Skip to main content
ai-cmo.dev connects to Google Search Console and Google Analytics 4 to bring real search demand into the platform. These are not optional add-ons — they provide the raw material for prompt generation and the offline signal for SEO measurement.

​
What GSC provides

Google Search Console gives the platform’s most important external data: real queries that people typed into Google, together with impressions, clicks, and average position. Each GSC row records:
  • Query. The search term a user typed. This is the seed for prompt generation.
  • Impressions. How many times the brand’s pages appeared in search results for that query.
  • Clicks. How many times users clicked through.
  • Average position. The brand’s average ranking for that query.
GSC data is the demand signal. It shows what people are actually searching for. Without GSC, prompt generation would rely on generic or invented queries.

​
What GA4 provides

Google Analytics 4 measures behavior after the click. GA4 data includes pageviews, sessions, engagement, and conversion events. For SEO measurement, GA4 shows whether search traffic converts — whether users who found the brand through search stay, engage, or take action. GA4 is complementary to GSC. GSC shows the surface of search (impressions and clicks). GA4 shows the depth (what happens after arrival).

​
Property formats

The platform requires specific formats for each property. GSC property. Two formats are accepted:
  • sc-domain:example.com — a domain property that covers all subdomains and protocols. Requires DNS verification in Search Console.
  • https://www.example.com/ — a URL-prefix property that covers a specific protocol and path prefix. Can use HTML file verification.
The property is stored as gsc_property in the workspace configuration. GA4 property. A numeric identifier only. This is the numeric property ID found in GA4 admin settings, not the measurement ID that starts with G-. The property is stored as ga4_property. See Connections for the setup steps.

​
What each connector unlocks

Without either connector, the platform still measures GEO — prompting and answer-engine measurement do not require GSC or GA4. But prompt generation uses generic sources, and the SEO score is unavailable.

​
How GSC seeds prompt generation

The prompt generator reads GSC queries, filters for question-shaped text (natural language questions that match the platform’s intent patterns), and uses them as seed material. High-impression queries become higher-demand seeds through the demand scoring system. This creates a feedback loop: real user queries produce measurement prompts, which test whether answer engines address those same queries. If ChatGPT mentions the brand for a question that drives 10,000 monthly searches, that is a meaningful visibility signal. If it does not, the brand has a content gap. See Prompt sets for how seeds become curated prompts.

​
Connector setup

Both connectors require inviting the workspace’s provisioned AgentMail address to the Google property with Viewer access. The operator holds the Google OAuth credentials — workspace members do not enter Google API keys. Configuration is stored in brands/<workspace>.yaml as:
The platform tests configuration by checking for the presence of operator credentials and the configured property. It does not query Google during the test. The first actual data pull confirms access was granted.

​
Limitations

  • GSC data has a 2–3 day latency. Today’s queries appear in 2–3 days.
  • GA4 data has similar latency for standard events.
  • The platform reads GSC and GA4 data at most once per day.
  • A successful configuration test does not prove the invitation was accepted. Check the first data pull.
  • Only one GSC property and one GA4 property per workspace.
See Connections for the full setup guide.