I’m an investor in Algolia so take my comments with a grain of salt.
But also, we deployed Algolia to SaaStr fairly early in their WordPress development (David Skok also did the same on his For Entrepreneurs blog which serves a similar audience), and we are a WordPress blog running truly at scale (many many millions of views), so the case study should be helpful:
Here’s what I’ve learned:
Algolia is obviously wicked fast. It’s epically faster than the out-of-the-box WordPress search we had before (not sure what technology that was), and about 100x faster than the native search on a different website we have on Squarespace.
It gets to great content very quickly in the search, & typo tolerance works well. See above, it’s clearly found the content people want even before the search is done. “Hire a great” in a fraction of a second -> the content most people really want, which is “How to Hire a Great VP of Sales”.
You do need to tag your content for it to work most effectively. We have so much archival content on Home – SaaStr (3000+ pieces) and a lot of the old content was never tagged and/or is not tagged properly. This stuff does not come up high on the Algolia search. I suspect this untagged, older, long tail content might surface more often (for better or worse) using other technologies.
Overall, it was a huge upgrade and you can just demo it at Home – SaaStr
It probably is important to have your content categorized at least half-decently to get the most bang for your implementation, I think. We’ve also never touched the settings since deployment. We should go back and do that, I think. But we haven’t needed to.