I am currently working on one of our publisher lists. Initially I left all of the pre-sets in Unsub as is, but see below #9. Here are the steps I've taken so far:
1. In Unsub, ranked by CPU and set instant fill rate at 75%. This gives a base set of 80 "selected" subscriptions shaded blue. The 75% is an admitted arbitrary set point, but since it appears to give a certain level of critical savings against our huge budget cuts, seemed like a good place to start.
2. Exported to Excel, and shaded the 80 starter titles in blue on export sheet.
3. Sorted export sheet by OA %. Flagged selected high (50%+) OA access titles for further review (aggregator db availability and embargo periods; YOP latest 5yrs review).
4. Sorted export sheet by Usage. Usage metric incorporates downloads, our authors' citations to it, and our author's publications in it. Flagged high usage titles (down to 120/year) that did not get selected for further review (same).
5. Sorted by CPU rank, and secondarily by Usage. This brings out lower use titles from original selection set we could perhaps do without.
6. Flagged low CPUs under $30 with low OA accesses. Possible adds to subs.
7. Look up coverage of remaining uninvestigated blue selected titles to eliminate any sufficiently covered in aggregators based on usage during embargo period.
8.
Sort by actual download-only data field and review. I found a number of titles had fairly low actual downloads, that received a high usage rank due to our authors' citations to the journal, or our author's publication within it. To me it seemed that in some, but not all, cases, the weighting of local author data skewed the importance too far toward selection.
I am just winging it here, and am probably undoing or duplicating work that is already represented by the straight data, but so far it seems like adding in aggregator and embargo data, as well as year-of-pub COUNTER data in conjunction with embargo periods, is a necessary step. Out of the initial selection of 80 based on CPU rank, I will DE-select 25 and possibly also ditch another 10 or more. The other steps also indicated titles we should add that were not in the initial group. So once again a huge messy spreadsheet with lots of notes, for final decision making.
The one aspect of the Unsub data I don't fully understand is ILL costs listed on the title display when it has been selected, but will go back and review the video or ask later if it seems critical.
Deb