How do you balance creativity and performance without sanding every idea down to something safe? Stop treating them as rivals and move the evidence earlier. When creative is pre-tested before the media money is committed, data stops being a verdict on brave work and becomes a tool for making it better.
The argument has fresh fuel. In an interview with Marketing Week, Sir John Hegarty called modern marketing "repetitive, cliché, boring" and blamed brands and agencies staffed by "left brain thinkers" who understand spreadsheets but not ideas. His evidence: "something like 95% of new product launches fail", a rate he says has not changed in 20 years despite all the extra data. He is right about the symptom. The diagnosis needs one more step. The problem is not data. It is when the data arrives, and what it is used for.
Why do creativity and performance feel like opposites?
Because of the order in which things happen. A familiar scene: the creative team presents three routes. The client's performance lead asks the reasonable question: "How do we know this will work?" Nobody knows, so the safest route wins. Or the work goes live and gets tested with real money.
If the bold route dips in week one, it is paused. The team draws the obvious lesson and plays it safer next time. Repeat that for a few years and you get the industry Hegarty describes.
In that sequence, performance data only ever turns up after the decision that mattered. It works as a post-mortem, and nobody does their best work while waiting for one.
Why it keeps happening: feedback that controls versus feedback that informs
Self-Determination Theory, developed by the psychologists Edward Deci and Richard Ryan, explains a lot here. People do their best work when they feel three things: autonomy (some control over the work), competence (that they are getting better at it) and relatedness (a connection to the people they work with and for).
The theory makes a useful distinction. The same number can land as control or as information. A dashboard used to judge work after the money is spent feels like control, and people respond by playing safe. A score shared early, while the team can still act on it, feels like information. It builds competence without taking the work away from the people who made it.
There is an audience side too. The insight behind this piece is that authenticity thrives in the cracks of perfection. The rough, human edge that makes an idea feel real is often the first thing a nervous approval process polishes away. Beth Johnson of RP3 Agency, writing in a Forbes Agency Council panel, warned that "without sharp strategy and human judgment, AI will only accelerate a tsunami of sameness." Used well, pre-testing gives the rougher idea the evidence it needs to survive the polish.
What is creative pre-testing, and what has changed?
Creative pre-testing means assessing ads or concepts before they run, to predict how they will perform and find weaknesses while they can still be fixed. Traditionally that meant panels and focus groups, weeks of fieldwork and a fee only a TV budget could justify.
AI has changed the economics. Concepts can now be scored for predicted outcomes, such as how likely people are to click, in minutes rather than weeks. That puts pre-testing within reach of the social and digital work that never had a research budget, which is most of what independent agencies make.
One caution. A score predicts the thing it was built to predict. A clickability score tells you about response to the ad in front of someone. It says nothing about distinctiveness, long-term brand building or whether the idea is true to the brand. Those still need people with taste.
How do I use pre-testing without killing bold ideas?
- Agree the job before you score. Is this ad meant to earn a click, a memory or a share? Only use a score for the job it actually measures.
- Score routes, not finished ads. Test rough concepts early, when changing them is cheap and nobody is emotionally attached to the final cut.
- Show the creative team first. Share results with the people who made the work before the client sees them. It protects their autonomy and turns the score into a brief for improvement.
- Read scores as a diagnosis. Look for what is dragging a concept down, such as the opening frame, the hook or the line, and fix that element rather than binning the whole idea.
- Protect one bold route. Agree in advance that the most distinctive concept goes forward if it scores within an acceptable range, even if it is not top of the table.
- Confirm in market, cheaply. Put a small, pre-agreed budget behind the leading routes before scaling, and feed the real results back so everyone learns what the scores do and do not predict.
Where does the evidence sit, before and after pre-testing?
| Stage | Without pre-testing | With pre-testing |
|---|---|---|
| Concept review | Opinion; the loudest voice wins | Scores inform craft judgement |
| Client presentation | "Trust us" | Evidence behind the bold route |
| Media launch | Real money tests unproven work | Budget backs routes already scored |
| Week-one dip | Bold idea paused | Checked against the pre-test, then adjusted |
| What the team learns | Play it safe next time | Which elements work, and why |
The shift is small on paper and large in practice. Evidence moves from the end of the process, where it can only judge, to the start, where it can help.
Where should AI help, and where should it stay out?
The same Forbes panel offers a useful split. Multiply's Matt Jayson says campaign setup and QA is the most automatable work: "For one high-growth client, that's easily 20 hours a week of invisible work." SWOON MEDIA's Natacha Grey describes value shifting "from making assets to making better decisions at scale."
That is the right division of labour. Let machines handle setup, QA, resizing and scoring, so people have more time for the idea itself. Hegarty's point stands: data can inform an idea, but it does not invent one.
This is where a scoring tool earns its place. Defy Monitor scores concepts for predicted clickability before media spend, and its scoring is validated against real ad performance, so a team can know a creative will work before paying for it. You can see how that fits with the rest of the Defy platform. The decision about which idea is worth making stays human.
What should I say to a client who wants proof before they buy the bold idea?
Say yes, and show them. A client asking for proof is not the enemy of creativity; they are usually a marketer whose boss will ask the same question. Give them a score, the reasoning behind it and a small in-market test plan. Then make the case for the idea on its merits, with the numbers on your side rather than against you.
