Google to Sunset Custom Search JSON API in January 2027, Prompting Developer Migration
With the API scheduled for discontinuation, developers face a nine-month window to transition to third-party SERP providers or open-source actors that replicate the original JSON schema.
Google has confirmed that the Custom Search JSON API (CSE) will cease operations on 1 January 2027, ending a service that has provided developers with public web search results in JSON format since its launch in 2006. The shutdown notice was issued in January 2026 via a Google Developers blog post and console email notifications, with the API remaining technically operational as of August 2026. This timeline provides approximately nine months for developers to migrate their integrations before the service returns an HTTP 410 Gone error.
While Google recommends Vertex AI Search as a replacement, the company’s enterprise-tier product is designed for semantic search over internal corpora rather than public web search. Consequently, Google’s official guidance directs developers to explore third-party providers for web search at scale. The recommended alternatives include established SERP (Search Engine Results Page) providers such as SerpApi, ScaleSerp, and Bright Data, though each emits a distinct JSON schema that typically requires significant code refactoring to replace the standard CSE structure.
To facilitate a smoother transition, open-source solutions have emerged that emulate the original CSE JSON schema. One such actor utilises Apify’s GOOGLE_SERP proxy to scrape results, allowing developers to maintain compatibility with existing code that parses fields such as searchInformation.totalResults and items[].title. This approach reportedly requires minimal code changes, potentially as few as two lines, while handling complex backend requirements including CAPTCHA rotation, IP reputation management, and TLS fingerprinting.
Pricing structures for these replacements vary significantly from the legacy CSE model. The open-source actor operates on a pay-per-result basis at approximately $0.005 per query, with no per-engine cap. For a mid-size application performing 100,000 searches per month, this model costs roughly $500, comparable to historical CSE costs at that volume. In contrast, traditional CSE users relied on pooling multiple engine IDs to bypass daily throughput limits, a strategy that will no longer be viable once the API shuts down.
Market reaction to Google’s technical roadmap adjustments has historically been sensitive. Previous reports of Google AI delays triggered substantial market slumps, underscoring the fragility of investor sentiment regarding the company’s infrastructure decisions. As the January 2027 deadline approaches, developers must weigh the compliance and stability risks of third-party proxy reliance against the operational costs of migrating to alternative SERP providers.
