Pre-testing creative: killing bad ads before they spend.

Picture the room. It is a Wednesday in March. The agency has just played the new fifteen-second cut for the brand team. Everyone in the room either made the ad, paid for the ad, briefed the ad, or stands to gain professionally from the ad succeeding. The room is wired against any honest assessment of the work. Heads nod. Notes are taken. Minor edits are agreed. The ad gets approved, scheduled, and launched. A month later, the sales lift is roughly zero, and nobody can quite explain what went wrong.

This story repeats itself, in slightly different forms, across most marketing departments on earth. It is not malice. It is structure. The internal economics of advertising approval make objective evaluation almost impossible. Pre-testing exists because someone has to be the honest broker, and it cannot be anyone whose career depends on the project moving forward.

What pre-testing is actually for

Pre-testing has two real jobs. The first is to kill creative that has structural problems before money is spent putting it into market. The second is to identify, in creative that does work, the moments and elements that are doing the heavy lifting, so that the final edit can be optimised before launch.

Pre-testing is not, despite what some sales pitches imply, a way to predict in-market sales with mathematical precision. The market is too noisy and too contextual for any pre-test to produce that. What pre-testing can do reliably is identify creative that is clearly failing on the fundamentals of attention, emotion, and brand linkage, and creative that is clearly succeeding, and the range of creative in between that needs editing.

Used well, it reduces the rate at which catastrophically bad work reaches air, and improves the rate at which already-strong work reaches its best version. That is enough to pay back the investment many times over for any brand spending meaningful media money.

What focus groups cannot do

Most brands believe they pre-test, because they show the creative to a focus group before launch. Focus groups are not pre-testing. They are a particular kind of theatre with refreshments, and they share the same fundamental problem as the original brand-team review meeting: the participants are constructing rational explanations for emotional responses they barely have access to, and the dynamics of a group inevitably amplify the loudest opinion.

Worse, focus groups tend to surface the kinds of objections that articulate, talkative people produce when asked to evaluate creative. These are not necessarily the objections that drive real-world response. The viewer who would scroll past your ad in a feed without registering it is not in your focus group. The viewer who would respond emotionally without being able to articulate why is in your focus group, but you are listening to the wrong half of what she is doing.

This is not an argument for never talking to consumers. It is an argument for understanding what kind of information a focus group can deliver. Focus groups are useful for understanding category context, for surfacing language people use, for hearing how brands are talked about in social settings. They are bad at evaluating creative for in-market performance. Anyone who has compared focus-group winners against subsequent market results has seen this. The correlation is loose at best, and inverse in too many cases for comfort.

What measurement can do

A serious pre-test measures three things, and triangulates them. Attention, because if no one is looking, nothing else matters. Emotion, because attention without emotional response is forgotten within minutes. Brand linkage, because attention plus emotion attached to no brand is a successful ad for the category, not for your business.

The instruments vary depending on stakes and budget. For a high-stakes piece of creative, full lab study with eye tracking, facial coding, and EEG produces the most complete picture. For lower-stakes work, online webcam-based eye tracking and facial coding can deliver most of the same answers at lower cost and faster turnaround. Implicit testing can supplement either approach by measuring whether the creative is shifting category associations in the desired direction.

What the measurement produces is not a single score. It is a trace, second by second, of how the audience engaged with the work. Where did attention drop? Where did emotion spike? Was the brand on screen when attention and emotion were both present? Was the ending strong, or did engagement collapse before the brand reveal landed?

The cheapest piece of advertising research you will ever buy is the one that stops you from spending media money on creative that was never going to work.

The shape of a useful diagnostic

A pre-test report that just says "this ad scored 67" is not very useful. A pre-test report that says "this ad earns attention well in the first five seconds, loses 40 percent of viewers between seconds 8 and 12, recovers slightly at second 18 when the music hook kicks in, and fails to land the brand asset before final attention collapse at second 23" is enormously useful. The first kind tells you whether to run the ad. The second kind tells you how to fix it.

The diagnostic should identify specific moments that are doing damage and specific moments that are doing work. Moments doing damage usually fall into categories: a confusing transition, a slow first second, a misplaced brand cue, an ending that does not pay off the setup. Moments doing work are usually concentrated in unexpected craft: a particular shot, a sudden change of pace, a piece of casting that lands.

The right output of a pre-test is a brief for an editor. Trim seconds 8 to 12. Move the brand reveal forward by two seconds. Pull the music in earlier. Hold the final shot longer. None of these recommendations require a survey scale. They come from looking at the engagement curve and matching it to the footage.

When to kill

Most creative is fixable. Some creative is not, and recognising the second kind is part of the job. The patterns that suggest a fundamental problem, not an editing problem, are these: attention that never lifts above baseline at any point in the ad, emotional response that is consistently flat or negative, brand linkage that is below the category average and cannot be improved by reordering scenes.

When the diagnostic shows these patterns, the responsible recommendation is to go back to the drawing board, not to launch. The cost of doing so is real but bounded. The cost of launching the wrong work, in lost media money and in negative brand impact, is larger and harder to bound. Pre-testing pays back most dramatically when it stops one of these launches.

The hardest part of pre-testing is not the measurement. It is the conversation that follows a "do not launch" finding. The agency is exhausted. The brand team is committed. The launch date is on a calendar with marketing manager bonuses attached. The honest broker function of the pre-test partner matters more in that conversation than any of the data they bring to it. A pre-test partner who softens findings to keep clients comfortable is delivering negative value, regardless of the price.

The Caribbean discount

One particular finding from our work in the region is worth flagging. Creative produced or adapted for the Caribbean often performs differently from the same creative tested in source markets. Material that scored well in a London or Toronto pre-test sometimes lands flat in Kingston or Bridgetown, and vice versa. Audiences read tone, casting, and pace through their own cultural filters, and a pre-test in the source market does not predict performance in the destination market with any reliability.

The implication is straightforward: if you are running adapted global work in the Caribbean, pre-test in the Caribbean. The cost is small compared with the media budget at stake. The variance in findings between markets is large enough that the source-market score is a weak guide. Local audiences read the work, and local audiences are the ones whose response matters.

Practical guidance for brand teams

Three habits separate brands that pre-test well from brands that go through the motions.

First, build pre-testing into the workflow before the creative is fully locked. The most expensive pre-test is the one that happens the week before launch, when there is no time to act on the findings. The most valuable is the one that happens at the rough-cut stage, when changes are still cheap and the editor's notes can still meaningfully reshape the work.

Second, treat the pre-test as a brief for editing, not a verdict. The instinct to read a single score and either celebrate or panic is hard to suppress. The discipline to look at the engagement curve, identify the specific moments doing damage, and direct an edit takes practice. Brand teams that develop this discipline get much more value from the same studies.

Third, resist the temptation to test for reassurance. A pre-test designed to confirm what the team already believes will dutifully confirm it, especially if the measurement is shallow and the sample is small. A pre-test designed to find problems will find them, which is the whole point.

The honest economics

For any brand spending real money on media, the economics of pre-testing are settled. A pre-test that catches one out of ten launches as fundamentally broken, and improves three out of ten through edit recommendations, pays for itself many times over inside a single year. Brands that do not pre-test are not saving money. They are buying smaller insurance against larger losses they have not yet absorbed, and the losses arrive on schedule, quarter after quarter, attributed to "market conditions" or "category softness."

The shift to pre-testing is not a budget shift. It is a discipline shift. The brands that adopt it consistently see their hit rate on campaigns rise, their media waste fall, and their internal conversations about creative get sharper. The brands that do not, keep running the same Wednesday-in-March meeting, approving work that nobody can quite say is wrong, and paying media to deliver it to audiences whose brains were always going to scroll past.

The work of pre-testing is not glamorous. It rarely produces a slide that anybody wants to present at the awards show. What it produces is a steadily smaller pile of failed campaigns, and a steadily larger one of work that earns its keep. Over a few years, the difference becomes material. Over a decade, it becomes a different brand.