Quick answer: Pre-test ad creative when the media spend behind it is large enough that being wrong is expensive, which for most Southeast Asian campaigns means anything above a modest digital burst. Test on branded recall, message comprehension, distinctiveness and brand linkage rather than on whether people say they like it. Use a monadic design so each respondent sees one execution without comparison, keep the stimulus close to the finished asset, and read every market separately. A single market read can field in as little as 24 hours. We build Vase.ai, so treat this as a method guide, not a pitch.
Is creative pre-testing worth doing at all?
It depends entirely on what you would do with the answer. Pre-testing earns its cost when there is a real decision attached: choosing between two finished routes, deciding whether an execution is strong enough to justify the media plan behind it, or catching a comprehension or cultural problem before it goes live across six markets. It is a waste of money when the creative is already booked, when the result will be overruled by whoever is most senior in the room, or when the test is being run to protect someone rather than to inform a choice. Be honest about which situation you are in before you commission anything. A pre-test that nobody will act on is an expensive form of paperwork, and the industry runs a great many of them.
What should you actually measure?
Likeability is the metric everyone reaches for and the least useful one, because ads that people enjoy and ads that shift behaviour overlap only partially. Four things matter more. Branded recall, meaning whether people can retrieve the brand and not just the ad, which is where a surprising number of well-liked executions fail. Message comprehension, tested by asking in an open end what the ad was saying before you show any prompted options.
Distinctiveness, meaning whether the execution is attributed to you rather than to the category leader, which is a real and common failure mode in Southeast Asia where category codes are strong. And relevance to the specific occasion or need the campaign is meant to own. Purchase intent is worth capturing but should be read as directional only, because stated intent overstates behaviour fairly consistently.
Should you use a monadic or a comparative design?
Monadic in almost all cases. In a monadic design each respondent sees one execution and evaluates it in isolation, which mirrors how people encounter advertising in the real world and produces scores you can compare against norms. In a comparative design respondents see several executions side by side, which turns the exercise into a preference task and inflates the differences between them. Comparative designs have a narrow legitimate use: very early-stage screening when you want to eliminate the obviously weaker of six rough routes and precision does not matter.
For anything close to final, run monadic cells with a separate sample per execution, which does mean your total sample multiplies by the number of executions. That cost is the price of a result you can trust.
How good does the stimulus need to be?
Closer to finished than is comfortable. Respondents cannot reliably imagine the finished film from an animatic, and the gap is worse for executions that depend on performance, music or timing, which is a great deal of Southeast Asian brand advertising. Animatics are acceptable for testing whether a message lands. They are unreliable for testing whether an execution is charming, funny or distinctive, which is often exactly what you want to know.
If you must test rough material, say so in the debrief and read the emotional metrics with visible caution. And test in the format people will see: a film destined for a six second bumper on mobile should not be tested as a thirty second cut on a desktop screen, because you will be measuring a different ad.
What does a sensible test design look like?
| Element | Recommended | Common mistake |
|---|---|---|
| Design | Monadic, one execution per respondent | Side-by-side comparison inflating differences |
| Base per cell | 150 to 200 completes per execution per market | Splitting one base across three executions |
| Audience | The actual media target, screened | General population for a targeted campaign |
| Stimulus | Near-final in the intended format | Animatic judged on emotional metrics |
| Key metrics | Branded recall, comprehension, distinctiveness | Likeability as the headline |
| Markets | Read and reported separately | One pooled Southeast Asia score |
Why does creative travel so badly across Southeast Asia?
Because humour, family dynamics, religious sensitivity, language register and category codes differ across the region far more than a regional media plan implies. An execution that reads as warm in one market can read as intrusive in another. Wordplay rarely survives translation, and a subtitled version is a different ad from a dubbed one. Festive periods carry heavy category expectations and an execution that ignores them can feel tone deaf.
The practical implication is that a regional pre-test with one pooled sample is close to useless: it will produce an average score that describes no market and hides the one country where the ad has a problem. Test per market, and be prepared for the answer that one execution works in three markets and needs a different route in a fourth.
How do you read the results without over-reading them?
Three habits help. Look at the gaps rather than the absolute scores, because absolute scores mean little without norms and most teams do not have norms in the relevant market. Read the open ends before the charts, since the verbatim answer to what the ad was saying will tell you more about a comprehension failure than any scale ever will. And check base sizes on every subgroup before quoting it, because a monadic design splits your sample and the cells get small quickly.
Then apply the rule that saves the most money: a pre-test is good at identifying a broken execution and only fair at ranking two good ones. If two routes score within a few points, the test is telling you they are equivalent, not that one is better, and the decision should be made on strategic fit instead.
How fast can this run, and what does it cost?
Faster than the creative timeline usually assumes, which is the main reason to build it into the process rather than treat it as a luxury. On a platform with an owned local panel, a single-market monadic test on two executions can field in as little as 24 hours, with responses building on a real-time dashboard as they arrive. Studies on Vase.ai start from around RM5,000 (about USD 1,000), with survey building and scripting from around MYR 1,500 and a full research report from around MYR 4,700 if you want a researcher to write the read. The multiplier to plan for is cells: two executions across three markets is six cells, and each needs its own base. Budget for that upfront rather than discovering it when the quote arrives.
When is Vase.ai the right fit, and when not?
We would back Vase.ai for creative pre-testing where you need fast per-market Southeast Asian reads on a verified panel and want the result back inside a creative approval window. Our panel covers 3.6 million verified Southeast Asian consumers, responses are AI-validated during fielding, results build on a real-time dashboard, studies run DIY or with our research experts, and more than 250 companies work with us. Where we are honestly not the best fit: if you need to benchmark against a large validated copy-testing norms database, Kantar and Ipsos own that ground and their normative context has genuine value, particularly if your organisation already reports against it.
NielsenIQ is the answer where the question is sales response rather than creative diagnostics. If you need biometric, eye-tracking or facial coding measurement, use a specialist that runs those instruments. If the campaign is global and needs one consistent methodology across every region, a global agency will handle that better than we will. Milieu Insight is a credible regional alternative and worth quoting alongside us.
Frequently asked questions
What is ad creative pre-testing?
Pre-testing is evaluating an advertising execution with a sample of the target audience before the media spend is committed. It typically measures branded recall, message comprehension, distinctiveness and relevance, so a team can catch a broken execution or choose between routes while changes are still cheap.
Should ads be tested monadically or side by side?
Monadically in almost all cases, meaning each respondent sees only one execution. Side-by-side comparison turns the exercise into a preference task and exaggerates the differences between executions. Comparative designs are only reasonable for very early screening of rough routes.
How many respondents do you need to test an ad?
Plan for 150 to 200 completes per execution per market, since a monadic design needs a separate sample for each cell. Two executions across three markets means six cells and six bases, which is the cost multiplier most teams underestimate when they budget a pre-test.
