There's no hard and fast answer to this question. A high traffic page seeing a significant change may decide if a test is a winner or loser after just a few days.
Other sites might need a week or more to collect sufficient data to make an accurate decision.
The degree of change you are seeing impacts how significant a result is. For example, a 20% improvement from 5 clicks to 6 clicks over a test period is inconclusive. But a 100% increase from 20 to 40 clicks is more likely to be valid.
You should also be aware of any significant dates around your tests. Holidays, weekends and sudden media coverage can all disrupt your data. It's good to compare your test performance against your wider site's traffic to ensure there are no obvious discrepancies.
