A pre-post analysis compares performance in a period before a change with a period after it, to judge whether the change made a difference. In Google Ads it is the most common way people evaluate a new bid strategy, a landing page, a match type switch or a budget increase.
How pre-post analysis works
You pick the date the change went live, choose a window of equal length on each side, and compare the metrics that matter: conversions, cost per conversion, conversion value, click-through rate. Google Ads makes this easy with its compare date ranges option, and the change history shows exactly when each edit was made.
A fair comparison needs a few rules:
- Equal and whole weeks. Compare 28 days with 28 days, starting on the same weekday, because most businesses behave differently at weekends.
- Skip the settling period. Bid strategies and new campaigns go through a learning phase; leave the first week or two out of the “after” window.
- Allow for conversion lag. If people take a week to enquire after clicking, the most recent days under-report. Wait until the lag has passed before reading the “after” period.
- Change one thing. If you changed bids, ads and the landing page in the same week, the analysis cannot tell you which one mattered.
Why it matters
Most small accounts cannot run formal tests on every decision, so before-and-after comparisons are often the only evidence available. Used carefully, they catch obvious failures quickly. Used carelessly, they credit or blame a change for something that would have happened anyway.
The biggest trap in the UK is the calendar. Seasonality moves demand constantly: bank holidays, Easter on a different date each year, school holidays, the January sales, Black Friday, payday at the end of the month and, for some sectors, the end of the tax year. A change made in late November will look brilliant for a retailer and poor for a wedding venue, regardless of what the change did.
Tracking changes cause the same problem. If a new conversion action was added, consent settings changed or a tag broke during either window, the comparison measures the tracking rather than the campaign. Check the conversion setup was identical across both periods before trusting the numbers.
Common mistakes
- Comparing windows of different lengths, or a period containing a bank holiday with one that does not.
- Reading results during the learning period and reversing the change too early.
- Ignoring wider market shifts. A competitor leaving the auction, or a news story, can move results more than your change did.
- Making several changes at once and attributing the result to the one you hoped would work.
- Treating a small difference on a handful of conversions as proof. Ten conversions against twelve is noise.
How to act on it
Strengthen the comparison with a control. If you changed some campaigns and not others, compare how both groups moved over the same dates; if the untouched campaigns rose by a similar amount, the market moved, not your change. Check the same weeks last year for seasonal patterns, and look at Auction Insights for changes in competition.
For decisions that matter, use a proper test instead. A campaign experiment splits traffic between the old and new versions at the same time, and a geo experiment compares matched regions, which gets closer to true incrementality. Keep a dated log of every change so later analysis has something to work from. Building that discipline into the account is part of my PPC management service.