How to plan an ad creative test for a small business

By Noe Guerrero III

Before paying for another round of ad designs, write down the question the next version is supposed to answer. “Make this one perform better” isn't enough. Better at getting noticed, producing quote requests, or bringing in the kind of work you actually want?

A useful ad creative test starts with one question, two clearly defined versions, a comparison you can explain, and a rule for reviewing the result. You can put that plan on one page. The point is to make the next decision easier, even if the test doesn't produce a winner.

Give the creative one job

Imagine a print shop in McAllen advertising custom team shirts. This is a hypothetical example, not a client result. Its current ad shows a close-up of printed fabric. The owner wonders whether showing a finished team order would help prospective customers understand the service.

That's a question a designer can work with: does the finished-order image produce more qualified quote requests than the detail image under the chosen test conditions?

Keep the headline, offer and call to action the same. If the second ad also adds free delivery and a discount, you won't know which change deserves the credit. Google Ads' experiment planning guidance recommends isolating a variable and choosing success measures before the test.

The test doesn't need to settle your entire brand direction. It needs to answer something small enough to use. “Which message attracts commercial work?” is still broad. “Does this photo help people recognize our team-order service?” gives the next creative a clearer assignment.

Put the comparison conditions in writing

Two ads existing at the same time doesn't automatically mean you have a controlled experiment. Before launch, ask whoever manages the account how people or traffic will be assigned to each version. Confirm what experiment options the particular platform and campaign support. Don't assume the platform will split everything evenly just because you uploaded two images.

Write down the audience settings, destination page, offer, placements and schedule you intend to hold steady. Note any differences the setup requires. If the platform changes where or how often each version appears, that belongs in the interpretation of the result.

Running the old ad this week and the new one next week can still provide a business observation. But treat it cautiously: those weeks could differ in demand, promotions or available inventory. Don't present that comparison as proof that the image alone caused a change.

For the print shop, the destination should describe the same team-shirt service in both cases. Sending one version to a focused quote page and the other to the homepage would turn this into a different question.

Copy this short test brief

Give this to the business owner, designer and person managing the ads before anyone builds another version. The filled-in example below is a proposed plan, not a promise of results.

• Question: Does a finished team-order photo help generate qualified team-shirt inquiries compared with the current fabric-detail photo?

• Version A: Current fabric-detail image, with the existing headline, offer and quote button.

• Version B: Finished team-order image, with that same headline, offer and quote button. Both images use approved assets without private customer details.

• Held steady: Intended audience, destination page, offer and campaign settings, subject to the documented experiment setup. Record any unavoidable differences.

• Primary outcome: Quote requests that meet the shop's written definition of a team-order inquiry. Decide that definition before launch.

• Checks and boundaries: Confirm that both ads and the form work. Agree on a spending limit and who can pause the test for a broken page, incorrect offer or other operational problem.

• Review plan: Set a review appointment, identify the reporting window and name the person responsible for checking results. Decide what would justify adopting a version, continuing the test or calling it inconclusive.

Add the campaign reference and links to both creative files. A screenshot of each final ad is useful too. File names such as “new-final-2” won't explain what ran when someone revisits the test next month.

Choose the decision before looking at the dashboard

If the business needs quote requests, define a quote request that matters to that business. Don't wait until results arrive to decide that a different, more flattering number was the real goal.

For the example shop, the owner might require an inquiry to identify the type of order and include usable contact information. That's a proposed definition, not a universal standard. Keep it short enough that staff can apply it consistently. Our guide to tracking calls and ad lead sources covers the measurement groundwork; this brief assumes you can connect the chosen outcome to the test.

A click can help explain what happened without settling the business decision. If people respond to the image but the inquiry form doesn't work, fix the form and document the interruption. Don't quietly combine the broken period with a clean period and call the total a creative verdict.

Also leave room for delayed outcomes. Google Ads explains that conversion lag is the time between an ad click and a conversion; recent cost-per-conversion figures can look worse before later conversions arrive. Choose the review window around the actual action you're measuring. A same-day dashboard check may be too early for a customer who needs to confirm an order with a team manager.

Write a result you can defend

At review time, return to the original question and the conditions in the brief. Check whether the intended comparison actually ran. Record interruptions, changes and missing information before interpreting the numbers.

Google Ads' experiment results documentation includes uncertainty information and recognizes results with no clear winner. A small apparent difference isn't automatically a dependable improvement.

Keep your written conclusion bounded: what ran, what the selected measure showed, what remains uncertain, and what action you chose. “We don't have enough evidence to replace the current image” is a usable decision. It can save the designer from being handed an unsupported rule such as “finished products always beat detail photos.”

If the comparison was disrupted, plan a cleaner attempt. If it ran properly but remained inconclusive, decide whether the question is valuable enough to revisit. More spending isn't an obligation, and another test should have a reason beyond wanting a more exciting chart.

A good creative brief gives the designer something specific to make. A good test brief gives the business something specific to learn. Write both before asking for the next ad.