OLIS

Subject: Update on Review of ScienceDirect usage data [3/30/09]

Dear Customer

As per our communication to you on March 10, 2009, we would also like to provide a further update on the error that was found in the August through December 2008 reports.

The analysis of the 2008 error has been completed and the impact to your institution is that ScienceDirect usage downloads are overstated by ca. 9% for the months of August 2008 to December 2008, the period when the problem occurred. Therefore your 2008 full year usage statistics were overstated by ca. 2%.

Background of this error

ScienceDirect usage reporting adheres to the COUNTER standard. COUNTER prescribes what reports publishers should make available to their customers and also issues guidelines on how usage should be measured.  One of the rules of the COUNTER standard is that if a user requests one and the same pdf multiple times within 30 seconds it should be counted as a single request.  The same applies to multiple requests for one and the same HTML article, although this time the period is 10 seconds. (Please note that downloading the same article in HTML and PDF form is not seen as a double click). During the release of the new Access & Entitlement system for ScienceDirect, the software code that filters out multiple clicks on ScienceDirect articles was removed for performance testing. Unintentionally, the code was not reinstated once the testing process was complete. As a result, ScienceDirect was not COUNTER compliant during this period.

Articles downloads counts affected

Please note that only full text article counts are affected by the overstatement problem.  There is no impact on other data in the customer reports, such as the number of users, searches and sessions.  Equally, Scopus figures are not subject to these inaccuracies.

The ScienceDirect customer reports that are affected are:

1a) Number of Successful Full-Text Article Requests by Month and Journal

1b. Full text usage per journal per entitlement

1e. Unsubscribed article views

1f. Full text usage per journal for all groups/all member accounts

4a. General overview

4d. Document usage by file type

4e. Document usage by entitlement

4f. Document usage by content type

5a. Document access routes

Next Steps

January and February 2009 usage reports are available on the reporting system now. Owing to the large amount of data involved, it will take one to two months to fully correct the August through December 2008 reports. We will inform you as and when this information is released.

Audit

We sincerely apologize for the inconvenience that this error has caused and will work diligently to complete the correction of the data over the coming weeks and months. Upon completion, Elsevier will voluntarily submit our systems and data to an additional external audit, on top of the annual audits executed since 2007. A rigorous system of cross-checks has been implemented to avoid a recurrence of any problem of this nature.

Thank you once more for your understanding. We welcome your comments and feedback.

With sincere regards,

Hanneke Steuten,

Director Usage Research & Project Management

Dear Elsevier Customer:                        [March 10, 2009]

Two weeks ago, we informed you [sic] of an error occurring in the August 2008 through December 2008 ScienceDirect customer usage data. This is the second in a series of forthcoming communications where we will update you on this topic.

During the release of the new Access & Entitlement system for ScienceDirect, software code related to the usage reporting was removed for performance testing. Unintentionally, the code was not reinstated once the testing process was complete.  The impact is limited to August 2008 – December 2008 usage reports; and we are currently running an analysis to assess the impact per customer. We will contact each customer individually in the coming month to communicate the impact to your data, i.e. an overall percentage per year, but we will not have the breakdowns per journal for another few months.

January 2009 data is now being run with the new customer usage reporting software and will be posted on the reporting system by end of March.  Owing to the large amount of data involved, it will take two to three months to correct the remaining August through December 2008 monthly reports.  Elsevier has long strived to achieve the highest standards of accuracy in usage reporting. For several years, we have voluntarily submitted our systems to annual external audit.  In fact, this problem was identified via an internal check well in advance of the external audit.

We sincerely apologize for the inconvenience and will work diligently to assess, communicate, and correct the data over the coming weeks and months. A rigorous system of cross-checks has been implemented to avoid a recurrence of any problem of this nature.

Thank you for your patience. As stated earlier we will continue to communicate with you on this topic once the assessment by institution is completed.

With sincere regards,

Hanneke Steuten, Director Usage Research & Project Management