Is fallback on by default?

No. Requests use Google only unless they send engine: "auto" (or a list of engines) or the API key has fallback switched on. An explicit engine always wins over the key setting.

When should I keep Google only?

For rank tracking, SEO audits and anything you compare against google.com. Positions from other engines come from their own indexes and aren't comparable.

Does a fallback cost more?

No. Every engine costs the same credits as Google, and blocked, failed and empty attempts along the route are free.

Can I choose which engines it may use?

Yes. Send a list such as ["google", "brave"] and only those engines are tried, in SerpKite's route order.

engine=auto

Search API with automatic fallback

For agents and chatbots, some answer beats no answer. engine: "auto" tries Google first and, only if Google is blocked or times out, answers from the next engine that serves the endpoint.

Get a free API key Choose search providers guide

How it works

Google first; another engine answers only when Google fails.

Behind every Google request SerpKite already retries internally before giving up. That failover is always on, costs nothing and never changes the engine. Fallback goes one step further: with engine: "auto", if Google is still blocked or times out, the next enabled engine that serves the endpoint answers instead.

The response keeps the same shape, so your parser keeps working, and it says what happened: meta.engine names the engine that answered and meta.route lists each attempt with ok, partial, empty, blocked, timeout or error and its duration.

Parameter errors (a bad country, an unknown parameter) never trigger a fallback: they return 400 straight away. And a Google-only request is only ever served a cached Google result.

Fallback in one parameter

cURL

TypeScript

Python

curl https://api.serpkite.com/v1/search \
  -H "Authorization: Bearer $SERPKITE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"q":"best espresso machine 2026","country":"us","engine":"auto"}'

Google blocked, Brave answered · illustrative

{
  "request": {
    "endpoint": "search",
    "engine": "auto",
    "q": "best espresso machine 2026",
    "country": "us"
  },
  "meta": {
    "credits_used": 1,
    "engine": "brave",
    "route": [
      {
        "provider": "google",
        "outcome": "blocked",
        "ms": 2140
      },
      {
        "provider": "brave",
        "outcome": "ok",
        "ms": 612
      }
    ],
    "parse_quality": "ok"
  }
}

Fallback order per endpoint

Google always goes first. Engines are tried in route order and only if they serve the endpoint.

Fallback order per endpoint
Endpoint	With engine=auto
`/v1/search`	Google → Brave → Bing → DuckDuckGo → Yahoo → Yandex
`/v1/news`	Google → Brave → Bing → DuckDuckGo → Yahoo
`/v1/images`, `/v1/videos`	Google → Brave → Bing → DuckDuckGo
`/v1/maps`, `/v1/places`	Google → Yandex Maps
Reviews, shopping, scholar, patents, autocomplete	Google only

Why it matters

What you get

Agents keep working

A tool call that would have failed returns real results from another index instead.

Per-key switch

Turn on "Allow fallback" for a key in the dashboard when your MCP client or framework can't send engine.

Failed attempts are free

You're billed once, for the engine that answered, at the same price as Google.

Never relabelled

meta.engine tells you when a result didn't come from Google. Rank trackers can keep engine: "google".

FAQ