Interlibrary Loan (ILL) is a key consideration in forecasting the effects of cancellation on fulfillment and cost. So naturally,, it's also an important part of Unsub's five-year forecasting approach, and is prominently displayed on your Scenario Overview page. To understand how we handle ILL, let's look at two parts of the equation: the expected number of **ILL requests** and the expected **ILL cost**.

## Forecasting ILL requests

We'll start by looking at how we forecast ILL requests for a single journal--let's call it *Journal X*. The approach has a few steps:

We start with the number of views

*Journal X*received last year, using data from your uploaded COUNTER report.We predict the usage of

*Journal X*on your campus for each of the next five years. To do this we use an algorithm to forecast the growing (or shrinking) of interest in this journal, based on recent publication and readership trends.For each of the next five years, we now generate a breakdown of how much of

*Journal X's*viewership can be fulfilled instantly, via your perpetual access rights and via open access. Whatever's left over after that is a turnaway, and must be fulfilled via ILL (unless of course you have chosen to subscribe à la carte to*Journal X*).We

*could*naively assume that every turnaway creates an ILL request. But we know that readers often procure paywalled articles by other means, or simply find replacements. In fact, the literature suggests that under 5% of turnaways generate ILL requests. To model this, Unsub simply multiplies the number of annual expected turnaways by 5%; this conservatively estimates the number of ILL requests expected for*Journal X*each year. (You can change this 5% figure in the scenario menu under "Parameters ➞ ILL ➞ ILL frequency, as percent of delayed access.")

That's it! Using these steps for *Journal X* we can predict the number of ILL request in each of the next five years. By giving all of the journals in your scenario this same treatment, we can estimate ILL requests across the whole package. Generally we'll display the results as a five-year average (rather than year-by-year) in order to simplify the report.

## Forecasting ILL cost

It's pretty easy to estimate the cost of ILL, once we have the number of ILL requests. We just need an estimate of the cost per request, and use that as a multiplier. Of course, this estimate needs to be the *overall* cost per request, averaging in both variable costs (eg. transaction fees) and continuing fixed costs like staff salaries.

Helpfully, there is a large body of literature estimating the the overall cost-per-request of ILL, which is summarized here. From that, we find that $17 per request is the most-cited figure, and so that's the value Unsub uses by default. Of course, if you know your own library's cost per request is different from this, you can change that $17 figure via the scenario menu ("Parameters ➞ ILL ➞ ILL transaction cost.")

Now that we have an average cost per ILL request, we just multiply that number by the number ILL requests forecasted for each of the next five years; that gives us forecasted ILL costs for each of the next five years. Again, we'll generally report the *average* annual cost over the next five years for simplicity's sake.