Connecting your data
Granting access to Search Console is where most owners stop. This page describes what is read, what it produces, and what is never done with it.
Short answer
Connecting Search Console to an SEO audit tool means granting a read-only Google authorisation. Atelier SEO reads the impressions, clicks, positions and queries of the sites already verified in your account. Nothing is written, nothing is resold, nothing trains a model. Access is withdrawn in one click, from the platform or from your Google account.
What is read
Two Google authorisations are requested, never more. Both are read-only: the platform consults, it does not write.
The scope is bounded by Google itself. A read authorisation technically cannot modify a property, delete data or submit an indexing request.
The authorisation covers the properties you have already verified. It is used to reconstruct the real trajectory of your pages in Google results.
The authorisation covers the properties you declare. It is used to connect a position gained to what visitors do next on the site.
The three moments
Data comes in, a deliverable comes out, and a boundary stays closed. The table below holds for the two Google connections requested.
| Connection | What is read | What it produces | What is never done |
|---|---|---|---|
| Google Search Console | Impressions, clicks, average positions, queries, by page and by period. | Pages losing positions, queries stuck on page two, gaps between impressions and clicks, action plan priorities. | No writing to the property, no resale, no advertising sharing. |
| Google Analytics | Sessions, users, traffic sources, declared conversion events. | Landing pages that convert, channels that actually feed the site, business value of the proposed work. | No cross-referencing with another client's data, no model training. |
| Site crawling | The public HTML of your pages, like a search engine. | A technical audit of more than 90 checks, internal linking, site structure. | No change to your site, no automatic publishing. |
The crossing
Search Console says what Google displays. Analytics says what the visitor does next. Crawling says what the page contains. Taken separately, those three readings describe. Crossed, they decide: they name the page to fix first. That is where data-driven SEO prioritisation starts.
One named page, one reason, one rank in the action plan
Search Console shows a page displayed in the results, with its impressions and its clicks. Analytics shows that visitors landing on it never trigger the conversion event you configured. On its own, each tool displays a value that alarms nobody. Crossing Search Console and Analytics names a page that attracts without producing anything, and puts it ahead of a page nobody sees.
Read in the Performance tab, which overlays landing pages coming from Google with the declared business actions.
A query already places you high in the results. The same address fails an audit check: a title cut off on display, an image too heavy, structured data missing. The position already earned makes the defect expensive, when it would stay secondary on a page nobody reaches. The crossing moves that page ahead of the others, because the fix pays off immediately.
Read by overlaying the Audits tab and the Performance tab; the action plan turns it into a dated line.
Crawling lists what your site already covers. Comparing it with the sites that surface on your queries lists what it covers nowhere. The gap is not a list of keywords to buy: it is a search intent none of your pages addresses, and that someone else addresses in your place.
Read in the Competition tab, which puts the queries you hold face to face with the ones you are missing.
The second crossing draws on the audit checks, and that is the only one of the three sources requiring no Google account at all: what the technical audit covers in detail describes the families of checks and what each one verifies.
These crossings are not a declared capability. The same question was asked twice on this site, without data then with it: the Iris demonstration on a real case reproduces both answers as they came.
The concrete steps
Granting access to Search Console takes under two minutes. No credit card is requested, at any step.
From your workspace settings, a button starts the authorisation. The page that opens is Google's own, never an intermediate form.
The screen names the read-only access to Search Console and to Analytics. You see the exact scope before accepting.
Only properties you already own or are a verified user of appear. An unverified property stays out of reach.
Reading starts immediately. How far back the history goes depends on what Google keeps for your property, not on the platform.
What comes out
Take the first crossing again, the page that gets impressions without converting, and follow it to its output.
The case is common and deliberately generic. What matters here is the path: three readings go in, one action plan line comes out.
Search Console surfaces a landing page piling up impressions and few clicks. Analytics shows that visitors arriving on it leave without triggering the declared conversion event. Crawling adds the reason: the title exceeds the width Google displays and gets cut off.
Rewrite the title so it fits the displayed width, placing the query that already generates the impressions. The expected gain is on click-through rate at constant position, not a jump in ranking.
The recommendation does not stay an isolated remark: it comes out in the Action plans tab, with its target page, its reason and its rank. Iris sets that rank, from the same data, which is what lets it answer about your site rather than in generalities.
What remains is who applies that line, and when: how the agency works describes the sequence, step by step.
The limits
These commitments are not declarative. They restate the Google API Services User Data Policy compliance disclosure published on this site.
Data obtained through Google APIs is neither sold, nor transferred, nor passed to a third party for its own use.
It is never used to target, segment or feed an advertising audience.
It trains no artificial intelligence model, general-purpose or internal.
Proposed fixes stay proposals. Nothing is published without your explicit approval.
The legal detail, the exact scopes, the retention period and the deletion procedure are published in two dedicated documents: the Google API Services User Data Policy disclosure and the privacy policy.
The objection
The question is legitimate, and the right answer is not a promise. It is a verifiable scope and an immediate way out.
The scope is visible on the Google screen before you accept. The way out is a button, and it depends on no email request.
One last point matters as much as the first two: the connection is not required to start. A full technical audit runs on crawling alone, with no access to your accounts.
Next
The audit starts from the site address, with no prior connection. It crawls the public pages and returns a score, ranked problems and an action plan.
The Google connection comes afterwards, and it changes the nature of the diagnosis. Without it, the audit says what is broken. With it, it says what is broken and what it costs you.
To see what that crossing produces over time, the site’s public journal tracks its own readings: the results measured on a real site. And to know who applies those recommendations, the three support plans describe the scope.
No. The technical audit works by crawling the public pages of the site. Connecting Search Console adds real impressions, clicks and positions, which allows prioritising by impact rather than by theoretical severity.
No commitment
The free diagnostic only needs a site address. You will see what the tool produces before deciding whether to entrust it with your Google data.