Skip to main content
The results page shows a small projected impact line next to a winning verdict. Something like:
Projected impact: +$4,200/month at current traffic.
Useful for showing a stakeholder “why we should ship this”. Easy to over-read if you do not know what it is assuming.

​
What it is doing

Simple math:
  1. Take the observed lift on the primary goal (say +3.2 % conversion rate).
  2. Take current traffic on the tested pages (visitors per month, from the ledger).
  3. Multiply. For a revenue goal, use the observed revenue per converter; for a click goal, use the value you configured on the goal.
That is the whole model. No forecasting, no seasonality adjustment, no ML. The math is transparent so the number is easy to trust or to question.

​
What it assumes

  • The lift holds. The observed lift is what the test measured. It can shrink in production (regression to the mean), especially on early or under-powered tests.
  • Traffic holds. Current-month traffic is a poor proxy for Black Friday.
  • The value holds. For a revenue goal, the goal’s average value is used. Big shifts in AOV break the projection.

​
Where it can mislead

  • Very small samples. A hundred-visitor test showing +3.2 % lift probably does not have a real 3.2 % effect. The projection uses the point estimate, not the lower bound of the CI. Read the CI alongside.
  • Guardrail-breaking winners. The projection assumes only the primary goal moves. If the variant tanks a guardrail, the honest impact is smaller than the projection shows.
  • Cannibalized traffic. If the test’s page competes with another page for the same conversion, a win in one place can be a loss somewhere else the projection does not see.

​
How to use it

  • As a magnitude. Is this a 200/monthtestora200/month test or a 20,000/month test? The projection is honest to an order of magnitude.
  • As a stakeholder-friendly summary. “This is worth shipping” reads better with a currency figure.
  • Not as a business plan. For a real forecast, take the lower CI bound, discount for regression, and add your own market assumptions.

​
Value for non-revenue goals

For a click or custom event, the projection reads the goal’s default value if one is set. Configure it in the goal builder’s “revenue value per fire” field. If no value is set, the projection line is omitted and the results page shows the raw lift only.