On-site search is a crucial feature for type distributors. A potentially overwhelming number of fonts are presented to designers, and making it easy to narrow thousands of fonts down is part of what these sites aim to offer.
I don’t think this is the case for solo and even medium-sized foundry sites, where the catalogue is an already digestible size.
It’s not a perfect parallel, but the independent user experience research organisation Baymard came to a similar conclusion for niche direct-to-consumer ecommerce sites:
While users may be drawn to search on larger sites that are likely to appropriately invest in a robust and fully featured search tool, prior experience with poor quality or unoptimized search on other small retailers may dissuade users from relying on search for product finding.
Indeed, using search as a last resort when navigating or browsing has failed — only to be disappointed with no or low-quality results — may be a worse overall experience than if search is not offered at all.
Again, this has nothing to do with incoming search that gets people to your site in the first place; this is purely for keyword searching the catalogue once you are already on their site.
So: direct-to-consumer companies with a smaller collection of specialty products don’t appear to benefit as much from having on-site search. Baymard also found that one of the only things people were searching for on these sites were “non-product” pages, ex. “warranty,” which may or may not be part of your site’s default approach to search. This is relevant for foundries too: if anything, is likely much more valuable to be able to keyword search a large collection of blog or support articles, than the details of a typeface—and depending on your platform, this might not be how the built-in search would work.
Their conclusion is that you “should invest elsewhere”—and given the choice, it’s much more valuable to be able to filter a medium sized catalogue rather than search it.
Even that can wait until a catalogue reaches a certain size. I don’t have a one-size-fits-all rule for what that size is, but: if the scope of your typefaces similar in script and character coverage, and choosing almost any filter is going to take the catalogue down to a single result, it’s probably still too early.
Let me know your experience with this, as a type-founder or a type-finder.