Price in ads: how to actually test it

Few questions divide marketing teams as reliably as whether to put price in the ads. One side argues price scares off clicks the brand paid good money to attract. The other argues those clicks were never going to buy. Both sides usually argue from taste, because the test that would settle it is easy to get wrong. Having run this test properly on both Google and Meta, here is how to do it, and what tends to happen.

What price does mechanically

Price in an ad is a filter that runs before you pay. The person who sees the price and does not click has disqualified themselves for free. The person who clicks anyway arrives knowing what this costs, which changes everything downstream: they progress further, select products at a higher rate, and reach checkout without the price-shock cliff that kills so many funnels. Click-through rate usually falls, cost per click sometimes rises, and the argument for price rests entirely on what happens after the click. Which is why you cannot judge this test in the ads interface.

Designing a clean test

The rules are simple and routinely broken. Same creatives with and without price, or a before-and-after on the same ads, never a comparison between shiny new price ads and battle-worn old ones. Matched time windows, because weekends, paydays and promotions all move conversion on their own. And one change at a time: if price went into the ads the same week the landing page changed, you have two suspects and no verdict.

If you are weighing a test like this on your own account, this is exactly the kind of thing I help with.

Read the funnel, not just the CPA

Purchases are the slowest, sparsest signal you have, and a fair price test can look flat on CPA for weeks simply because the numbers are small. The leading indicators live in the funnel: at what rate do sessions from price ads view products, select a product, reach checkout? In one test I ran, sessions from price-led ads selected a product at roughly double the rate of sessions from otherwise comparable ads without price, and the gap opened exactly where the theory says it should: not at the landing, but at the moment of commercial commitment. When the funnel moves like that, the CPA follows as volume accumulates. If the funnel does not move, no amount of waiting will save the CPA.

The confounds that ruin verdicts

Three appear constantly. First, learning phases: newly launched or heavily edited ads get algorithmically penalised for days, so a new price ad loses unfairly against a settled control. Second, mature ads inherit returning visitors, sessions from people who clicked weeks ago and came back, which flatters their conversion rates in ways a new variant cannot match. Third, bulk edits: platforms record mass changes, and an unexplained account-wide edit landing mid-test quietly resets your baseline. Pull the change history before you read the results, not after.

When price loses

Price in ads is not a law of nature. It loses when your price is genuinely uncompetitive, in which case the ads are honestly reporting a commercial problem the marketing cannot fix. It loses for premium positioning where the price only makes sense after the value story. And it struggles where pricing is too complex to state fairly in a headline, when a fragment of price can mislead worse than none. In each case the funnel data will tell you: if price-exposed sessions progress worse rather than better, believe them.

The verdict most tests reach

For subscription and considered purchases with visible market pricing, price-in-ad usually wins, not by attracting more clicks but by spending money only on people who already accepted the number. But “usually” is not your account. Run the clean version of the test, read the funnel before the CPA, check the change history, and let the evidence make the call. It is a cheaper argument to settle than to keep having.