Search volume is behind a login. Keyword Planner wants an active ads account and a human to sign
in to it — a wall an autonomous agent cannot climb. Here’s how an agent reads real demand
for a list of terms in one call, with no account anywhere.
§01 · demand, measured
Keywords in, volume out.
POST a list of terms and the keyword-volume stall returns monthly search volume, CPC, competition, suggested bid range and a twelve-month trend for each. That trend is the part guessing cannot replace: it shows whether a topic is growing, seasonal or already over. Pay-per-call over x402, no key.
keyword-volume
# Up to 200 keywords in ONE call — priced per call, not per keyword.
curl -s -X POST https://true402.dev/api/v1/keyword-volume \
-H 'content-type: application/json' \
-d '{"keywords": ["x402 protocol", "agent payments api", "pay per call api"]}'
# → { keywords: [ { keyword, searchVolume, cpc, competition,
# competitionIndex, bidLow, bidHigh,
# trend: [ { year, month, volume } ] } ],
# requested, returned, location, language, limits: [...] }
§02 · batch, always
200 keywords cost the same as one.
This endpoint is priced per call, not per keyword. Up to two hundred terms travel in a single request for a single flat price, so an agent should send its whole candidate list at once. Sending them one at a time buys the identical data for up to two hundred times the money — the most expensive mistake available on this endpoint, and the easiest to avoid. Over the cap the request is refused rather than silently truncated, because a caller who asked about 500 terms and quietly got 200 has no way to tell which vanished.
§03 · estimates, and it says so
What the number is.
These are banded advertising estimates reflecting ad-auction behaviour — right for comparing demand between terms and reading trend direction, wrong for claiming an exact count of humans. Every response ships that caveat in a limits[] field rather than assuming the caller knows it. A term the index has never measured comes back with no volume field at all, never a zero, alongside a requested and returned count — because zero would assert nobody searches it, which is not what an absence of data means.
§04 · for agents
No ads account, no key.
An unpaid request returns HTTP 402 with payment requirements; the agent pays USDC on Base and retries. Need candidates before you can measure them? Keyword ideas turns one seed into a list. Want to know what a domain already ranks for? Ranked keywords. New to the protocol? What is x402? Full reference: API docs.
§05 · questions
Answered for machines.
How can an AI agent get keyword search volume without a Google Ads account?
It calls a keyword endpoint that already holds the data. true402’s keyword-volume stall does this: POST a list of keywords to /api/v1/keyword-volume and it returns monthly search volume, CPC, competition and a twelve-month trend for each one. Keyword Planner requires an active Google Ads account and a human login, which an autonomous agent cannot pass. This is one HTTP call, paid in USDC over x402, with no account at all.
Is it cheaper to send keywords one at a time or all at once?
All at once, by an enormous margin — and this is the single most important thing to know about the endpoint. It is priced per call, not per keyword. Up to 200 keywords in one request cost the same as one keyword. Sending them individually buys the identical data at up to two hundred times the price, so always batch.
Are these volumes exact search counts?
No, and the response says so in a limits[] field. Search volumes are banded advertising estimates that reflect ad-auction behaviour, not a census of every search performed. They are the right tool for comparing demand between terms and for spotting trend direction, and the wrong tool for claiming an exact number of humans searched something.
What happens to a keyword the index has never measured?
It comes back with no search-volume field at all, rather than a zero, and the response reports how many terms you requested against how many came back. That distinction matters: zero would assert nobody searches the term, when the truth is that the index holds no measurement for it. The service will not turn an absence of data into a finding.
What can an agent actually do with search volume?
Decide what to write, what to name a product, and which of several phrasings people actually use. A content agent can rank candidate titles by real demand instead of guessing; a research agent can size a topic before committing to it. Pair it with the keyword-ideas stall to discover the candidates in the first place, and with ranked-keywords to see what a domain already wins.