Comparing a recognisable design with a stripped-back variant

Compare two ads for the same product to see whether a full-brand treatment or a stripped-back treatment performs better. Keep the other ad details consistent, split the audience at the same time and choose the primary outcome before launch.

Define the change and outcome

For version A, use the logo, brand palette, type system, imagery style, iconography and tone. For version B, keep the same logo, imagery, tone and copy, but remove the branded palette, type system and iconography.

Keep the product, offer, destination, headline, discount, imagery and essential copy identical. This pair tests the combined effect of the palette, type system and iconography; it does not isolate the effect of each one.

In Google Ads, use the Experiments page to set up the test. Randomly and simultaneously divide the same eligible audience between the two versions, and keep the format and placement the same.

Record the actual assets served, audience and test period. Set the test’s start and end before launch so both versions run over the same period.

Read the result cautiously

Choose one primary outcome before launch, such as clicks, completed purchases, downloads or sign-ups. Treat the other outcomes as context rather than changing the goal after seeing the results.

If completed purchases are the primary outcome, more clicks alone do not make a version the winner for that goal. Keep the conclusion tied to the tested audience, format and placement.

Compare versions that ran simultaneously over the same period; do not declare a winner from a tiny sample or from tests run in different seasons. If performance differs by placement, check whether either asset was cropped, since that means the comparison no longer isolates brand treatment.

Use the result to refine the treatment, then retest if the creative or channel changes substantially.

Key Performance Indicators for Ad Testing