How to Test an Ad Spokesperson Before You Sign
Run the same ad with three different spokespeople before you commit. A casting scorecard shows what each candidate helps people believe, remember and attribute to the brand.
Before you commit to a spokesperson, find out whether the person makes the product's promise easier to believe and easier to connect to your brand. A familiar face can get people talking about the ad while leaving them unsure who made it or what it offered.
Take a fictional commuter bag brand with a working demonstration of a laptop sleeve that opens with one hand. Its agency can cast a recognizable actor, a product designer with relevant expertise, or a regular commuter who has used the bag. The decision is which candidate earns a callback and a production quote. The casting test should tell the team why.
Give each candidate the same job
Ask each person to deliver the same factual script and demonstrate the same feature. Keep the running time, shot sequence, brand exposure, captions, sound level, and call to action as similar as possible. Compare rough cuts at the same stage of finish. A better-lit expert cut paired with a badly recorded customer cut tests two productions at once.
Use authorized footage. If the expert or customer would introduce personal experience or a credential, give each version a truthful, comparable introduction. If an introduction needs to change with the role, document it as part of the package you're testing. Michael Kamins and Kamal Gupta varied both spokesperson type and fit in a study of endorser congruence. The distinction is useful when you write the casting brief. Spokesperson and product congruence study
If you can afford a fourth cut, use a demonstration with a neutral voiceover. The baseline helps you see what each casting route adds to a product explanation that already works on its own.
Ask what stuck before naming the talent
Assign each eligible viewer to one cut at first exposure. Show it in the placement you intend to buy, including the same mute and caption conditions. Ask people to describe what they saw, name the advertiser, and explain what the bag lets a commuter do before you ask about the spokesperson or their familiarity with them.
Collect delayed message recall after the same interval in each group. A next-day check is one option. Track who came back by cut and retain the response base. If one group has substantially fewer returners, investigate that difference before interpreting recall.
Copy this casting scorecard
Measure | What to record for each cut |
|---|---|
Candidate and exposure | Cut ID, spokesperson, placement, eligible viewers assigned, and completed exposures. |
Credibility | Full distribution for “How believable was the demonstration?” and reasons given in the viewer's own words. |
Unaided brand attribution | Correct advertiser, competitor named, or unsure. Record the eligible answering base and the scoring rules. |
Delayed message recall | Count of returning viewers who accurately describe one-handed access, total answering, and follow-up completion rate. |
Casting economics | Fee, agreed usage window, reshoot availability, and any production constraint that would change the decision. |
Decision | Advance, revise and retest, or drop. Record the measure or production constraint that drove the call. |
Choose the candidate whose advantage is worth paying for
Agree on the minimum acceptable brand attribution and the improvement that would justify a higher fee before you see the results. If the actor generates recognition but viewers regularly name another advertiser, test a stronger brand cue before committing. If the expert improves credibility and buyers can still retell the benefit the next day, the expert has earned a production discussion. If all three look similar within the study's precision, use production feasibility and cost to decide which version earns the next test.
The test ends with a casting recommendation and one remaining question for the final cut. For help turning that decision into a study brief, see Sapien's creative testing services.
