Skip to main content
LLM Metrix
Search · Google measurement

Your Google numbers,
derived honestly.

Search Console is first-party measurement of a property you verified, the trust anchor the AI numbers sit beside. Connect it and this page fills with your real clicks, impressions and rankings straight away, on every plan.

Start freeRead the guide

Free forever plan. No card required

app.llmmetrix.com/dashboard/search
Illustrative, sample figures in the product's real layout
Google Search performance · 28 days
Clicks
12,480
Impressions
341,905
CTR
3.65%
Avg. position
8.4
Top queries
crm for startups1,204 clicks#3.2
acme crm pricing980 clicks#1.4
crm slack integration412 clicks#9.7
best crm remote teams268 clicks#11.3

CoverageThese queries account for 71% of your clicks. Google withholds queries searched by only a few people, so the list never sums to your total.

Works with the AI engines your customers use

ChatGPTPerplexityGeminiClaudeGrokMeta AIDeepSeekGoogle AI OverviewsMicrosoft Copilot

Why it matters

Search Console numbers, with nothing hidden.

Included on every plan

Search Console is available on free too, split out from the GA4 and Cloudflare connectors on purpose. It is the one source that shows a brand-new account something true on day one.

Real history on connect | in two windows

The first sync backfills roughly 16 months of DAILY TOTALS, so the trend chart is populated immediately. Query-level and page-level rows backfill about 90 days: they are far bulkier and Google withholds their long tail anyway. Both windows then extend day by day.

Impression-weighted average position

Position times impressions over total impressions, not a plain mean across rows. A plain mean lets a three-impression query pull as hard as a thirty-thousand-impression one, and it is the figure a marketer screenshots and challenges.

Query coverage, stated

Google withholds queries searched by only a few people, so the query table always undercounts the property total. That gap is reported as a percentage rather than papered over by summing the query table and calling it "total clicks".

Days that line up with Google's

The day window is computed in Pacific time, because that is what Search Console reports in. Deriving it in UTC shifts the boundary hours early for part of every day and silently disagrees with the tab open next to you.

Top queries, top pages, striking distance

The queries that earn the clicks and the landing pages that receive them, plus the queries sitting just off page one with real impression volume, the cheapest wins available, and a branded/non-branded split taken against itemised clicks, never the property total.

Mechanics

How it works, end to end.

  1. 01

    Authorise and bind a property

    Sign in with a Google account that can see the property, then bind it to your project. Nothing is estimated while the connection is unbound or unsynced, empty panels say exactly that.

  2. 02

    Backfill on the first sync

    The first sync pulls roughly sixteen months of whole-property daily totals plus about ninety days of per-query and per-page rows in one run, fired by the hourly schedule, sooner via Sync now.

  3. 03

    Derive the honest aggregates

    Impression-weighted position, CTR from summed counts, the query-coverage percentage and the itemised-click branded split are all derived from stored rows, never naively averaged, never fetched live at page load.

  4. 04

    Promote proven queries to prompts

    Select rows in the top-queries table and import them as tracked prompts: they land inactive, stay behind your plan's activation cap, dedupe case-insensitively, and pick up an estimated monthly volume where configured.

Who it's for

Who lives in these numbers

SEO manager reporting classic GSC metrics beside AI dataOne report finally carries both halves of visibility.

You own weekly or monthly organic reporting and have been bolting AI slides onto Search Console screenshots.

  • Stat cards for clicks, impressions, CTR and weighted position sit beside the AI score trends instead of in a separate deck.
  • Every query row carries its previous-period Δ Clicks, computed from stored history rather than re-fetched.
  • The striking-distance list hands content priorities straight from the same window you report from.
  1. 01Connect Search Console in Settings → Integrations and bind the property to your project.
  2. 02Press Sync now on the Search page so the two-window backfill starts immediately.
  3. 03Build this month's deck from the headline four plus the striking-distance list.

Metrics this role tracks: Clicks / impressions / CTR · Impression-weighted position · Query coverage

Open Search performance
Analyst reconciling query coverage honestlyYou need the shortfall published, not hidden.

You reconcile dashboards against their source tools, and lose patience with vendors who sum the query table and call it total.

  • The query coverage percentage states itemised-versus-total outright, so reconciliation starts from the truth.
  • Missing previous-period rows render a dash rather than a fabricated zero.
  • Day windows are computed in Pacific time, so figures agree with the Search Console tab open beside them.
  1. 01Connect Search Console in Settings → Integrations and let the first sync land.
  2. 02Sum the top-queries table yourself and check it against the published coverage note.
  3. 03Spot-check impression-weighted position against Search Console for the same window.

Metrics this role tracks: Query coverage · Impression-weighted position · Clicks / impressions / CTR

Check the coverage note
Freelance consultant delivering monthly GSC reportsMonth one looks like year one, because the history arrives first.

You retainer-report across several client properties and rebuild the same Search Console exports every month.

  • Bind each client property to its project and the headline four populate from backfilled history instead of building up from zero.
  • Every query and page row carries its previous-period Δ Clicks, computed from stored rows, no re-pulling Google mid-call.
  • The coverage percentage answers the client's "why don't these sum?" question before it is asked.
  1. 01Create one project per client and connect each property in Settings → Integrations.
  2. 02Press Sync now per project so every backfill runs before the first reporting call.
  3. 03Import each client's top queries as tracked prompts to bridge into AI tracking later.

Metrics this role tracks: Clicks / impressions / CTR · Impression-weighted position · Query coverage

Open Search performance
Marketing lead who needs the Google baseline firstTrust the familiar numbers before trusting the new ones.

You are evaluating AI-visibility tooling and want a source of truth you already trust inside it on day one.

  • Search Console ships on every plan including free, populated with backfilled history from the first sync rather than week one.
  • The branded/non-branded split shows the demand you own versus the demand you compete for.
  • Once the baseline earns trust, the AI metrics rendered beside it inherit credibility.
  1. 01Connect Search Console in Settings → Integrations, free plans included.
  2. 02Verify one week of figures against your own Search Console tab after the first sync.
  3. 03Run a brand scan to render the AI metrics beside the baseline you just checked.

Metrics this role tracks: Clicks / impressions / CTR · Query coverage · Mention rate

Connect Search Console
Content strategist feeding the prompt backlogReal queries that earned clicks beat guessed prompts.

You maintain the tracked-prompt list and want it grounded in demonstrated search demand.

  • Select rows in the top-queries table and track them as prompts in one click.
  • Imports land inactive, so activation stays inside your plan's active-prompt cap.
  • Case-insensitive dedupe plus estimated monthly volume ranks new prompts alongside hand-added ones.
  1. 01Open the Top queries table on the Search page.
  2. 02Select proven queries and track them as prompts, imports land inactive.
  3. 03Activate the imports from the Prompts page, inside your plan's active-prompt cap.

Metrics this role tracks: Tracked prompts · Clicks / impressions / CTR · Query coverage

Review tracked prompts

The headline four

What does impression-weighted position mean?

Average position here is impression-weighted: each row's position multiplied by its impressions, summed, then divided by total impressions. A query with three impressions cannot drag the average around the way it would in a plain mean across rows, and a plain mean is the single most common way to compute this number wrongly. It is also exactly the figure a marketer screenshots, cross-checks against Search Console, and challenges when it disagrees.

The same arithmetic discipline runs through the rest of the panel. Click-through rate is computed from summed clicks and impressions over the window, never averaged across daily rows, a three-impression day with one click is a 33% day, and averaging it into a rate flatters nobody accurately. Rows without a position contribute nothing rather than counting as rank zero, which would pull the average toward an impossible number.

  • Headline figures derive from whole-property daily rows, the same reporting Search Console's own Performance tab shows you.
  • A previous-period delta with no underlying history renders a dash, never a fabricated zero.
  • Everything is recomputed at read time from stored rows, so opening the page never calls Google.

Reconciliation

Why don't my query clicks add up to my total clicks?

Five kinds of row are stored and deliberately kept apart: whole-property daily totals, per-query, per-page, per-device and per-country rows. They are never summed together, because Google withholds queries issued by only a few people, so the named query rows always undercount the property total. That shortfall is published as a coverage percentage rather than papered over; summing the query table and calling the result "total clicks" is quietly wrong in a way you can catch in the next tab.

The branded versus non-branded split takes the same care. Shares are computed against itemised clicks, the rows actually listed, never against the property total, because Google's own filters never sum back to the unfiltered figure and a share taken against the total can never reach one hundred percent. Taken against itemised clicks, the split describes the demand you own versus the demand you compete for.

Getting started

What happens when I connect Search Console?

Connecting is a standard OAuth flow: authorise with a Google account that has access, bind the verified property to your project, and the first sync runs on the hourly schedule, or immediately, via Sync now. That first run backfills two windows at once: roughly sixteen months of daily totals behind the trend chart, and about ninety days of query-level and page-level rows, which are far bulkier and whose long tail Google withholds regardless.

Freshness has a known shape after that. Search Console's own reporting lags around two days, so every window stops two days short of today rather than pulling partial rows, and day boundaries are computed in Pacific time, the zone Search Console reports in, so figures line up with the tab you cross-check against. Once the backfill is in, later syncs re-pull only the recent tail; until a sync lands, panels stay empty rather than estimated.

FAQ

Search Performance questions, answered.

What does Search Performance tracking show?
Your Google Search Console numbers, clicks, impressions, click-through rate and average position, rendered beside the AI visibility metrics the rest of LLM Metrix derives, so classic search performance and answer-engine presence sit in one view. Every figure comes straight from Google's reporting for the property you connected; none of it is modelled.
Which plans include Search Console?
All of them, free included. It is deliberately separated from the GA4 and Cloudflare connectors, which sit on higher plans: Search Console is the only connector that backfills real history on its first sync, and it costs nothing to serve because the data, the API and the quota are Google's.
What do I need to set up GSC reporting here?
A Google account with access to a verified Search Console property, then a standard OAuth connect: you bind the property to your project, and the first sync runs on the hourly traffic-sync schedule or immediately via Sync now, pulling the backfill described below. Until that sync lands, panels are empty rather than estimated, nothing is invented to fill them.
Why don't my top queries add up to my total clicks?
Because Google withholds queries issued by only a few people, so the per-query rows never sum to the whole-property figure. Five kinds of row are stored and kept apart: whole-property daily totals, per-query, per-page, per-device and per-country rows, and they are never summed together. The shortfall is shown as a coverage percentage.
How far back does the history go on the first sync?
It depends on the row kind, and the difference is worth knowing before you connect. Daily totals, the trend chart and the coverage denominator, backfill roughly 16 months, Search Console's full retention window. Query-level and page-level rows backfill about 90 days, because they are far bulkier and Google withholds their long tail regardless. Both windows then grow forward from there as syncs run.
How fresh is the data?
Search Console's own reporting lags by around two days, so the window deliberately stops two days short of today rather than pulling partial rows, and the page tells you the most recent day it has complete data for. Syncs run on a schedule and only re-pull the recent tail once the initial backfill is in.
Can I turn my best GSC queries into tracked prompts?
Yes, select rows in the top-queries table and import them as tracked prompts in one click. Imported queries land inactive until you switch them on, so activation stays inside your plan's active-prompt cap; they are deduped case-insensitively against prompts you already track; and they carry an estimated monthly volume where a keyword provider is configured.
Does opening the page call Google?
No. The dashboard reads rows the sync job has already written, so a page load never depends on a third-party API being up. The sync job owns quota, pagination and token refresh.
Is this the same as the Rankings page?
No, and they are kept apart on purpose. Rankings means your position inside an AI answer; this page means your position in Google's results. They are different measurements of different surfaces, and merging them would corrupt the product's core metric.
How is this different from the Search-AI Gap page?
This page is the Google half alone, classic search performance from Search Console, reported straight. The Search-AI Gap page joins these stored queries to your scanned prompts and reports only the disagreements between Google and the AI engines. Work the striking-distance list here; work the ranked-but-not-cited list there.

Keep exploring

Related features

Search-AI Gap

Join your Google Search Console queries to your scanned prompts and see the two places the surfaces disagree: ranked but not cited, and cited but not ranking.

AI Traffic

Which AI crawlers read your site and which AI answers send visitors back. Every crawler hit is marked verified or unverified against the operator’s published IP ranges or reverse DNS.

Multi-Engine Monitoring

Track your brand across ChatGPT, Perplexity, Gemini, Claude, Grok, Meta AI and DeepSeek: the seven major AI models in one view, plus Google AI Overviews, Google AI Mode and Microsoft Copilot on a weekly cadence for paid plans.

In depth

AI Search and the End of the Session Metric

Your traffic can fall while your influence grows, and no dashboard built on sessions will show you the difference. Here's what to report instead, and an honest account of how much of the gap is genuinely unmeasurable.

8 min read

How to Audit an AI Visibility Vendor's Numbers

Every tool in this category will show you a confident dashboard. Nine questions separate the ones measuring something from the ones generating something, and we've answered all nine about ourselves at the end.

9 min read

One Brand, Seven Markets, Seven Different Answers

A global brand measuring AI visibility in one region is reporting a number about one market with a label implying all of them. Here's why answers diverge by market, what a region setting actually changes, and how enterprise governance usually gets this wrong.

8 min read
Start on the free plan. No card, no time limit.

The AI era of search
is already here

See what AI engines say about your brand before your competitors do. Start free today. No card required.

Start freeBook a demo