The campaign flopped. The media mix model said why.
A color safe laundry brand went dark for five years. Then it spent real money on a comeback campaign. The model measured the effect of that campaign at close to zero. That null result was the most valuable thing the brand bought all year.
Five years of quiet
The logic for going dark seemed smart. Premium product, good margins, so why pay to promote it. Let it sell itself and pocket the difference.
It did not sell itself.
With nobody explaining what the product was for, shoppers guessed. They saw an expensive bottle of bleach, figured it was a fancier version of the regular kind, and used it on their whites. Once a product looks like a pricey take on something cheaper, all that is left is the price. People reached for the cheaper bottle.
Then the second thing happened, and that is the one that threatens the business. When a product stops moving, stores pull its shelf space, and in this category the shelf is almost the whole game. Lose it and you are not competing, you are disappearing. Five years of quiet had turned a margin play into a slow bleed.
The comeback
So the brand came back. Big campaign, high hopes. The message was clear and it was correct. Whatever regular bleach does for your whites, this does for your colors.
It flopped. Not softened. Flopped. After all that buildup and real money, it barely moved a thing.
A true message, competently made, behind real budget, and it did nothing.
The comfortable answer, and why it was wrong
Underperformance after a big launch has one comfortable explanation, and it is usually wrong. Not enough. Not enough weight, not enough time, not enough frequency. That answer is popular because it requires nobody to admit the work was bad, and it comes with a built in stay of execution. Give it another quarter.
Without a model, that is the reading, and buying more is the move. They could have put millions more behind a creative that was never going to work, while the real problem kept bleeding underneath. They were losing distribution. More advertising was never going to fix that.
The elasticity nobody was looking at
The model showed more than the flop. Going back across the silent years, price elasticity kept climbing. Shoppers were reacting harder to every price change, well before any new ad ran. That is what happens when people stop seeing the value in a product. Strip away the reason to pay more and all that is left is the number on the shelf.
No campaign report produces that finding. It only appears if the model lets price sensitivity move over time instead of fitting one number for the whole period.
That reframed the problem. This was not a media weight problem. It was a meaning problem, with a distribution problem downstream of it.
The wrong person was holding the bottle
So the team asked the most basic question in marketing, the one that gets skipped because everyone assumes it was settled years ago. Who actually buys this, and why.
Moms, shopping for their kids' clothes.
Now look at the failed spot again. Two professional women getting stains on their clothes, and a line about doing for your colors what bleach does for your whites. Functional, true, and pointed at the wrong person. The mom holding the bottle is not weighing bleach chemistry while she scrubs. It is personal. They are her kids' clothes.
So they scrapped it. The new spot showed kids playing in the grass, wrecking their colored clothes the way kids do, with the product saving them. Same function, now with feeling behind it, talking straight to the person holding the bottle.
The model popped. Same brand, same product, same channels. The creative was the only real change, and the model saw it.
A media mix model is not just a channel calculator
That is the part almost nobody says out loud. With good data underneath it, a model can tell you whether the creative landed.
The mechanism is not exotic. When the product, the price structure, the distribution and the channel mix hold roughly steady and the creative changes, you have a controlled comparison sitting in your own data. The model holds the other drivers constant and reports what is left. That is the difference between two ads, measured in sales rather than in opinions.
One model, four answers:
- It caught a flop before millions more in wasted spend went behind it.
- It found years of rising price sensitivity and named the real problem, which was distribution, not advertising.
- It pointed the team at the actual buyer.
- It proved the rebuilt ad worked.
Channels, pricing, distribution, creative, all read out of the same model. It also changes who wins creative arguments, which today are usually settled by whoever is most senior in the room.
Why most models cannot do this
None of that required an exotic method. It required data most advertisers do not have in one place. Spend has to be clean and separable down to the individual creative. Sales and distribution have to line up on the same calendar. Price has to be in the model as a live variable, free to move. History has to reach back far enough to cover the quiet years, not just the last four quarters.
That is the real reason most media mix models never produce anything but a channel ROI table. Not the math. The plumbing. Self serve measurement platforms assume you have already built it. Most mid market advertisers have not, because it takes a data team they do not have.
What to do with this
Three questions worth asking about your own measurement.
- Can it tell one creative from another, or does it only see channels.
- Can price elasticity move over time, or is it fitted as a single number for the whole period.
- When a campaign underperforms, can it separate not enough media from the wrong ad.
If the answers are no, we can help you setup for this level of measurement. Reach out and we will take a look.
Real media mix modeling work, brand withheld. The pattern is not rare, it is closer to normal. We build the data foundation and then run the model on top of it, refreshed weekly, for advertisers spending five to fifty million a year on media with no analytics team of their own. Get in touch.