Airtable SEO rank tracking: weekly Google ranks and AI Overview flags

The Airtable connector is Airtable's own MCP server at https://mcp.airtable.com/mcp, which lets an AI client such as Claude read and update your bases. Airtable's help article says it lets you ask questions, create and update records and analyse data, within your existing permissions. The Agent Index lists 12 tools: 5 reads, 6 writes and a ping.

Three limits matter here. Creating records is capped at 10 per request. Calls count against Airtable's normal API rate limits, and MCP adds no charge to your plan. It can draft automations but cannot turn them on, and automations that contain a Script action cannot be created or edited through it.

The usual answer comes from Airtable's own SEO workflow guide: a keyword research table with fields such as parent keyword, search volume, ranking difficulty, current position and target URL. Its examples of automation include refreshing rankings weekly and notifying an owner when a target keyword slips out of the top ten.

Data Fetcher's guide goes further. It describes Pages and Keywords tables, a separate time-series Performance table for trend reporting, and SERP APIs for rank tracking by keyword and location. It also names the weak point: data freshness, because imports need scheduled syncing and reliable record matching.

Both guides assume the rank data arrives from somewhere. The Airtable article does not say where, and Data Fetcher points to a paid SERP API. With Linkeddit, Claude pulls the weekly checks that Search already records and writes them, which removes the sync you would otherwise build.

Because the two do not move together. In the test run, cookieless analytics returned a rank of null, so the domain was outside the top 10, yet the AI Overview Google showed listed the domain as a source. The same held for gdpr compliant analytics. A table holding only the position would file both as a loss.

The reverse appeared too. privacy friendly analytics ranked 3 with no AI Overview shown at all, and google analytics alternative ranked 2 with an Overview that cited the domain. Four keywords gave three different combinations, which is the case for separate fields.

Add the top domain as a column. Whoever holds position 1 is who you are chasing, and a grouped view by top domain shows which competitor owns the most keywords.

Rank data is noisy, and people who watch it say so. An SEO at an agency wrote on Reddit about a page that kept falling out of the results: "Only for that page to drop out of the top 100 rankings for a period of time (sometimes a few days / sometimes a week or so)." That is why each weekly row should carry the check date and the raw result, so a drop can be read against what was actually fetched.

A null rank means the domain was not in the top 10 on the checked results page. It is a result. Store it as a select option, not blank, or your formulas will treat it as missing.

A first check has no change. The test run returned change_text of first check with a null change, because there was no earlier check. The change is the previous rank minus the latest, so a positive number means the keyword moved up, and it is null when either check was outside the top 10. Do not convert that null to zero.

Keep one record per keyword per date, as the time-series design above does, and ask Claude to skip a pair that already exists, so a rerun cannot duplicate a week. The history view returns the series per keyword over the last 8 weeks by default.

Only when Search Console is connected in Linkeddit. The aio_impact view splits keywords into those with and without an AI Overview and compares Search Console clicks before and after, with 28 day windows by default. On the test project it said Search Console was not connected, and Claude should pass that message on instead of estimating.

Until it is connected, the base still answers a useful question: for which keywords does Google show an Overview, and does it name you? Data Fetcher also describes storing Search Console clicks, impressions, CTR and position in Airtable, so you can bring the same numbers in yourself if you prefer.