A campaign's most recent days always appear to perform worse than earlier ones. This is usually an artifact of timing rather than a real decline.
The buying event and the outcome are separated in time
An impression happens at a fixed moment, while the conversion it contributes to may occur hours or weeks later depending on the decision involved.
Reporting attributes that conversion back to the original impression date, so the number recorded against any given day keeps rising after the day has ended.
Yesterday's figure is therefore incomplete by construction, and the more recent the day, the more incomplete it is.
The lag is a property of the purchase
Low-consideration purchases largely convert within a day, so their reporting settles quickly and recent data is close to final.
Considered purchases involving research, comparison or approval from someone else can take weeks, and a substantial share of conversions arrive long after the click.
Business-to-business advertising sits at the far end of this range, where a lead created in one quarter may close in another entirely.
Reacting to incomplete data creates a cycle
An advertiser who cuts budget on a campaign that looks weak in its most recent days is often cutting one that will look fine once the data matures.
The cut then reduces impressions, which reduces the conversions that would have arrived later, appearing to confirm the original judgment.
This pattern is common enough that many teams adopt a rule against acting on any window shorter than the campaign's typical lag.
Measuring the lag is straightforward
Comparing a completed period's conversion total as first reported against the same period's total weeks later shows how much arrives late and over what interval.
That distribution gives a usable maturity point: the number of days after which a period's figures stop changing meaningfully.
Reporting can then exclude days that have not reached maturity, or apply a correction factor derived from historical patterns rather than presenting raw partial numbers.
Attribution windows interact with the lag
A conversion arriving after the platform's attribution window closes is never credited at all, so a short window systematically undercounts slow-converting products.
Where the lag exceeds the window, the campaign appears permanently weaker than it is, and no amount of waiting will correct the figure.
Choosing a window that reflects the real purchase cycle is therefore a measurement decision with direct budget consequences, not a default to be left alone.