Coverage

Coverage is the share of the vocabulary catalogue that has at least one clip indexed. It is the number that decides whether a feature built on this API feels complete or feels broken.

curl https://clips.vocaflare.com/v1/corpus

{
  "languages": [
    { "language": "en", "clips": 252, "channels": 56, "words": 8640, "wordsCovered": 2758, "coverage": 0.319 }
  ]
}

Read it before you build, and read it again before you promise a feature to your users.

Designing for partial coverage

Coverage rises daily but will never be total, and a feature that assumes a clip exists for every word will fail in front of a user. Two approaches work:

Ask first, render second. Query GET /words with hasClips=true for the words in your lesson, and only show the "watch it used" affordance on the ones that came back. The empty state disappears because it never renders.

Degrade in place. Show the affordance, and on 404 replace it with the didYouMean suggestions. Cheaper to build, slightly worse to use.

Coverage is not uniform

The corpus is built from real speech, so frequent words are covered far better than rare ones. A B1 word list will feel well covered; a C1 academic list will not. Sample your own vocabulary against GET /words rather than reasoning from the headline percentage.

It moves in one direction, slowly

A daily run adds clips. A word that has no clips today may have three next month, and clip counts for covered words keep rising. If you cache coverage, refresh it daily; if you cache a "no clips" answer, give it a short life.