serp_snapshot

Cost: 5 credits per call plus 8 credits per keyword — 1 to 10 keywords per call, so 13 to 85 credits.

serp_snapshot measures where a domain appears in Google's organic results for each of your keywords — on one location, one language and one device — and stores every reading, so keyword_positions can show the series later. It is the only part of the rank tracker that contacts a search engine.

What it does

Pass a project_id (or any target domain) and up to 10 keywords. Each keyword is a separate live search, so the list is what the call costs — that is why the price has a per-keyword part. Duplicates are refused rather than quietly removed, and the list is never trimmed to fit: a shorter answer to a longer question is not the answer you asked for.

Who can run it

serp_snapshot needs a paid credit balance. Each keyword is a separate live search bought from a paid third-party provider, so it is not available on trial credits, however many are left; the refusal arrives before anything is reserved and says outright that you were not charged. Buy any credit pack and it unlocks straight away — your existing credits are untouched and keep working for crawls, audits, reports and Search Console tools.

This is the one part of the rank tracker that is gated, and gating is a separate question from price. track_keywords is free; keyword_positions charges for the analysis but contacts no search engine. Neither is gated — both run on a trial account, and each page states its own cost.

If live DataForSEO access is unavailable on this deployment, the tool says so and charges you nothing — no credits are reserved or spent. SeoGrep never returns sample or placeholder positions as if a search engine had really returned them.

Three answers, and none of them is a number you can misread

A keyword comes back as exactly one of three things, and they are never collapsed:

• Found — with DataForSEO's own rank_group (its rank among organic results) and rank_absolute (its rank among every element on the page, including featured snippets and ad blocks). Both are reported where the vendor sent them, because they disagree whenever a SERP feature sits above the result and that gap is itself the finding; either one the vendor left out is stated as not reported rather than filled in.
• Searched for and not found — together with how many organic results were actually examined. That is the scope of the claim: it is not position 0, and it says nothing about results beyond the ones counted.
• Not measured — the request or the response failed, so the position is unknown. Nothing was examined, so this is not a statement that the domain is absent.

SeoGrep computes no visibility score, no share of voice and no ranking of its own.

What else was on the page, including AI Overviews

Every measured keyword also reports the SERP features DataForSEO listed for that page — a featured snippet, a "people also ask" block, a video carousel, an AI Overview — counted and named under the vendor's own identifiers. Nothing is renamed into friendlier words, because that would be inventing a mapping onto a taxonomy SeoGrep does not own, and nothing is dropped for being unfamiliar: a feature type the vendor ships tomorrow reaches the page tomorrow. Where a page carries more features than one line holds, the extras are counted, never quietly discarded.

This is also the direct explanation of a gap between rank_group and rank_absolute: the features sitting above your result are what separate the two numbers.

An AI Overview here means one thing only — DataForSEO reported the block on that results page for that keyword. It does not say whether your site is cited inside it. Google builds its AI features by query fan-out: it runs further searches of its own across sub-topics and draws on a wider set of pages than the single keyword measured here. So one keyword's snapshot is not a measure of a site's AI visibility, and the answer says exactly that wherever it reports one.

What a reading is scoped to

A position is a measurement at a moment, not a property of a site. Every reading is taken on one search engine, one location, one language, one device and to one fixed depth, and the answer states all of them — Google returns different results and a different layout on desktop and on mobile, so a desktop reading says nothing about a mobile one.

A result counts as yours only when its host matches yours exactly, after lower-casing and removing a leading www.. A subdomain does not count: blog.example.com is not example.com, because "our blog ranks" and "our site ranks" are different findings and only you know which one you asked about.

It measures on demand, and only on demand

Nothing here runs on a schedule. A snapshot happens because you asked for one, so no credits are ever spent while you are not looking. track_keywords records which keywords you want watched — that is free and takes no measurement — and this tool is what turns a watched keyword into a reading.

Example

Ask your MCP client in plain language:

Take a SERP snapshot for "seo tools" and "rank tracker" on my example.com project, on mobile.

Run list_projects first if you need the project_id.

Input

Returns

One block per keyword, in the order you passed them, each carrying its own answer of the three above — and, where the domain was found, every placement it was found at with DataForSEO's own ranks and URLs. The reply states what the snapshot was measured under (search engine, location, language, device, depth requested and how a domain was matched), separates DataForSEO's own account of when it measured from SeoGrep's clock, and confirms how many readings were stored.