You rewrote your pricing page three weeks ago. Your manager asks whether it changed anything in ChatGPT. You open your tools and find a curve that went up a little, on a month when everything went up a little. Was it the page, or the market?
The Progress page answers that question, action by action.
What problem does Progress solve?
AI answers move all the time, with or without you. From one month to the next, a large part of the sites cited on a given prompt changes. So when your visibility goes up after a change, you cannot tell whether your change did it, and when it stays flat, you cannot tell whether the change failed or simply has not been picked up yet.
Most teams end up guessing. They keep doing what they believe works, and they stop doing what they cannot prove. Progress replaces the guess with a measurement.
What do you see on the Progress page?
Three things, from the top of the page down.
At the start and today. Three figures: your brand mentions across all your prompts, the number of pages of your site cited by at least one answer, and the AI health score of your site. Each one compares the 4 weeks before your first logged action with the last 4 weeks, or for the site, your first crawl with the latest one.
The Impact tab. One line per action that targets prompts: the page fixed or the recommendation done, the date, and how far into its measurement it is (D+12/28, for instance). Next to it, mentions and citations before and after, in points, and one badge per AI engine: moved, waiting, or not yet. Lines are grouped into three panels: It worked, Being measured, No visible effect.
The History tab. The log of everything done on the project, month by month, with a tag on what an AI agent logged through MCP and on what Mentionable detected by itself.
How does Mentionable measure the effect of an action?
On the prompts the action targets, Mentionable takes the share of AI answers that name your brand and the share that cite your page, over the 4 weeks before the action and the 4 weeks after. A project scanned only every two weeks or once a month gets a longer window, so that each side still holds enough scans.
Then it does what a careful analyst would do: it measures the same change on your other prompts, over the same dates, and subtracts it. If every prompt went up 6 points, an action that went up 6 points did nothing on its own. If its prompts went up 14 points while the others went up 4, the action is credited with 10.
Three rules keep the verdict honest:
- No figure before 3 scans on each side. Until then, the page says how many scans are left and when the next one runs.
- It worked as soon as the action's own effect on mentions or citations reaches 5 points.
- No visible effect only once the window is over. An action is never written off for lack of data.
Each engine also has its own badge, because they do not move at the same speed. An engine stays on "waiting" while its usual delay to pick up a change has not passed.
Which actions are measured?
Page fixes, and recommendations you mark done that target prompts: a source to get mentioned on, a local action, a mission from your Action plan. Each one carries the prompts it serves, so the effect is read where it should show.
Everything else still goes into the History tab and onto your visibility chart in the dashboard: a published page, a press article, a change made outside Mentionable.
Where do the entries in the log come from?
From four places, so that the log stays complete without becoming a chore.
- You. "Add an optimization": what you did, when, and optional details. Backdating is allowed.
- The Action plan. Marking an action done writes it to the log, on the day you did it.
- Your AI agent. Through the Mentionable MCP server, an agent like Claude or Cursor can log what it just changed, and read the log back by date range for a monthly report. Its entries carry the "via MCP" tag.
- Mentionable itself. A plan page found online at the next crawl of your site is checked off and logged. When you run a page fix again after changing the page, Mentionable checks which points you applied and logs them. Those entries carry the "Detected at recrawl" tag.
Every entry can be edited or deleted.
Who is it for?
The marketing lead who has to say, at the end of the month, what the AI visibility budget produced. The team that wants to stop doing what does not move anything. And agencies, who need to show a client which recommendations paid off and which ones did not, with the method on the page instead of a slide full of curves.
Which plans include Progress?
Every Mentionable offer, for brands and agencies, and the 7-day free trial. The measurement runs at the end of each scan: there is nothing to configure, and nothing to start.