Every tag on APIs.io carries a quality lift — how much better, on average, the providers wearing that tag score than the catalog as a whole. It is a small number sitting in the tag record, and we had never ranked it.
Ranking it across the sixty largest tags turns out to answer a question people actually ask us: what kind of company has a good API?
Not the one you would expect.
The ranking
72.6 open-source posture
54.6 webhook capability
51.1 software-as-a-service posture
45.1 mcp capability
43.4 agents market + capability
43.3 authentication capability
41.2 marketing market
41.2 graphql capability
38.4 developer-tools persona + capability
35.3 analytics market
35.1 payments market
...
12.8 consumer persona
10.5 pharmaceuticals market
10.2 biotechnology —
3.5 acquired posture
2.1 defunct posture
The top is almost entirely capability and posture tags — things that describe what a company does technically. The bottom is industry tags — things that describe what market it sells into.
acquired and defunct at the floor are the control. Those are postures that should predict a worse API, and they do, which is a small sign the number is measuring something real.
Why this is not obvious
It would be reasonable to guess the opposite. Financial services is a regulated, integration-heavy industry with decades of API practice; biotechnology is full of extremely sophisticated companies. You might expect those industries to carry high-quality API surfaces.
They do not, and the reason is worth being precise about.
A capability tag is a claim you can only make by shipping something. You cannot carry webhook without a webhook. You cannot carry graphql, or mcp, or authentication, without publishing the thing. The tag is downstream of the work, so the population it collects is self-selected for having done the work.
An industry tag is a claim about who you are. biotechnology sits on 858 companies, and identity_share on that tag is 1.0 — every single one claims it as their own identity. It is accurate. It is also nearly free to be true.
The honest caveat
The low end of this ranking is not a judgement about those industries. It is substantially a fact about what an API catalog can see.
Of the 858 companies tagged biotechnology, the number of APIs carrying that tag is one. Not one per company — one. The companies are real, the tag is correctly applied, and most of them simply do not publish a public API, because a clinical-stage therapeutics company has no reason to.
So biotechnology scoring 10.2 does not mean biotech builds bad APIs. It means the tag collects a population that mostly does not build APIs at all, and the handful that do — Metabolon, Element Biosciences, Biodock — are averaged against hundreds that never will.
That distinction matters for anyone reading these numbers as a market signal. Low lift on an industry tag is a coverage fact. Low lift on a capability tag would be a quality fact. They are not the same measurement even though they share a column.
What we are doing with it
Two things.
It changes which tags are worth investing in. We are working through the catalog’s tags one at a time — checking who is in them, whether the ranking makes sense, and what is missing — and this says where that effort pays. A developer landing on webhook or mcp is looking at a population that ships. A developer landing on an industry tag is mostly looking at companies we can tell them very little about.
And it gives us a cheap detector for a tag that is not doing work. A tag with a large membership and a lift near the floor is a label that has been applied widely and predicts nothing — which usually means it is being applied loosely. We found one that way: consumer, on 1,359 providers at a lift of 12.8, with business-to-business companies carrying it. That is now filed as a defect rather than a curiosity.
If you want to poke at this yourself, every tag record on the tags surface carries quality_lift, identity_share, provider_count and api_count. The ranking above is four fields and a sort.