Skip to main content

​
Impact Tracking

When you make changes to improve your AI visibility, you need to know whether those changes actually worked. Impact Tracking gives you a clear before-and-after picture so you can see exactly what moved the needle.

​
How It Works

Impact Tracking follows a simple three-step process:

​
1. Baseline Capture

When you mark a recommendation as done, we capture your current performance across the tracked metrics at that moment. This snapshot becomes your “before” measurement. Nothing is captured until you mark the action done.

​
2. Continuous Monitoring

From then on we re-measure the same metrics every day, indefinitely, for as long as the action stays open. There is no fixed measurement window: AI engines can start citing a page weeks or months after you change it, and we keep counting.

​
3. Before and After

At any point you can see what changed and by how much, comparing today’s numbers against the baseline. When you are finished with an action, archive it and we stop measuring it.

​
What Gets Measured

Impact Tracking monitors four metrics:
How many AI answers cite the specific page you changed. This is the sharpest signal that a content change landed, because it is scoped to the exact URL rather than your whole site.
Cumulative visits to that page from AI surfaces since you took the action. Requires Crawler Analytics to be active.
The raw number of AI answers that mention your brand. Useful for spotting trends even when Share of Voice stays flat.
The percentage of tracked AI responses that mention your brand. This is the single most important indicator of your overall AI visibility.
Reddit experiments track a slightly different set, scoped to the post you are watching: whether your post has been cited, by which engines, and for which prompts.

​
How Long to Wait

There is no fixed measurement window, but the shape of the change is usually the same:
  • Early on: AI platforms may not have re-crawled your updated content yet. The baseline is what matters here.
  • After a week or two: changes typically start appearing as AI models incorporate new information.
  • Beyond that: trends stabilize and you can see the sustained impact of your changes.
You can start tracking a new action at any time. If you make a second round of changes to the same page, record that as its own action so the two do not blur together.

​
Trusting the Results

Not all changes in metrics are meaningful. Impact Tracking helps you separate signal from noise:
  • Trend direction: A metric that moves consistently in one direction over multiple data points is more reliable than a single spike.
  • Magnitude: Small fluctuations are common and may just reflect normal variation in how AI generates responses.
  • Consistency across prompts: If your SOV improves on most tracked prompts rather than just one or two, the improvement is likely real.
These are judgement calls for you to make. The platform reports the before and after numbers; it does not run a statistical significance test on them.
AI responses are not deterministic. The same query can produce slightly different answers each time. Impact Tracking accounts for this by averaging across multiple observations rather than relying on single data points.

​
Getting Started

  1. Open Actions in your workspace sidebar.
  2. Make the change to your content, or post the Reddit thread.
  3. Mark the recommendation as done. That is what captures the baseline and starts measurement, a tracking run on its own does not start anything.
  4. Come back to Actions after a week or so for early signals, and keep checking: the numbers carry on updating until you archive the action.