How pre-post analysis works
Pre-post analysis compares a period before a change with the period after it. Spend went up in one month and sales went up in the next, so the extra spend worked. It needs no special setup, it is the most natural way to look at a result, and it is the method most marketing reports quietly rely on.
The structure has one thing going for it and one thing badly against it. In its favour, the comparison is against your own business rather than against an assumed benchmark. Against it, there is no control: nothing in the design separates the effect of your change from everything else that moved at the same time.
Why the weakness matters
Almost every business has a season. Demand for trekking, education consultancy, remittance and retail all move with the calendar, so a comparison across two different months is partly a comparison of two different demand levels. Add competitor activity, a price change, a public holiday, a news story or a platform update, and the later period is not the earlier one with a single variable altered — it is a different set of conditions.
The consequence is expensive rather than academic. Budgets get scaled on effects that were really seasonal, and channels get cut for a dip they did not cause. A weak method used confidently is worse than no measurement at all, because it produces decisions.
Where pre-post analysis goes wrong
Two failures show up repeatedly. The first is choosing the comparison period after the fact — sliding the start date around until the story improves. If the window is not fixed before the change goes live, the analysis is a search for a favourable result rather than a test of anything.
The second is stacking changes. Bids, creative, landing page and budget all move in the same week, then the following month is credited to whichever one the team likes best. A before-and-after read can only ever tell you that the whole picture changed. It cannot say which part of it did the work.
How to use it responsibly
Use it where a real experiment is impossible, and be plain about the limits when you report it. Compare like periods rather than adjacent ones, so the same season sits on both sides. Fix the window before the change goes live. Change one thing at a time. Look at parts of the business your change did not touch over the identical period as a rough control — organic traffic, a region you left alone, a product line you did not promote.
When a decision is big enough to matter, replace the method rather than dressing it up. A geo experiment or a holdout test costs more effort and returns an answer you can defend in front of whoever is paying. Pre-post is a reasonable first look and a poor basis for spending more money.