Google’s Custom Search API Has an Expiration Date

Google has put a hard expiry date on its Custom Search JSON API. It’s the server-side service behind applications that send a query to a configured Programmable Search Engine and receive web or image results as JSON. The API is closed to new customers, and existing integrations must move to another solution by January 1, 2027.

The service was easy to treat as plumbing. Existing customers get 100 free queries per day, then pay $5 per 1,000 queries, with a limit of 10,000 queries per day. That can be a tiny line item in a product’s budget while still sitting deep inside its search, research, monitoring, or content workflow.

Google’s January 20, 2026 announcement gives those customers two routes. Vertex AI Search is the recommended option for searches covering up to 50 domains. Customers that need full-web search are sent to an interest form for a separate solution. Google says that solution is available, but the public transition page does not provide its pricing or promise compatibility with the old API.

The successor changes the product shape

The published Vertex AI Search route involves more than exchanging one endpoint for another. Google Cloud’s current website-search setup starts with a website-content data store that must be attached to a search app. Basic website search uses Google’s existing index for the sites you include. Advanced website indexing adds features and control, along with extra cost and a domain-verification requirement.

Those may be useful capabilities, but they create a different operating model from an API key plus a Programmable Search Engine ID. The transition announcement does not describe a drop-in migration. A team has to decide what corpus it wants searched, which product tier fits, how the new response maps into its application, and whether the result quality still matches what users expect.

The first boundary to check is 50 domains. A documentation search across a known set of company sites may fit the recommended path neatly. A research tool that searches a broad or changing slice of the web has a different problem and needs details about Google’s separate full-web offering or another provider. Image search deserves its own check because the old API supports web and image results, while Google’s announcement describes the alternatives mainly by domain coverage.

Then inspect the assumptions that grew around the old response. Which fields does the application render? Does it depend on pagination, filtering, language or country parameters, custom ranking, or a particular quota pattern? What happens when results arrive in a different order or the service is unavailable? The replacement decision should follow that inventory. A successful API call can still produce a broken product experience.

This shutdown exposes the difference between consumption cost and switching cost. A team can spend almost nothing on Custom Search JSON API and still owe substantial product and engineering work when its scope changes. Google has supplied a deadline. The practical move is to document the dependency now and test replacements against real queries, because a small API bill will not make the migration small.

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com