For a public library, we get input from the public and also read Publisher's Weekly. We may get a curated title list from publishers on various subjects.
We regularly generate reports on what is being read and checked out, but those are mainly meant to determine what to remove from the shelves.
- Most librarians who are responsible for collection development in a given subject area have subscriptions to/access to blogs, newsletters, and publications covering upcoming books. There are also more general places like Kirkus[0].
- Reports analyzing what is checked out and/or read tells you what your audience is interested in/needs, and some of Librarianing(TM) is being able to analyze what those books have in common and seeking out new materials that might also meet that interest or need. We also (as mentioned elsewhere) pull reports on what hasn't been checked out and track down why if possible. Sometimes, librarians can also pull this information or request it from other libraries.
- Publisher reports: Multiple libraries I've worked in have all the staff go through publishers' upcoming titles and make notes of titles that would be useful in their subject area.
- Audits of the existing collection against standardized lists: the library world creates collection development lists and circles them around. Collections are also audited for outdated information.
- Bundling isn't uncommon when it comes to things like academic periodical/database access. Some library purchasing decisions look more like choosing a cable package than shopping at Barnes and Noble.
> Is it as simple as looking at these two lists to know what is popular? Do you care about the content of the books or just the general category? Do you look for best sellers?
Past popularity is only one variable to consider when it comes to collection development. For example, you have to break down general popularity versus popularity in the audience you're serving. Trend cycles in general also have to be considered: Just because something has been popular in the past doesn't mean it will be now - for both fiction and non-fiction, things go in and out of style. Think westerns and the Atkins Diet. Eventually you learn to recognize the cycle and plan your buying accordingly.
In addition, collection development requires considering the collection as a whole and how the material will age. For example, one problem with a lot of best-sellers/'airport reads' (think Danielle Steele, James Patterson, etc.) is that the demand curve is very skewed: The first few months one of the books is out you may have 300 people wanting it but after a year there's next to no demand. So how many copies you buy is going to depend not just on your own collection but what format they're in/what licensing you have for your e-book collection, whether or not you're in any consortia that can help alleviate the number of requests, etc.
Content matters, but so does category. In addition, as unpopular as it is to admit, so does ideology. Librarians are split on the question of whether curation can be a neutral undertaking, and some libraries have funds for staff recommended purchases regardless of analytics.
A positive to this is it helps expand what's available/introduces people to new things they didn't know they'd like and also can allow for collection development for a minority (numerically) population: basing everything on analytics and popularity also kills the library's ability to serve the long tail or act as an information repository for non-beginner information. The negative to this is that most librarians come from upper-middle class, urban, professional backgrounds due to the MLIS requirement and think the world as depicted in Slate, the Guardian, Vox, etc. is the only world which can massively skew collection development.
In the film world, there's often a big disconnect between critics and popular sentiment, such as when a RottenTomatoes critics score and the popular score are basically opposites. Does this happen frequently in the book world, and how would you go about measuring that?
The books that are good but for a small audience are the ones that it's hardest to decide on. In the geek world, think of books like Godel, Escher, Bach.
It's usually measured informally: You get an idea of what audience a tool is chasing or speaking to and then compare that to the audience you're providing materials for.
Also librarians should be decently well-read in their subject area, which means that they're also familiar with both internal and external criticisms, how the genre/category has been evaluated historically, etc. This means we can tell the difference between a sci-fi novel that's poorly written and one that is good/decent but appealing to a small audience. And between clickbait/politically motivated criticisms and criticisms of the work's structure or content.
Basically, part of the job is having an awareness of the meta-conversation and history of a category and reading the critics critically. This is also idealized and it gets much messier on the floor as we have less time for the 'non-productive' knowledge gathering than we used to.