KMWorld CRM Media Streaming Media Faulkner Speech Technology Unisphere/DBTA
Other ITI Websites
American Library Directory Boardwalk Empire Database Trends and Applications DestinationCRM EContentMag Faulkner Information Services Fulltext Sources Online InfoToday Europe Internet@Schools Intranets Today KMWorld Library Resource Literary Market Place Plexus Publishing Smart Customer Service Speech Technology Streaming Media Streaming Media Europe Streaming Media Producer Unisphere Research

For commercial reprints or PDFs contact David Panara (

Magazines > Computers in Libraries > June 2017

Back Index Forward
Vol. 37 No. 5 — June 2017

Ebook ROI: A Longitudinal Study of Patron-Driven Acquisition Models
by Yin Zhang and Kay Downey

Analysis of triggers, pre-purchase activity, and other relative data helps to detect patterns to assist in predicting cost and ROI.
Kent State University Libraries (KSUL) has been using the patron-driven acquisition (PDA) ebook purchasing model since January 2012. Now, 5 years of PDA data makes it possible to conduct a systematic comparison to evaluate different PDA scenarios based on ebook usage patterns at KSUL between Jan. 1, 2012, and Dec. 31, 2016. This study specifically compares the economies of the straight PDA business model and the PDA model that uses the short-term loan (STL) component. This study may help libraries gain a better understanding of how PDA works in different scenarios and how to choose the scenario that works best to fulfill their needs.


In 2012, KSUL conducted a pilot proj­ect for a PDA ebook purchasing model, at which time we loaded 20,000 ebook discovery records into the library catalog. KSUL used ebrary as the ebook provider and YBP Library Services (now GOBI Library Solutions) to manage the content profile, invoicing, and deduplication services. Its success led to continual implementation of the PDA model for KSUL. Auto-purchases are triggered by the 10-10-1-1-1 paradigm with no STL component. This paradigm is defined as 10 page turns within the body of the text, 10 consecutive minutes of use, one print, one copy, or one download. Each week, new discovery records are added to the local catalog and become accessible to Kent State faculty and students. To date, more than 78,000 freely accessible discovery rec­ords have been added to the catalog. Of those, close to 6,000 merited enough use to trigger a purchase. As of this writing, we have not weeded the discovery pool. 

Today, accumulated PDA data makes it possible to conduct a systematic comparison of different PDA business models based on ebook usage patterns. This study specifically compares the rolling cost of the straight PDA business model versus the PDA model that uses the STL component in which books are not immediately purchased, but loaned at a percentage of the purchase price. The goal is to take a close look at PDA business models and consider the variables in determining the best value or deciding which acquisition practices are the best fit for the library. We wanted to answer some basic questions, such as, how do we evaluate ROI comparing one model with another, particularly with the STL PDA model? And how do we identify trends to help us learn how well PDA serves our library users?


PDA Business Models

Most commonly, PDA (also called demand-driven acquisition; DDA) is the model whereby a publisher or ebook aggregator provides the library with free access to a pool of ebooks that, after having been discovered and receiving substantial use, are automatically purchased by the library, usually seamlessly. This model is often combined with the STL option. 

The PDA model has been implemented in a number of ways. Some libraries provide access to an entire platform compendium for PDA. Some libraries—such as KSUL—provide access to a restricted profile through YBP, and others create the PDA pool via manual selection. Popular variations on PDA models include 1) STL in combination with straight PDA purchasing after STLs, 2) straight PDA auto-purchase without the STL option, and 3) STLs only with no purchase outcome. 

With the STL option, ebooks that receive substantial use are not immediately purchased but loaned for a fraction of the cost of the ebook purchase price. The ebook must receive a number of loans before a final purchase is made. The library presets the loan period generally for 1 day or 1 week, and the cost varies with the duration of the loan. Depending on the publisher, the 1-day loan fee ranges anywhere from 5% to 30% of the ebook’s list price. The 1-week loan fee ranges from 15% to 45% of list, although a few publi­shers may charge up to 90% of list. Once a certain number of STLs (most often three) have been triggered, the ebook is purchased. The final cost is the sum of each rental fee, plus the price of the ebook. The concept behind this model is that it will save the library from paying the purchase price for ebooks that receive low use and should help the library realize overall savings in the PDA program.

However, beginning in June 2014, many publishers began to increase the cost of STLs, issuing embargos on frontlist content, while some publishers stopped offering STLs altogether. Publishers were experiencing a drop in print sales and the added indirect costs of ebook production such as platform maintenance and the provision of MARC records. They believed that the PDA/STL model was unsustainable, created sales unpredictability, and was a major contributor as to why they had been experiencing revenue loss. At the American Library Association’s 2014 annual conference, a representative from Wiley reported that average revenue for a frontlist ebook offered via the STL model was approximately 86% less than that of an ebook purchase. As such, the PDA/STL pricing levels were not a feasible means to support ebook publishing operations (Thompson and Zusman, 2015). 

The publishers’ cost increases of 2014 and 2015, along with the content reductions, placed a strain on library finances for ebook acquisitions. Librarians from the library at St. Edward’s University reported an increase of nearly four times the amount of previous years. To mitigate the increased expense, they made programmatic changes while maintaining a limited STL component. In conclusion, they noted that the changes are not effectively reducing costs (Buck and Hills, 2017). Since KSUL did not participate in the STL model, the programmatic changes did not impact the budget. However, they did impact service in terms of the content currency available to its users. 

Despite these changes, many librarians continued to endorse the cost savings value of STL (Roll, 2016). Because the benefits of STL have not been supported in the Kent State PDA program, the focus of this study is to compare the long-term cost of straight PDA versus PDA with the STL component, to find out what, if any, benefit can be had using the STL model. 

Evaluation Approaches and Principles

Usage patterns remain the key measure of whether PDA ebook acquisitions serve the needs of library users. We argue that if the goal of acquisition is to support users, usage data should be an important component in model assessment and budget justification. In addition, we adopt a longitudinal approach to examine whether ebooks acquired through PDA usage-based triggers are a result of a one-time random act or if they receive sustained usage after the initial purchase trigger. 

The data sources we have gathered for the evaluation include the following (during the period Jan. 1, 2012, through Dec. 31, 2016):

  • Ebook records in the program’s PDA ebook discovery pool
  • ebrary Trigger Reports about purchased titles via PDA usage trigger
  • ebrary Title Reports about all PDA ebooks’ usage
  • ebrary Book Report 2 about all PDA ebooks’ detailed monthly usage
  • Post-trigger/purchase usage report from ebrary representative


1. Acquired PDA Ebooks Overview

During the 5-year duration of the KSUL PDA program from Jan. 1, 2012, to Dec. 31, 2016, a total of 78,393 ebook records were uploaded to our local library catalog as the PDA discovery pool for patrons to search—12,725 (16.2%) were used, and 5,701 (7.3%) received enough use to trigger a purchase. Table 1 (on page 6) summarizes the annual acquisitions, associated costs, and average cost per ebook. While the PDA program maintained a steady acquisition level in the first 3 years, there has been an increase of 300 to 400 ebooks each year in the last 2 years. The average cost per ebook is around $100. 

Table 2 indicates that page views have been the most popular and steady trigger event, followed by downloads. With the full download feature activated in November 2014, download has become the second most popular trigger option, and it gained popularity in the most recent years. Print has also gained popularity in the same time period, with copy as the least-common trigger event.

Table 3 shows the top publishers of ebooks acquired through the KSUL PDA program. University presses constitute the top publishers, consisting of 24.2% of ebooks purchased. Among the 102 university presses, Cambridge University Press is the largest contributor to our triggered titles with 279 titles (5% of total acquired). Taylor & Francis comes in as a close second, contributing 22.7% of ebooks acquired in the PDA program. One major publisher missing from this list is Springer, for the reason that KSUL acquires Springer ebooks via the OhioLINK consortia, which are excluded from the local PDA program.

The question of whether the PDA program can ensure the acquisition of more recent books is also a concern. Table 4 shows the annual breakdowns of acquired ebooks by year published. It appears that in each acquisition year, ebooks published in the most recent 3 years have been most commonly acquired, suggesting patrons prefer to choose the most current titles in the discovery pool. 

2. Usage of Purchased Titles

An important measure of a successful PDA program is whether ebooks continue to receive usage after the initial purchase. That is, is a patron’s usage-initiated trigger a random act or an indicator of need that leads to sustained use? As shown in Table 5 (on page 7), all except four purchased ebooks (5,697/5,701 = 99.93%) received post-trigger uses as measured by the number of user sessions—80% received more than one post-purchase user session. 

Table 6 (on page 7) details uses of acquired PDA ebooks over time by annual cohort, including pre-trigger uses, uses in the trigger year, and uses after the trigger year. A small portion of ebooks receives free, light uses before they are triggered. Evidently, all acquired ebooks receive sufficient uses in their trigger year. There is a decline in usage of purchased titles after the trigger year, but close to 20% of ebooks receive steady usage over time, even in their fifth year, for titles purchased the first year. 

3. STLs vs. Direct Purchase

One debate in implementing a PDA program is whether to include STLs or not. The KSUL PDA program initially adopted the direct purchase PDA model without STLs. However, the question remains whether it would make sense to have STLs for our PDA program. The actual PDA ebook usage data makes it possible to construct and compare different what-if scenarios to indicate the best program option for us. The two commonly adopted STL/PDA program scenarios for our comparison are three 1-day STLs before purchase and three 7-day STLs before purchase. 

Using the 5-year KSUL PDA program data, based on the ebrary PDA trigger threshold for an STL and each acquired ebook’s actual usage, an ebook would fit in one of the following categories: one STL (one trigger), two STLs (two triggers), three STLs (three triggers)—or purchase after three STLs (four or more triggers). Figure 1 shows the predictive trigger breakdown for KSUL purchases by program year as calculated by the PDA/STL scenario formula. 

Overall, about 20% to 24% of ebooks are not STL-eligible, in which case they have to be purchased without going through STLs, due to publishers’ license restrictions. Over time, if an ebook receives continued uses, the cumulated uses would lead to three STLs and eventual purchase. As shown in Figure 1, of the ebooks acquired in year 1, 85% would have been purchased during the program duration (by the end of year 5) regardless. Sixty-five percent of these were eligible for STLs and would have been triggered beyond three STLs, leading to purchase. For ebooks acquired in year 5, without the benefit of following years, 53% are already in the STL stage with two to three STLs, and 46% would have been purchased—24% of which were STL eligible. 

The number of STLs and the purchase status for each ebook based on its usage—along with its publisher’s STL rate, list price, and purchase price—make it possible to calculate the cost for each title under different PDA program scenarios. Table 7 shows the costs under different PDA program scenarios. For the ebooks purchased in program year 1, the following is true:

  • The actual cost of direct purchase is $86,115.
  • The projected cost for the 1-day STLs scenario is $132,089.
  • The projected cost for the 7-day STLs scenario is $144,295.
  • The projected difference between PDA without STL and 1-day STLs scenario is $45,973 less.

There are savings across the board for the direct purchase PDA scenario, and the projected cost-savings increase over time. Except the most recent year, year 5, with a large portion of ebooks still in the STL stage, both 1-day and 7-day STL scenarios would have cost us more for the same set of ebooks we purchased (see Figure 1 on page 7). It is expected—as supported by the usage pattern of ebooks acquired in early program years—as ebooks receive continued uses, there will be additional costs associated with additional STLs and even purchase after three STLs. 


To conclude, the STL option does not make sense for KSUL. The longitudinal study illuminated the potential impact of STLs in terms of budget and collection building, and scenario analysis determined sufficient use in correlation with PDA STL charges. The data shows that for the PDA ebooks purchased within 1 year, 80% would have received four triggers and been purchased anyway, if we used the STL option. Moreover, this data is consistent from one year to the next. Had the STL option been used, KSUL would have paid substantially more each year for the same content. 

Determining which PDA model is best for a library depends on established goals. Perhaps the most basic consideration depends on the philosophy and mission of the library. Today, more than ever, libraries have become a “just in time” service point that meets the immediate needs of the user. But even though this is the current mode in many libraries, traditional collection concepts continue in tandem, balancing ownership versus lease. Examining the right data will provide a means for assessing whether or not the program meets the goals. Analysis of triggers, pre-purchase activity, and other relative data helps to detect patterns to assist in predicting cost and ROI. Because ebook PDA is becoming an established acquisition model, research for other longitudinal studies is now possible. We are currently conducting a study that compares PDA/STL investment from eight large academic libraries to determine if findings from KSUL are consistent with other programs. Results of this and future studies may help libraries gain a better understanding of how PDA works in different scenarios and how to select the scenario that works best to fulfill their needs. 


Thompson, J. Michael and Zusman, Laura (2015). “The Long & Short of Short-Term Loans. A Report of the ALCTS Acquisitions Managers and Vendors Interest Group (AMVIG). American Library Association Annual Conference, Las Vegas, June 2014.” Technical Services Quarterly, 32(1), 76–83. DOI:

Buck, Tina Herman and Hills, Sara K. (2017).Notes on Operations Diminishing Short-Term Loan Returns. Library Resources & Technical Services, 61(1), 51–56. DOI:

Roll, Ann (2016). “Both Just-In-Time and Just-in-Case: The Demand-Driven-Preferred Approval Plan.” Library Resources & Technical Services, 60(1), 4–11. DOI:

[See also: Downey, Kay, Zhang, Yin, Urbano, Cristóbal, and Klingler, Tom (2014). “KSUL: An Evaluation of Patron-Driven Acquisitions for Ebooks.” Computers in Libraries, 34(1), pp 11–12 and 30–31.—Ed.

Yin Zhang   ( is a professor at Kent State University’s School of Library and Information Science. Her research areas include information uses and users, information-seeking behavior, and information systems. She has a B.S. and an M.S. in information science and a Ph.D. in library and information science. 

Kay Downey
 ( is an associate professor and collection management librarian at Kent State University Libraries. Her research interests include library acquisitions and collection development. She has a B.F.A. in painting and drawing and an M.L.S. in library science.