A download count looks like a simple number, but it is one of the most slippery measurements in scholarly communication. Does a count include automated crawlers? Does reloading a page count twice? Does a publisher count a request that was refused because the institution lacked access? If every platform answers these questions differently, then comparing usage across publishers, or even across two journals, is comparing things that were never measured the same way. COUNTER exists to make usage statistics trustworthy, comparable and consistent, and it is the reason a library can put figures from different providers side by side and mean something by it.
The COUNTER Code of Practice
COUNTER, which stands for Counting Online Usage of Networked Electronic Resources, is a standard maintained by a not-for-profit organisation supported by the library, publisher and vendor communities. Its central artefact is the Code of Practice, a detailed specification that defines exactly how usage is to be recorded and reported so that statistics from compliant providers are consistent and comparable. The Code sets out precise rules: how to filter out robot and crawler traffic, how to handle rapid repeated clicks so that double-clicks are not counted as multiple uses, how usage events are to be classified, and the exact structure and content of the reports a provider must produce. A publisher or platform that is COUNTER-compliant has agreed to follow these rules, which is what gives the resulting figures their credibility.
The current generation is Release 5, with the refinement Release 5.1 bringing further clarification and consistency. Release 5 was a significant simplification and rationalisation of earlier versions, designed around a smaller, clearer set of report types and a consistent vocabulary of metrics, making the reports easier to generate, interpret and compare than the more sprawling earlier releases.
What COUNTER reports measure
The key to understanding COUNTER is its precise terminology, because the words have exact meanings under the Code.
- Requests are the core unit: a request is a use of content, such as the viewing or downloading of an item. COUNTER distinguishes item requests, uses of an individual item such as a single article or chapter, from title requests, uses aggregated at the level of a whole title such as a journal or book. This separation lets a library see both granular article-level activity and higher-level title activity.
- Total versus unique. Release 5 introduced a careful distinction between total requests and unique requests, where unique counts collapse repeated activity within a session so that a user reading the same article several times in one sitting is not inflated into many uses. This makes the figures a more honest reflection of genuine demand.
- Access denials. COUNTER also records when a user is refused access, for example because the institution does not have a licence or because a simultaneous-user limit was reached. These denial metrics are valuable in their own right: a high level of turnaways tells a library that there is demand for content it does not currently provide, directly informing acquisition decisions.
By standardising these categories, COUNTER lets a library answer practical questions, which resources are genuinely used, where demand is being turned away, what cost-per-use a subscription represents, on a consistent basis across all its compliant providers.
SUSHI: automating the harvest
Standardised reports still have to be collected, and doing so by hand across dozens of providers would be unmanageable. SUSHI, the Standardized Usage Statistics Harvesting Initiative, is the protocol that automates it. SUSHI defines a standard way for systems to request and retrieve COUNTER reports programmatically, so a library’s statistics system can automatically pull the latest reports from each provider rather than relying on staff to log in, download spreadsheets and consolidate them. The combination is what makes COUNTER usable at scale: the Code of Practice ensures the reports are consistent, and SUSHI ensures they can be gathered automatically and brought together for analysis.
IRUS and repository usage
COUNTER’s logic applies not only to publisher platforms but to repositories. IRUS, the Institutional Repository Usage Statistics service, applies COUNTER-conformant processing to repository usage, so that downloads from institutional and other repositories are counted according to the same rigorous rules, with robots filtered and double-clicks handled, and reported in a standardised, comparable form. Operated in the UK context through Jisc as IRUS-UK and extended more broadly, IRUS gives repositories something they otherwise struggle to produce: credible, standardised usage figures that can be compared across institutions and aggregated nationally. For the open-access and repository community, this is significant, because it lets repositories demonstrate the reach and impact of the content they steward on the same evidential footing that publishers enjoy.
Why standardised usage data matters
The payoff of all this machinery is decision-making grounded in comparable evidence. For libraries, COUNTER data underpins collection management: deciding what to renew or cancel, calculating cost-per-use, and using access-denial figures to identify unmet demand. For repositories, IRUS provides the usage evidence that supports the case for open access and helps institutions understand how their outputs are actually used. For the wider system, a shared standard means that usage, one of the few direct signals of how research is consumed, can be reported and compared honestly rather than asserted with incompatible numbers. This consistency is itself a form of good data practice, akin to the shared definitions a CASRAI data dictionary promotes and the interoperability goals of FAIR data: when everyone counts the same way, the resulting figures can be trusted, combined and acted upon.
