pieces of information at different points in the implementation. The sample reports were not always accurate.
Communication both internally and externally
We feel we were used as the (unpaid) development team for our auditors. They seemed not to fully grasp what was required to implement C5. One of the auditors appeared to have limited experience with usage statistics or publisher platforms. We also felt put under pressure to complete the audit to their timescales to avoid a delay for other customers. Technical documentation was updated on several occasions, but we weren’t made aware.
The standard was not locked and there was a significant amount of change after development was complete.
Losing stats for open access content in the standard reports was confusing
1-Consorsium reports not in COUNTER 5, Consortium still ask consortium level COUNTER reports. 2-There is no reporting option for back-file usage similar to COUNTER-4 JR 1a.
The documentation wasn’t clear enough in places. We had to restructure the way we do consortium reporting.
Early in the development process, the technical specification were unclear and could be difficult to find. As time progressed, they improved.
As stated before, SwaggerHub documentation of the SUSHI API expectations was neither complete nor stable when the implementation period started after “release” of the standard. neither was the publishing of the COUNTER documentation. This is wasteful of time & money for implementers; and it is frankly unprofessional for an organization trying to produce a de facto standard.
Of the six who were still having problems, all six reported that lack of staff time was hampering them. Three said that the documentation was inadequate. Two “other” responses:
We have experienced some inconsistencies in communication – e.g. sometimes what the validator tool flags as an error is different from what the COP says; the understanding of what we had to do in order to become compliant when we signed our declaration of compliance with COUNTER was different from what the auditor told us we needed to have in order to be compliant when we contacted them to schedule an audit. And some of the updates to the documentation, while helpful for clarification, have required us to go back in and make more technical changes (e.g. changing the way we log outlinks to Altmetrics since that is now considered part of Investigations).
Implementation has been an absolute nightmare. Provided examples (e.g., the Excel examples, and the SUSHI website) were incomplete and contained errors. Requirements are constantly changing – some changes are disseminated through email, but not always updated or explained in official documentation. There are multiple sources for documentation, and I never know which one to use, or which one is the most accurate. The validation tool has contained mistakes, and has changed over the course of our audit. Overall, there has just been no consistency or clarity. I just want one source of documentation that’s complete and accurate that I know I can 100% rely on, and that absolutely does not exist.
We also asked about the implementation of COUNTER_SUSHI. 8 publishers found it straightforward, 10 said it was about how they’d expected, and 9 said it was complex.
Validation tool. Most publishers who had used the validation tool had found it useful or very useful; only two reported finding it not very useful. However, it had been developed too late for 7 publishers, and another 6 said they weren’t aware of it (this may again reflect the timing rather than a lack of promotion).
Costs. About half stated that the costs were about as expected. Most of the remainder (11/25) said that they were higher, and only 2 said they were lower.
Audit. 19 respondents had passed an audit, with three more in progress. Of these, 8 found it onerous or extremely onerous, and only 2 found it straightforward or extremely straightforward. Half found them to be about what they expected; all but one of the remainder found them quite (4) or extremely (4) expensive. One publisher (medium-sized) found them trivial.
Support materials. As with the libraries, the Friendly Guides were well used and highly rated. More people made use of the webinar recordings than watched them in real time, but they don’t get a lot of “Particularly helpful” ratings. The Foundation Classes were well used. Nobody rated the Newsletter as particularly helpful.
We asked what else might have been useful. There were requests for more on SUSHI, and for more case studies and examples, and more translations were requested. Not everyone was happy:
The release 3 documentation was perfect. We printed it out, put it in a binder, and could always rely on it to get the exact information needed. It had useful examples, and was clear and concise. With R5, there’s way too many documents and examples and emails floating around, all with out-of-sync information. Forgive my bluntness, but R5 was a complete mess, and as a developer, it was one of the most stressful implementations of my 14 year career, not because of the technical difficulty, but because of the scattered, and constantly changing requirements.
Consistency and completeness. We found numerous conflicting resources and our auditor brought new items to the table repeatedly. We would make changes to address items, then newly identified and unrelated problems would arise.
The frustration I had was having to read through an entire document whereby only small elements of it actually meant anything to us as a publisher. What COUNTER could have done is given more thought to each publishers content model and made clearer what the transition means if you fall under say certain categories. I should add though, our supplier did provide us with an understanding as to what we needed to change with respect to feeds etc to be able to support the new standard. However, if I took that information….and the information contained within the training documentation supplied by COUNTER, you wouldn’t necessary think they were working towards the same goal.
None of the documentation contains instructions for manipulating the data into a usable format, i.e., pivot tables.
Standardized examples need to be available earlier and when they available they need to be valid.
The guides were comprehensive but what was lacking is the ability to relate the information in those guides to our content model. Many publishers offer content in precisely the same way so perhaps more thought could have been given in relating the change to specific content models as a means of providing needed clarity. The examples that explained the new measurements were all journals related…not enough information given to other content models such as books which created challenges in understanding the new rules.
Create one single document with all information, and robust, complete, accurate examples. Edit updates in-line with strike-through text, and append a change log at the end.
About half of respondents agreed that it was easy to find what they needed in the support materials, while a quarter disagreed. The same proportion disagreed with the suggestion that the materials were clear and comprehensive.
Overall, the support materials were found helpful by most, with the reservations noted above.
Direct support from COUNTER. 18 publishers reported having contacted COUNTER. 14 found the responses prompt and helpful, 4 slow but helpful. As noted with the librarians, good customer service is one of COUNTER’s strengths.
Requested amendments or improvements in the Code of Practice. We asked what changes might be necessary.
Definitely the book reports need a distinction between full book downloads and chapter downloads.
Totals in all reports.
C5 doesn’t include IP address usage identification. We still have a custom report (created for C4) to extract this information which is available to each customer on their MPS account.
It needs to be better for OA stuff, and libraries need to understand that we can’t always tie OA usage to them.
In my honest opinion, sometimes less is more. I would love to see very simple standard reports with just the most important metrics, e.g. unique_item_requests for journals and unique_title_requests for books. All the other metrics are rather nice to have really.
Where the code of practice ended up is very good. The process of getting to this point has been painful at time.
The issues of improving the book reports and making better provision for OA came up in the focus groups as well, as did the suggestion that simplification and consolidation rather than further complication was desirable.
How frequently should future updates be made? Over half (57%) thought that revisions should be rolled up and spaced out evenly. A quarter felt that they should be left until absolutely unavoidable, and only 17% were in favour of updates on a rolling basis.
Customer feedback. We asked what sort of feedback publishers had received from their customers:
What could be expected – need to learn about the new metrics, how to compare 2019 to 2018.
Customers found it confusing that CoP doesn’t allow book usages to be reported in chapters as before.
A few were concerned with the JR_1 reports leaving out titles with 0 usage. They prefer to see these titles. But we have a relatively short list of titles, so including these was never a problem for us.
Many are still using R4 and find R5 confusing. We have had to provide training to our staff and to customers to help them better understand all the new features and information. The biggest area is helping them understand how to compare to R4 reports, which reports to use and which metrics are comparable.
Some confusion around our books usage whereby one customer thought we had processed this incorrectly because the well as requests vs investigations usage were the same. We made clear this was because we only offered books and chapters and no other content items, hence why those measurements will show the same numbers.
Most customer inquiries requesting CR5 are related to harvesting via SUSHI. Some customers are confused by how to insert the additional parameters in the URL based on what they are trying to retrieve.
As previously stated, some customers prefer the simpler R4 reports. Otherwise, users are happy with our provision.
Most feedback has been positive, but we are doing a lot of work before sending reports to customers (already in pivot tables; only providing item request data) ensure the reports are easy to understand. We are providing a lot of support to new customers and we’re pretty sure if we were to simply send the raw data outputs there would be a lot more questions. One of the primary questions was around comparison with R4 reports and there hasn’t been a sufficient answer to that. We wonder if a sufficient cross-section of the librarian community was reached during the R5 consultation to understand how the wide variety of librarians use usage reports, and not just those with a keen interest in this metric (it feels like R5 caters to the technically proficient).
For customers who are using R5 reports they are grateful for the additional information. These same customers will also report that they are still unable to understand what reports they would need to use to see the information they are looking for.
Our customers in Asia and non-academic customers were less aware of R5 and still prefer to have R4 data.
Customer are still transitioning from R4 to R5. We are getting more questions on R4 reports and some are related to R5 reports.
Feedback on R5 reports has been surprisingly little. I have the impression that most customers still have to get their heads around it.
Early on, yet after the deadline to be at R5, we had library customers that were unprepared and confused by the new reports. It seemed like they were completely out of the loop regarding the new standard.
The item reports tend to time out because of the size.
Audit. Nearly half (45%) of publishers thought that improvements were needed in the audit process:
For our supplier to incorrectly process usage that contained clear anomalies in the way it was being measured, and for this data to be passed as correct by the party auditing it, shows clear weakness in this process. It raises questions as to whether remains a level playing field in how publishers are deriving these numbers in accordance with the rules which is concerning to us.
Generally our audit experience has been good, but for R5 we had a different contact who did not instil us with confidence. We had to handhold the auditor and explain how our platform worked.
Once an audit is complete and the organization has passed many changes occur and no new audit happens. Also, COUNTER themselves will then inform the organization to make a change to meet a new standard that again passed for the audit. The audit is not thorough enough.
Improvements needed for bot activities and how the data should be filtered from bot activities.
Well, that’s a difficult one. On the one hand, the audit process should be as lean as possible. On the other hand, I have the impression that the auditors test only the basic usage scenarios, and sometimes seem to not have detailed knowledge about the crucial points or the various platforms. So the quality of the auditors seems to vary significantly, at least that’s our impression based on the questions you get back. It’s a bit of a lottery…
The audit process is unnecessarily onerous. Also, if platform changes have not been made in the past 12 months, new audits should not be required.
The chief source of issues revolves around SUSHI implementation, which the audit does not address. A passing audit should ensure full compliance; it would seem the audit should include some of the technical implementation
Our audit has lasted nearly a year so far because new items keep coming up that were not previously brought to our attention. Also, the R5 requirements keep changing, so we’re trying to hit a moving target. It’s costing our small company a ridiculous amount of time and money to constantly have to revise our work.
As mentioned earlier, all reasons for not passing the audit need to be clearly defined earlier in the process. Also there were situations where the documentation said one thing, the auditor said another, and we had to make the changes to meet the auditor’s findings.
Overall view of R5. 59% viewed the update as a necessary evil. 27% said that it was a welcome improvement, and three publishers said that it was unnecessary and unwelcome, and they felt pressured to be compliant. That’s not a large proportion, although of course none would be preferable. Feeling was split between those who agreed that R5 represented a good ROI, those who disagreed, and those who were noncommittal.
COUNTER data was used internally by all participants to support selection/deselection decisions, in support of renewal discussions, and to support library assessment and budget bids. More specifically, focus group participants reported they use the data and reports to:
and this is where we have hit some problems in moving from 4 to 5. There is a blending of what we have available and some subtle changes to how we interpret this. Ebook download statistics would have had each chapter, and there is some mess in there in terms of cross-product comparison.
Consortia provide member libraries with aggregated reports across publishers and vendors, and also use the data to compare across institutions to flag institutions whose reported usage or cost per use is out of line with peers so they can dig in and understand why. Some consortia and members use COUNTER information for tracking the effectiveness of their own marketing activities. For example, if a resource is being promoted then usage information can tell them whether these efforts have been effective or not.
All participants were clear that they must combine COUNTER data with financial data, and this was generally done at the package or database level rather than the book/ journal level. The mismatch in granularity can be especially challenging if the data are all being managed by hand.
We did a project on various demand-driven and evidence-based acquisition programs and matched this to COUNTER data to see which models and platforms and approaches worked best and generated usage.
We have our own in-house search platform and so R5 lets us differentiate federated searches from user access. This lets us dig in to see if there are searches but no requests, and to try and understand what is going on.
In a consortial setting we are going to be doing some harvesting and comparison to see if some members are not using resources as we would expect which could flag up a training need/gap.
We do a lot of matching to consortial entitlements. We need to weed out titles in the COUNTER reports that are not part of our subscription. That is a lot of work, but it is worth it for us to get it right.
We look at ILL requests/journal and turnaways once per year. We take turnaways with a grain of salt and are not convinced that high turnaways is always reflective of real demand. Part of the reason for this is that we get high turnaways from journals that we subscribe to. We suspect for journals that these are turnaways from archives we have not purchased, although we do not know for sure. There is not a turnaway-by-year-of-publication-report, so we do not exactly know from what users were turned away.
R5 was agreed to be a big step forward, with fields more clearly named and easier to process. SUSHI is more effective with R5. The API is much easier to handle, even for non-developers, and helpfully more publishers appear to be providing SUSHI because of this.
Looking at usage by year of a journal or by year for a publisher’s portfolio of journals is helpful, particularly where Unsub does not hold data for a publisher. I have really valued these recent developments in COUNTER, and we really need to leverage this.
We use it to evaluate Patron Driven Acquisition, to see if books are used. If so we will purchase them. The year of publication in the title reports is an important new addition that we like for this use case.
We have been using R5 and since they reformatted TR_J4 it has got extremely easy to use. We look at year of use and year of publication.
In our Tableau visualizations we are calculating cost/use and year of use vs year of publication. For [a publisher] we are paying loads of money for recent content because that is what our subscription is really covering; we have already paid for all the old content. They claim we have a low cost/use, but we use the cost per year of use vs year of publication statistic to really push back at them. We have been able to cancel c. 20 journals for which only older content was being used. We monitor 2020 ILL requests for 2020 content for the same reason. That is a new use because R5 is easier to integrate into Tableau.
You can see which publishers make their journals look more current, and which are more accurate in their actual dating of content. There is a lot of variety here. [A publisher] is particularly bad about this.
Much easier to harvest data under R5 than it was in R4. We can build our own in-house tools too which makes things significantly easier.
Some challenges also emerged during this part of the discussion:
The shift to the new metric types in R5 clearly represented a steep learning curve. COUNTER experts needed to develop written materials to support their subject librarian colleagues with this, but once climbed the change is viewed as worthwhile.
That having been said, the roll-out of R5 was felt to be chaotic and disorganised which made things difficult, and librarians told us it would be better to finalise the Code of Practice, give publishers enough time to implement fully and properly, have a narrow window for going live timed to the end of the calendar year, and continue with both reports in parallel for a year or even two.
This has been very problematic to libraries as they have had to manually pull R4 and R5 data together across multiple reports, and this has broken reports and visualizations in various ways. Different vendors have made this change in different years, and some have yet to transition, so this was not a one-off pain point.
Libraries would have preferred to stick to the R4 reports for a vendor until there was widespread confidence that the R5 data and reports were right, and then to have a transition that coincided with a change in calendar year.
A key challenge remains the fact that many vendors are still not compliant with R5, and so perhaps some of these lessons can be taken on board as these vendors come online.
A lot of publishers are still not on R5, or not accurately so, and this means we can’t fully leave R4 behind even though vendors like Ebsco are switching it off. We still need that data for our use cases. That means we must go to publishers individually for relevant data, and it is extremely difficult to work out trends over time. And then we are having to check and reconcile even more versions of the data, if the SUSHI data doesn’t match the Platform reports – this is tricky. It would be better to have an aligned joint running time of one or two years, and then all switch off the older versions at the same time rather than doing this whenever they want.
A really accurate correlation table is essential for the old and new releases. Librarians told us there were errors in the COUNTER correlation table, and it was not comprehensive. There were other correlation tables produced by intermediaries that seemed to be more accurate. But no one was clear which intermediaries those were and whether the correlation tables on the COUNTER website now are correct or not.
There needs to be clear guidance that if you are looking for something like R4’s JR1 GOA that you then add these two new numbers together and get its equivalent.
The correlation table for databases is still unclear. What should I be doing? Should I be totalling up regular clicks? I need to align R5 to R4 and I don’t know the best way to do this. What is the equivalent to DR 1 in DR_D1?
Consortia had in some cases invested to develop their own harvesting systems and needed to make changes to these systems due to the transition from R4 to R5. This was said to be difficult, especially as the Code of Practice wasn’t stable in the initial stages.
Libraries and their consortia experienced other technical challenges too:
From a technical perspective, the implementation of the actual API definition by Swaggerhub, which if you are not familiar with YAML is brutal. I understand wanting to use the hottest, trendiest technology of 2019, but honestly a GetPost list would be much more approachable for normal humans.
IMS global maintains the LTI specification and they have test suites and if you are writing an LTI application you must pass the test suites. There are inconsistencies in vendor implementations for COUNTER, so examples and test suites would have been helpful to have for COUNTER. This would mean when I am writing a harvester then I can be confident I am getting the proper returns and in proper formats for all error and non-error conditions so I can be confident I’m handling all cases appropriately. I think COUNTER should have a test suite that publishers should need to pass and run through to completion to prove that they have implemented correctly.
Agreed. When we were testing, we just had to trust that if we got back correctly formatted data that what we were receiving was correct. Subsequently, we have been able to verify via the publisher web platforms, but this perspective was not available to us when we were implementing.
We have run into issues that the data coming back from the COUNTER interface. When compared to downloadable spreadsheets from the publisher websites, we have data discrepancies.
There were a number of helpful suggestions including:
It is in the IR report, which is a little weird. Just having number of views does not help us. Some of the streaming services provide incredibly detailed information. We want something in the middle: time viewed for example. COUNTER is a little behind on media and the importance of media collections is really growing. We are putting a lot of money into AV streaming since the Covid-19 shut down.
Interactive products such as anatomy illustrations are closer to computer programs than anything else, and there is really no way to track their use right now.
A fully OA world will still cost libraries money, and they need evidence of usage and value. A recurrent theme was that there is a need for more OA-related data and insight. A big feature of R5 was to remove OA content from the journal request reports, but there seems to be some regret about this. While it is the case that for hybrid titles usage needs to be separated for open and closed articles, it is essential that libraries are easily empowered to compare and understand both types of usage.
For a long time we wanted OA out of the journal request reports. That finally happened so we finally know about what we are paying for. But now that we are looking at potential cancellations and we are needing to review read and publish deals and the statistics show 0 downloads for the OA content… retrospectively I am kicking myself as we now must go digging for those numbers. That data is hard to find from different reports for different years.
Article level metrics are now recognized as more important, and especially so in an OA world and both for hybrid titles and for fully OA titles. To support new types of OA agreements, libraries need to understand the usage for their ‘read’ articles separately from their ‘publish’ articles. For open content, libraries would ideally be able to understand the global usage and the local usage of these same articles.
There was interest in usage information about discovery service providers in a more OA world, and the choices users make about which article version to access.
COUNTER is focused on publishers as content providers but not as discovery service providers… what about usage of these services? This would help us in an OA world. To understand this, we would also need to know about use in a linkresolver – what was clicked, and was what was used e.g. green, gold, or whatever open version.
Interestingly, no focus group participant reported using COUNTER for understanding the use of their institutional repositories. Instead they access usage statistics via their IR platforms e.g. DSpace.
We explored with librarians the costs and benefits of being COUNTER members. There was a roughly even split between participants from libraries that were members, that were not members, and who did not know the answer to this question. This triggered some interesting discussion, and perhaps highlights the need for better marketing about the value of COUNTER membership.
I do think that libraries have huge concerns about membership funding at the current time. Is there broad enough financial support for COUNTER? Do we recognize how important it is? Would more be prepared to pay something?
It just never came up before. We are not consciously not a member.
Many of our [consortium] members use COUNTER statistics, but they don’t know it… they think this information comes just from the publishers or platforms. We need to prepare a promotion event re COUNTER with our members.
The membership fees at their current levels did not seem to be a barrier, and the opportunity to have a say in COUNTER’s development seemed attractive (with some reservations about the time commitment this could entail).
Participants reports a broad range of internal uses for the COUNTER data including:
Smaller publishing houses may not be so aware of the range of uses for COUNTER and may have limited time to climb this learning curve to fully exploit the available data. More training on what is possible, why it is useful, and how to do it with limited resources would be helpful.
Publishers all reported that a lot of education and support was needed by customers around the introduction of R5 reports: there is a very long adoption curve, it’s not immediately clear how to compare R4 and R5 reports, and R5 looks intimidating. Larger library customers were more comfortable with R5 reports than perhaps the long tail of smaller customers. There are some specific skills needed, for example R5 is more usable if you are comfortable using pivot tables in Excel. Some customers struggled with the terminology, saying it is tantamount to learning a new language.
Customers were quite concerned with the loss of 5-year longitudinal data and continue to ask questions about how to compare elements of R4 and R5 reports. Backwards compatibility is highly desirable. Some publishers reported that they ‘went cold turkey’ and launched R5 with only the mandatory time overlap with R4 and this took place without much complaint or problem. In the library focus groups, however, this practice was highlighted as particularly unhelpful.
Some participants reported that customers continue to push back; continue to be confused about R5 reports; and still request R4 reports with which they are more familiar for their analyses and decision making. Other participants reported that this was their early experience but after a great deal of effort (e.g. communications, education, FAQs, training for sales and marketing colleagues, etc.) this challenge is largely overcome.
COUNTER provides a lot of support materials for publishers to reuse, but a consistent complaint from publisher participants is that these materials aren’t kept updated as small amendments are made which significantly reduces their value. Inconsistent information is given in the different support materials.
More translations of these support materials would be really helpful, both for textual documents and videos.
There was not great awareness of COUNTER, what it is for, and how it could be useful within the born-OA publishing community. Mixed-model publishers were not clear how COUNTER supported their fully OA titles nor their Read & Publish (or other OA transformative) agreements. Better support from COUNTER would be welcome as OA publishing is increasingly important. Usage information can help libraries ‘value what is free to them’ and publishers of open content benefit from insight about use by unaffiliated researchers as well as researchers affiliated with research institutions.
OA benefits readers as well as authors, so usage information continues to play an important role. However, it may be less central to purchasing decisions going forward. One participant reported that Read & Publish customers use TR_J1 and the publisher directs them to TR_J3 instead.
From an OA books perspective, a real challenge is the number of platforms across which books are accessed. A consistent approach to understanding usage across all these platforms is hugely helpful both to the publishers and to customers. Customers want to understand usage on the publisher’s platform vs. other platforms, and publishers would like more insight about and ability to compare the usage of closed and open books.
We are looking for a lingua franca. Anything we can say about/to customers who are not purchasing directly from us would be helpful.
This lingua franca for book usage needs to be more granular in recognition that there are many flavours of open, for both books and journals. For example, there is a lot of temporarily open content [NB: publishers describe this as free content to distinguish it from open] in response to Covid-19 and this needs to be distinguished from different flavours of permanently open content. One participant reported tracking the historical usage of their free content, but not of their open content.
There was some discussion about global usage reporting by COUNTER, and different views: some felt this would be helpful, and others felt this should remain out of scope for COUNTER because there are other providers who could do this much better and it is very difficult to properly eliminate non-human usage. This seemed a rather technical conversation, and it could be helpful to take a step back from how to do it and to begin the conversation afresh from the perspective of [publisher] user needs, i.e. what information could help an OA publisher provide customer support, gain editorial insight, do marketing and sales, and provide insight to authors and Society publishing partners?
Publishers’ priority lists include:
10 intermediaries replied to the online survey, of whom only 6 answered any substantive questions. More informative were follow-up interviews with a subset of intermediaries. From these conversations it became clear that while all ingest COUNTER data and reports from publishers and provide these to libraries – thus positioning them to have a really clear understanding of operational issues – some are more publisher-focused, and some are more library-focused. This influences their understanding of how the data will be used, with which other data it will be combined, and the workflows and tools that will be used in support. For example, one more library-focused intermediary told us that COUNTER is for collections management rather than product management, but without an understanding of the array of editorial, royalty calculation and reporting, sales and marketing uses that a publisher could make of the data and the systems and procedures with which it would need to be integrated in order to do so.
This stakeholder group acknowledged that R5 had been challenging and expensive to implement but feel it was very worthwhile. These stakeholders were convinced by the business case change: greater automation, support for a wider spectrum of resources, the ability to answer a broader array of strategic questions, and greater flexibility to drill down into it to answer specific questions that may arise. All worthwhile goals and phrased rather differently than COUNTER’s stated objectives of consistency, simplicity, flexibility, and clarity.
We were struck by the duplication of effort across this community and between this community and COUNTER itself related to communication with libraries and publishers, development of communications and support materials, and the provision of librarian training.
Every intermediary (and indeed publisher) we spoke to reported that their library customers found the R5 terminology confusing and that a lot of education was necessary. COUNTER provided materials that could be reused and shared – for example the Friendly Guides – but earlier engagement and collaboration with organisations that provide librarian training and accreditation would be extremely useful. More broadly, this sort of connection could also be leveraged to create a more robust feedback loop to COUNTER about what librarians would like to do with usage information and how they would like to do it.
To illustrate the scale of duplication of effort JUSP, for example, works with 300 libraries, 100 publishers, and c. 50k reports/month and engages directly with publishers and intermediaries if problems are found. Some publishers and intermediaries reported to us that they primarily rely on JUSP rather than COUNTER for information and feedback. COUNTER has developed its friendly guides and videos and JUSP has developed its own guides for librarians. JUSP begins with COUNTER tools to do its preliminary validation steps, but then has developed its own tools as well. JUSP reports that it is doing a great deal of work with Highwire and Silverchair on compliance, and that, as a result, many missing publishers will have their data become available in R5.
On operational issues, the intermediaries we spoke with were very helpful in surfacing and explaining challenges with the CoP and made practical and thoughtful suggestions about how to improve future roll outs.
On Release 5:
Regarding its planning and roll out:
We should perhaps note that there was uneven awareness of how to engage closely with COUNTER and to whom membership was open (one intermediary participant believed that only publishers could be members), but all recognised that importance and value of being closely engaged. This is one of many topics that might be dealt with in NISO groups that have been established to choreograph the library supply chain and its players (e.g. around ERMs). More active engagement by intermediaries direct with COUNTER would be useful for them and for COUNTER too, as there is a wealth of operational and practical experience in these organisations. There are some perceptions of conflicts of interest between intermediary business models and the evolution of the CoP which could be better managed through a more inclusive approach, and each intermediary has rich communications channels into the library and/or publisher communities that could be leveraged in support of COUNTER.
In addition to historic rollout challenges, two further broad types of challenges emerged for COUNTER during our project. The first challenge is the number of potential providers and users who are not currently aware of, or using, COUNTER or Release 5. We focused on two groups in particular – born Open Access publishers, and libraries in regions with low COUNTER engagement/membership – and there are clearly opportunities to engage meaningfully with both. The second challenge relates to the fragmentation of feedback and feedback loops, and we believe there are some real opportunities here to improve services and decrease duplication of effort.
In this community, participants wanted to know basic information that is currently not readily accessible on the COUNTER website:
It is easy to get the impression that COUNTER is for ‘traditional’ publishers and their library customers. COUNTER is in fact engaged with born-OA publishers and considering how best to support them going forward. There are some important differences between them and mixed-model publishers. For example, their users do not log in and so tracking what they do in a way that can be linked to a specific institution is impossible. The global usage of OA content might however still be relevant to a librarian paying an OA publisher fees via APCs or transformative agreements. And the global usage of open content across many platforms is of great interest to born OA publishers.
Once interested, born OA publishers need information not about how to transition from Release 4 to Release 5 but rather about how to get started with COUNTER at all. Information that would be helpful at this point includes:
Language barriers and low awareness appear to be two key challenges for COUNTER in Africa, Asia, Asia Pacific, the Middle East, and in South America. Institutions in some of these regions are very sophisticated in their use of data, and it is perhaps surprising that they are not already more engaged. Other institutions in these regions will be handling usage data in a very manual way, if at all, and so once aware of COUNTER usage data and reports would likely need further support and training to make use of them.
In these parts of the world, publisher-supplied Excel spreadsheets with information about downloads, denials, searches, and views can be important sources of usage information. One participant reported that she had used COUNTER R4 data in the past, in part supported by INASP and training that INASP provided, but with the change in remit at INASP is now on her own to figure out how to transition to R5.
Consistent themes in these conversations were the need to provide:
Publishers were very helpful to the project team, and proactively helped us connect with global customers. They were keen to encourage broader awareness of and use of COUNTER data in these regions as it would save them having to respond to bespoke direct requests for data and spreadsheets.
The audit process is the mechanism that triggers publisher compliance with the CoP and needs further consideration. Publishers expressed to us their gratitude to intermediaries/platforms for bearing the brunt of audit efforts, even where they are also involved. However, publishers also complained that:
Some publishers offer “COUNTER-like” reports, presumably because they are put off by the cost and resource implications of the audit process. There is, obviously, no way of telling how closely these follow the CoP.
We note that only one auditor (responsible for about 50% of audits) copies COUNTER in on failed audit reports; this makes it hard to judge whether the process is consistent across auditors, and weakens the feedback loop to COUNTER
The audit routine looks exclusively at the Standard Views, on the theory that if the SVs are correct, the underlying reports must be sound. This view is not held unanimously, and at least one interviewee felt that the Master Reports should be audited directly; the SVs would then just need a simple formatting check.
The audit process doesn’t address SUSHI at all, and it was suggested to us that it should, given the increased importance placed on supporting automation with Release 5. One intermediary reported that differences in how strictly publishers were parsing SUSHI had caused them problems. Another suggestion was that the audit process might test how reports worked with IMLS vendor products.
Transparency was raised as an issue:
The audit reports need to be more clear about what was tested, what reports were tested, and what was found so that we can be confident that the right things were audited.
It was also noted that:
Right now publishers can be given conditional passes or passes, and they are treated the same way while the conditional passes sort out whatever problems there are. This is encouraging and discreet and generous, but not as transparent as it could be.
The validation tool has been extremely helpful in reducing costs and the burden of audit, and the possibility of providing further automated checking of data and reports is under investigation. This is sensible, and we encourage it. The tool may also be helpful to auditors, although they are likely to require training in how to leverage it.
One platform provider asked why a repeat audit is required after 12 months if neither the platform nor the Code of Practice has changed. It had put this to COUNTER previously but didn’t feel that the reasons had been adequately explained. If there are good reasons to require repeat audits under these conditions, it might be helpful to make the reasons transparent. If the reason is that some publishers pass audit and then their data and reports still cannot be processed, then whatever is the cause of this needs to be addressed rather than requiring unnecessary audits of the whole community.
User experience of implementing and deploying COUNTER data, reports, and services triggers evolution of the Code of Practice and needs further consideration. We were struck by the degree of duplication of effort involved in providing support and training to implementers and to users of COUNTER data and reports, and also by the rather informal way this information is fed back to COUNTER and the limited resource available to act upon this insight. There would appear to be opportunities for better alignment, coordination of effort, sharing, and reuse of outputs. This in turn would lead to more accurate, relevant, and timely messages and materials.
Based on the input we have received, there appear to be opportunities to improve:
For the current and future CoP we recommend:
There is feedback and explicit or implied recommendations woven throughout this report and it is not our intention to reprise all of them in this concluding section. Instead, we’ll try here to suggest some larger themes or directions that might become conversation starters for COUNTER. As is so often the case in assessing the work of not-for-profit organizations, the familiar themes of communications, fiscal sustainability, and organizational growth are highlighted as an invitation for deeper consideration by COUNTER leadership. We don’t expect that any of this will be new to you, but we do think these issues should be continually revisited as the COUNTER initiative and broader information ecosystem continue to evolve.
As COUNTER seeks to engage with libraries, you might think beyond the transactional interests of collection librarians and consider the strategic value of quality usage statistics for Library Deans and Directors. How can COUNTER data be marshalled to attract increased campus investments, or to strengthen relationships with college deans and influential faculty members? Further, Library Deans might be interested in how COUNTER can support an organizational pivot to more data-driven decision making. To unleash the full power of COUNTER to inform libraries and the academies they serve, COUNTER might consider more direct engagement with deans and directors, perhaps by inviting their participation on the Board or hosting an invited symposium for deans to provide input on the metrics that would be most useful to support their strategic agendas.
There are similar opportunities to engage with publishers to ensure they are maximising the benefit they receive from their efforts to implement COUNTER data collection and reports. Supporting the development and exchange of information and case studies will help publishers understand how they can use COUNTER data for a range of their internal purposes. These materials would need to be sensitive to the fact that smaller publishers will have fewer resources to climb this learning curve and review the data and provide appropriate ideas and examples. A forum could perhaps be established to enable publishers to share information about how they are implementing COUNTER with customers, intermediaries, and other publishers or COUNTER could potentially work in partnership with publishing Trade Associations to ensure this information is more widely available.
Thinking more broadly about opportunities to enrich organizational communication, COUNTER might consider reactivating the currently dormant Communications and Outreach Group, which should be responsible for the accuracy and timeliness of information made available about COUNTER and its implementation. The Group should ensure that information being shared addresses the needs of novice libraries and publishers as well as the needs of experienced users pushing the boundaries in terms of their use of automation and R5, and ensure that communications and implementation support materials are correctly aligned with future iteration of the CoP. Of course, improved communication and outreach would require investment and additional administrative attention/time. The Board would need to weigh the expected benefits and costs, as well as if there are other more pressing strategic activities that would be unfulfilled or delayed if resources are directed to enhancing communications.
There may also be opportunities for closer collaboration with other organisations. For historical and geographic reasons, the roles of the COUNTER Executive and JUSP seem rather intertwined. Some of the services that JUSP provides seem rather core to COUNTER and are needed broadly across the community of COUNTER participants.
Our stakeholder interviews with publishers, libraries, and intermediaries surfaced opportunities to re-imagine the audit process. ‘Interviewees from all sectors felt that that COUNTER’s deeper involvement with the audit process would be welcome. This would ensure greater consistency in the audit process and would alert COUNTER to ongoing or more specific issues. COUNTER should also consider the development of tools for further automating the audit process This needs careful consideration particularly regarding the financial and technical resource required.
A key finding of this review is that there is clearly a warm regard for the COUNTER team and its efforts in the community. While trust in both the integrity and competence of COUNTER leadership is high, there is also a sense that the community trims its expectations to what is reasonable given current staffing levels rather than petitioning for what is needed. The Board should be assessing whether COUNTER resources are at an optimal level to support community needs. Toward that end, consideration might be given to ways to increase membership support, perhaps as part of an enhanced communication campaign as discussed above. Our discussions with libraries suggest an openness to membership, but many thought they hadn’t been asked or didn’t know the costs of joining or the benefits of membership. To the extent that revenue limits opportunities for stronger COUNTER engagement, improving member recruitment might be a means of surmounting that hurdle.
Regarding future updates and releases, the input we gathered surfaced the following concrete suggestions:
Proposed three-year revision cycle
We are grateful to everyone who completed our online survey, everyone who helped disseminate it, and to those who took part in focus groups, interviews, or meetings. We would particularly like to call out:
Larry Alford, Chief Librarian (University of Toronto Libraries, Canada)
Andrew Anderson, President and Executive Director (Library Information Resources Network)
Chris Banks, Director of Library Services (Imperial College London)
Romy Beard, Licensing Program Manager (EIFL)
Tony Bocquet, Vice President for Sales—Asia (Springer Nature)
Romain Boistel, Documentaliste (École des Ponts ParisTech)
Helen Buchanan, Content and Discovery Manager (Imperial College London)
Sunshine Carter, Interim Collection Development Officer and Electronic Resources Officer (University of Minnesota)
Ean Lee Cheng, University Librarian (National University of Singapore)
Barry Clarke, Manging Director of Singapore Office (Taylor & Francis)
Thomas Connertz, Senior Vice President for Electronic Journals and Books (Thieme)
Patrick Danowski, Manager of the Library (Institute of Science and Technology, Austria)
Jennifer Eddy, Collection Development and Analysis (University of Melbourne)
Jill Emery, Collection Development Librarian (Portland State University)
Lorraine Estelle, Project Director (COUNTER)
Noam Fass, Product Manager (Ex Libris)
Celeste Feather, Senior Director of Content and Scholarly Communication Initiatives (Lyrasis)
Sueli Mara Ferreira, Professor of Information Science (University of São Paulo)
Karolina Firman, Communications Officer (ALIA—Australian Library and Information Association)
Richard Fisher, Academic and Policy Correspondent (Independent Publishers Guild)
Gustavo Fonseca, Software Engineer (SciELO)
David Gaither, Library Associate for Digital Resources Management (Louisiana State University Shreveport)
Pam Geldenhuys, E-Resources Librarian (Exeter Health Library, Royal Devon & Exeter NHS Foundation Trust)
Vikas Gupta, Managing Director (Wiley India)
Karen Harker, Collection Assessment Librarian (University of North Texas)
Kazuhiro Hayashi, Senior Research Fellow (National Institute of Science and Technology Policy)
Hannah Heckner, Product Strategist (Silverchair)
Christopher Hegert, Data Analyst (University of North Texas)
Catherine Hill, Director of Publications (British Ecological Society)
Athena Hoeppner, Discovery Services Librarian (University of Central Florida)
Youngim Jung, Senior Researcher (Korea Institute of Science & Technology Information)
Conni Junge, Senior Research Manager (Wiley)
Asaf Kline, Vice President for Product Management (Ex Libris)
Caroline Kobusingye, Librarian (Makerer University)
Jo Lambert, Project Manager (JUSP)
Bastien Latard, Head of Research and Development (MDPI)
Wonsang Lee, Senior Researcher (Korea Institute of Science & Technology Information)
Liz Marchant, Global Portfolio Director (Taylor & Francis)
Shahajada Masud, Senior Assistant Librarian (BRAC University, Bangladesh)
Blessing Mawire, International Development Consultant (South Africa)
Stuart Maxwell, Chief Operating Officer (Scholarly iQ)
Regina Mays, UT Hodges Assessment Librarian (University of Tennessee)
Devon McArdle, Product Owner (EBSCO)
Sue McKerracher, Chief Executive Officer (ALIA)
Paul Meehan, Senior Development Officer (JUSP)
Tasha Mellins-Cohen, Director of Technology and Innovation, Microbiology Society
Pedro Milliet, Consultant (Brazil)
Cindy Minor, Senior Marketing Manager (Elsevier)
Jeremy Morse, Director of Publishing Technology (University of Michigan Press)
Ian Moss, Chief Executive Officer (STM)
Paul Needham, Technical Advisory Group, Chair (COUNTER)
Roseanna Norman, Research and Analytics Manager (Taylor & Francis)
Daniel Nunn, Sales Support Administrator (Royal Society of Chemistry)
Anne Osterman, Director (Virginia Academic Library Consortium)
Abel Packer, Director (SCIELO)
Juliet Palmer, Customer Service Executive (Royal Society of Chemistry)
Kimberley Parker, Executive Council Chair (Research4Life)
Oliver Pesch, Chief Product Strategist (EBSCO)
Andrea Powell, Director of Outreach and Publisher Coordinator (STM)
Tim Preuss, Technical Team Lead (Silverchair)
Jason Price, Director of Licensing Services (Statewide California Electronic Library Consortium)
Christopher Rennie, Lead Product Manager—Platform (ProQuest)
Liz Robinson, Product Manager (Royal Society of Chemistry)
John Sack, Founding Director (Highwire)
Wayne Sime, Chief Executive (ALPSP)
Corrie Snijder, Information Specialist (Wageningen University, Netherlands)
Susan Spilka, Director of Strategic Communications (Workplace Equity Project)
Mark Sutherland, Executive Director (CAUL)
Simone Taylor, Publisher (Workplace Equity Project)
Jeri Wachter, Strategic Insights (Workplace Equity Project)
Lucy Walton Digital Discovery and Services Librarian (Western Sydney University)
Charles Watkinson, Associate University Librarian for Publishing (University of Michigan Press)
Harald Wirsching, Vice President, Strategy and Market Intelligence (Springer Nature)
Laura Wong, Library Analytics (JUSP)
Monica Young, Electronic Resources Librarian (Chinese University of Hong Kong)
Michael Zeyfert, Head of Data Systems and Strategy (Oxford University Press)
We invited COUNTER users to supply details of any uses or approaches to COUNTER data that they thought might be of help or interest to others. We are grateful to those who responded and have provided summaries below.
Romain Boistel, Electronic resources manager, Department of Library and Scholarly Communication, École des Ponts ParisTech
Romain blends COUNTER data (R4 to 2018 and R5 from 2019 where available) to create dashboards in Tableau. The visualizations calculated for each publisher include:
He plans to add further visualizations when he has time (for example using the YOP data).
Tableau needs the data formatted in a certain way: a single-entry table with one column per variable and one row per record, with the header on row 1. COUNTER data is formatted as a double-entry table with one row per title and one column per month, and the header is on row 14 (for R5) with a few semi-empty rows above with the metadata. Romain preformats COUNTER data with a VBA macro that he coded, which:
He then performs calculations directly in Tableau to match COUNTER R4 and R5 data.
This might be of interest to other users who wish to analyse data using Tableau, and in particular it might be helpful to share the VBA macro to make this easier for people to do.
David Gaither, Digital Resource Management, Noel Memorial Library
David has struggled at times when his library has purchased a specific collection to get the correct usage for the collection at title level. For example, EBSCOhost offers various eBook Collections at additional costs. While the Database Master Report can, and does, list usage by collection, it does not break things down into title level usage and can sometimes skew incorrectly as a result (in the case of EBSCOhost, it doesn’t properly break down the eBook collection usage), which is the level of granularity desired by David’s institution. The Title Master Report offers all the title information one could want, except it doesn’t list from which collection/database titles are originating. As a result, he has had to use vendor-specific non-counter compliant reports that he has indexed with the Title Master Report to be able to sort by collection. A column in the TMR that shows database or collection origination would surely save quite a bit of time and effort.
The suggestion about the Title Master Report is worth noting for the future, and this approach may be informative for other users with the same challenge.
Jennifer Eddy, Collection Profiling Librarian, The University of Melbourne
Evidence Based Selection (also known as Evidence Based Acquisition) is a purchase model that has been developed for electronic resources that are one-time purchases, and the University of Melbourne uses it as part of the eBook purchasing strategy. The basic premise is that the library nominates a deposit amount and then the publisher agrees to open up access to a list of eBooks to a total value of X times the value of the deposit (usually somewhere between 3-7). All these titles are added to the catalogue and the library has access for twelve months, effectively as a “try before you buy”. The underlying assumption is that not all of the titles that the library selects will be of interest to staff and students but by selecting a wide range staff and students are allowed to (unknowingly) do the selecting for the library which purchases the titles that get used the most. While selecting nearly 5000 eBooks might seem like a large number, there are over 50,000 students and over 10,000 staff and the library will purchase fewer than 20% of the titles in the package. The rest get turned off again.
The COUNTER report lists usage for all the Taylor & Francis eBooks that we have access to, not just the EBS package, and only lists eBooks that have been used. There are two measure of use of eBooks, Unique Title Requests, the number of times an ebook was accessed in a single patron’s session (if a single user downloads multiple chapters from an ebook, or accesses the ebook several times during a session, all of these would count as 1 unique title request) and Total Item Requests, the total number of requests or interactions with a title. Because Unique Title Requests is independent of the way the eBook is accessed we regard it as the most reliable measure to compare eBook usage. Both these metrics are triggered by staff and students interacting with these eBooks in the course of their studies or research.
The last tab in the workbook is a list of all Taylor & Francis eBooks that were made available from May 2019 – April 2020 with the COUNTER metrics over that period added. We have used the ISBN of each title to match the usage data to the right title. Where a title isn’t listed in the report the usage can be assumed to be zero as the reports are designed to only include titles with use. The list is currently sorted in usage order from highest to lowest so all the zero use titles are right at the end.
The first five tabs all relate to selecting eBooks for purchase. The next three tabs are the COUNTER reports and the lookup tables that I derived from them. The next 6 tabs are Pivot Charts and Tables that analyse the usage by subject area rather than at the title level, with the overall analysis in Chart 3. The last three tabs are title lists where Low use covers all the titles that were used 0-2 times in 12 months, High Use covers all the titles that were used 3 or more times in 12 months and then the final tab that lists all the titles.
The TR_B1 reports have been combined with title list data from the vendor, firstly to select titles for purchase based on usage over our EBS access period. The collection as a whole was then analysed to inform decision making on the selection of a new set of titles for our current Evidence Based Selection package.
We have used the new Unique Title Requests as the primary selection parameter and then added in Total Item Requests to round out selections. The analysis of uptake across the various subject areas just looks at Unique Title Requests, given that is the metric designed to allow cross comparison.
Jennifer kindly gave us sight of her Excel Workbook. This is a nicely implemented and straightforward use of the COUNTER data. If the actual data were stripped out and a small amount of dummy data used, it might form a helpful example for others to follow.
Larry Alford, Klara Maidenberg, Sabina Pagotto, and Cristina Sewerin, University of Toronto
Larry shared a slide deck prepared by the UTL team giving an excellent overview of COUNTER and usage stats generally. It doesn’t cover any unusual uses, but gives a good overview for people new to the subject, and might conceivably (if appropriate permissions were obtained) form the basis of a presentation to be downloadable from the COUNTER website.