Donor count
How many peer sites went into your control. Below 30 we show an honest before/after, not a causal claim.
The synthetic control is only as trustworthy as the pool of peers it is built from. The donor count is simply how many eligible, consented peer sites were similar enough to yours to be candidates for that control.
We enforce a hard floor: 30 donors. With fewer than 30, the control can be shaped to fit almost any story, which would let us fabricate a causal claim. So below 30 we refuse the causal label entirely and show a plain before/after instead, clearly marked as correlational.
A before/after tells you what changed around the time you acted. It does not net out seasonality or algorithm updates, so it cannot rule out coincidence - and we say exactly that on the label rather than dressing it up as proof.
Worked example
Your niche has only 18 consented peer sites resembling yours. That is below 30, so instead of a causal lift card you see "Before/after only - not enough matched peers for a causal estimate", linking here. Grow the consented cohort past 30 and the causal estimate unlocks.
Honest limits
The floor guards against false confidence, not against every error - even at 30+ donors the estimate still carries a confidence interval and stated assumptions. A large donor count improves reliability but never turns an estimate into certainty.