Webflow’s October 15, 2026 MCP update lets connected AI tools read data from Analyze, Optimize, and AEO analytics. The release adds access to page traffic, goals, experiment results, and information about AI discovery.
Google Gemini also joins Claude and ChatGPT as a connector. Webflow says Gemini support for Enterprise plans is still in development. Check availability against the account you intend to use.
What the connection makes possible
MCP, short for Model Context Protocol, is a way for compatible AI tools to work with connected services. In this case, the new data access gives an assistant more information about what is happening on a Webflow site.
A marketing team could use it to prepare a page-performance summary or review responses to tracked buyer questions. The value depends on whether the resulting answer identifies the right data and helps someone decide what to investigate.
Our recommendation is to start with one reporting task. Choose a question you can also check in the source report so you can assess the output.
Give the reporting task a clear definition
A prompt such as “How is the website doing?” leaves too many decisions unstated. Define the date range, comparison period, pages, and metric you want reviewed.
For example, ask for a comparison of selected landing pages over two specified periods, including traffic and the relevant recorded goal. Request the underlying values and a separate list of possible explanations. This is an example reporting brief, not a result from a tested connection.
Check that the tool preserves the metric’s definition. A goal completion is not automatically a qualified sales lead. A mention in an AI response is different from a citation linking to your site.
Check the access and the source
Webflow documents AEO analytics access through MCP as read-only. Its AEO analytics product requires a Team or Enterprise Platform plan with Analyze and the Workspace AI setting enabled.
The wider MCP connection can also support site changes. Webflow’s MCP overview says it respects the connected user’s existing permissions. Review the role and tools available to the account before treating an analytics workflow as a purely read-only connection.
Ask the assistant to identify the data it used, then compare a sample with the original report. Keep unavailable or missing values visible. A confident explanation is not a substitute for the underlying numbers.
Keep decisions with the team
If the assistant identifies a weak page, inspect it before changing it. The issue might be traffic quality, an offer that changed, missing tracking, or a small sample. Agree what evidence would justify an edit or experiment.
Document the reporting prompt and metric definitions once the team is satisfied with the result. That makes future summaries easier to review and gives the next person a clear starting point.
For a first pilot, use a recurring report that currently takes someone time to assemble. Measure the time saved and the corrections needed. That will tell you more about the workflow’s usefulness than a polished sample answer.
Related reading
Written by
