Google Is Ending Full-Web Programmable Search in 2027: A Webz.io Migration Guide

If your product uses Google Programmable Search Engine (PSE) to search across the web, plan for a change before January 1, 2027. Google’s announcement is about full-web search: it says existing engines configured for “Search the entire web” can continue only until that date, and new engines have been restricted to “Sites to search” since January 20, 2026.

That does not mean every PSE setup is going away. Google says the Search Element can still search a list of up to 50 domains. For Custom Search JSON API users who need up to 50 domains, Google points to Vertex AI Search. For full-web use cases, Google asks customers to register interest in its full-web search solution and learn about its capabilities and pricing. Read Google’s transition announcement before choosing a replacement.

First, check what your current search does

The right migration depends on how your application uses PSE:

• Search across your own site or up to 50 selected domains: Google’s site-specific Search Element may still meet the need.
• Search across the full web: Google says the current option ends January 1, 2027. Compare alternatives against your actual queries and source requirements.
• Call the Custom Search JSON API: Check whether the application searches a defined set of domains or relies on full-web results. Those are different migration paths.

Google’s announcement also describes a full-web solution for partners who need its entire index, with a contact form for details. If that could be the closest fit, ask Google about availability, capabilities, and pricing while evaluating other providers.

Where Webz.io fits

Webz.io is worth evaluating when the job is to find and work with open-web content through an API—especially recent news. Its News Search API accepts a natural-language topic or question and can find articles by meaning, not only by matching the same keywords. Results include article titles, URLs, publication times, relevant excerpts, match scores, and source metadata. Filters can narrow by language, publisher domain, source country, date, topic, named entities, and other fields.

The boundary matters: Webz.io News Search is a news-search product, not a copy of Google’s entire web index or a drop-in replacement for the PSE search box. Its documentation describes News Search coverage as the last 30 days. If your product needs arbitrary pages across the whole web, a longer archive, or a specific domain set, validate that use case and the relevant Webz.io product before committing to a migration. Webz.io also offers separate APIs for news, blogs, forums, and reviews; their coverage and query behavior should be assessed for the content types you need.

Make a first search

You can call the News Search API directly or use Webz.io’s Python client. The example below searches for recent supply-chain disruptions affecting European automakers and limits results to English-language sources from Germany, France, or the UK.

Install the client and set your token on the server:

pip install webzio-news-search
export WEBZ_API_TOKEN="YOUR_WEBZ_API_TOKEN"

Then call the API:

from webzio_news_search import WebzNewsSearch

client = WebzNewsSearch()  # Reads WEBZ_API_TOKEN

response = client.search(
    "recent supply-chain disruptions affecting European automakers",
    k=10,
    days=7,
    language=["english"],
    country=["DE", "FR", "GB"],  # Source country
)

for result in response:
    print(result.score, result.title, result.url)
    print(result.text)

The days=7 argument sets a recent publication window. The result excerpt is useful for a results page or for passing relevant evidence to a downstream model; it is not the full article body. The response also includes fields such as total_results and credits_used. Keep the API token on your server, and use the returned URL and publisher metadata to make sources visible in your interface. See the News Search quickstart, filters, and response format.

If you are building an AI agent, Webz.io also documents a News Search MCP server. MCP changes how an agent connects to the search tool; it does not change the underlying data, search window, or API permissions. See the MCP setup guide.

A practical migration plan

1. Inventory your PSE setup. Record whether each engine searches the entire web or a domain list, whether the app uses the embedded Search Element or JSON API, and which filters or result fields the product depends on.
2. Choose the matching search scope. Keep site-specific search in place where it still meets the need. For news discovery, test Webz.io News Search. For blogs, forums, or reviews, assess the corresponding API.
3. Put an adapter behind your existing interface. Have your server send the query and filters to the selected API, then map the response into your product’s existing result cards. A data API gives you results to render; you own the search box, ranking controls, pagination, and presentation.
4. Replay real queries before switching traffic. Compare relevant results, source coverage, freshness, languages, response time, and total cost. Include queries that currently return few results, and check how each system handles domain restrictions and dates.
5. Roll out with room to fix gaps. Start with a limited share of searches, review misses and empty results, and complete the migration before Google’s January 1, 2027 cutoff if you rely on its full-web option.

Choose by the search job

For recent news search, Webz.io gives developers a contextual API that returns structured, source-linked results and excerpts. For a search product that depends on Google’s entire index, first define the sources and time range it must cover, then test providers against representative queries. Similar-looking search boxes can rely on very different indexes and return different results.

Documentation

• Google: Updates to our Web Search Products & Programmable Search Engine Capabilities
• Webz.io News Search API
• Webz.io News Search Python client
• Webz.io APIs documentation