Site search results limit capped at 85?

We have been trying to implement the Sitesearch function on our site and have encountered a few strange things.

First, the limit attribute does not seem to work above 85. Is this a hard limit? Would be cool to raise it.
Second, the text indexing may be missing words at the end of sentences, maybe because of the closing period. Not sure about this as pages might be missing of the 85 limit, but I thought I’d ask your opinion, two birds, one stone.

Thanks!

@mat_jack1 @s.verna can you guys look into this sometime?

Hi @sharedcodingsans! The enforced limit is 100, so if some results are not present the reason has to be found elsewhere.

If you send us more details here (project, exact query, missing results) we can try to investigate a bit more: Support page

In general, we released DatoCMS Search years ago, way before Algolia published its site crawler, which is way more powerful than ours! GitHub - algolia/algoliasearch-netlify: Official Algolia Plugin for Netlify. Index your website to Algolia when deploying your project to Netlify with the Algolia Crawler For this reason we’re maintaining our Search functionality, but we’re not investing any more time in it.

Thanks for the input. The 100 limit is weird as I always get max 85 results in datetime descending order.

The algolia plugin sounds cool, but we are using vercel / nextjs so i think this is not a real option for us. Or is it? Or maybe you have any recommended alternative for our environment, some free 3rd party solution?

We’re also using a Next.js stack, but we haven’t built the site search component yet. I’ll report back if we either succeed or run into issues.

Algolia does have a free plan (10,000 search requests/mo). It doesn’t include the crawler, which means you’ll have to index your resources yourself. But it should be easy with graphQL.

Backend: DatoCMS → graphQL → simple script → Algolia JS SDK → Algolia index

Frontend: Algolia index → your frontend, probably using the react-insantsearch component

1 Like

Thanks for the insight Roger, sounds like a pretty doable stack.

Will take it up with the client, if you guys run into any trouble along the path, I would be grateful for a headsup. Will do the same.

Will do, and likewise! Thanks.

PS there are also open source alternatives to Algolia like TypoSense and MeiliSearch, but you either have to self-host them or deploy to their supported clouds but manage scaling yourself (CPU/bandwidth/etc.). If you’re going with Vercel you’re probably interested in serverless to begin with (same here), so the idea of managing more infrastructure just for search seems… not ideal.

I’ll definitely report back if we find any other solutions later, or if Algolia works out.

You really should put this on the documentation for this feature so people don’t waste time trying get things fixed/changed if it’s not working for them.

1 Like

hi @jacob1 thanks for pointing it out. I’ll update the doc!

1 Like