The Perception Engine: Why Marketing Fails by Design

In 2009, Mike Cessario was living in Denver. Some friends of his were playing the Vans Warped Tour, and he went to watch.

Monster Energy sponsored the tour, so the bands drank from Monster cans on stage. The cans didn't hold Monster. The musicians had swapped the energy drink for water, so they could stay hydrated through their sets.

The sponsor got the can. The musicians got the water.

Cessario kept turning over one question after that. Why does all the funny, loud, cool branding go to junk food?

The idea that waited

Five years later, he got a chance to use it. In 2014 he was working in advertising, on a public service campaign about the health risks of sugary energy drinks. He pitched a stunt: canned water, made to poke fun at energy drinks.

The client didn't like it.

So he kept working on it in his spare time. In 2017 he trademarked a name: Liquid Death. Then he spent two years taking it to people who could fund it.

In his telling, investors and people in the beverage business said the can looked too much like beer. They said retailers would never stock anything with "Death" on it. "Nobody was going to just write a check for that idea," he told CNBC.

I don't know the reasons behind each of those no's. I can guess at them, and none of my guesses is foolish. A public service client has a message to protect, and an investor has to explain every check.

In 2018, Cessario stopped asking. He made a 3D image of the can and set up a Facebook page, so Liquid Death looked like a real product.

He shot a two-minute commercial for $1,500. Then he put a few thousand dollars of his savings behind it as paid media.

There was no water yet. There was only the ad.

"Four months in, the video had 3 million views," he said. The page had almost 80,000 followers, more than Aquafina had on Facebook at the time.

Hundreds of people wrote in with the same questions. Is this real? Where can I buy it? Beverage distributors called, looking for a salesperson.

That was enough for investors. In January 2019 he raised $1.6 million in seed money, and Liquid Death started selling cans online that same month.

It sold $2.8 million of water that year. By 2023, its products rang up $263 million at the register. In March 2024, a funding round valued the company at $1.4 billion.

So what changed between 2014 and 2018?

The idea didn't.

Same water. Same can. Same joke.

What changed was what stood between the idea and the people who would buy it. In 2014, a client stood there. In 2018, nothing did.

The 2014 version

If you work in marketing inside a company of any size, you know the 2014 version of this story. You may have lived it.

The idea starts sharp. Then it meets the review, the research, the format that fits the media plan and the quarterly check-in.

Each step is sensible. By launch, the edges are gone.

Then the report comes back. Everything the agency controls is green: reach is up, the impressions were delivered, viewability hit its target. Everything the business cares about is flat.

Someone says the creative played it safe. The agency gets replaced, the tagline gets rewritten, and the next campaign launches with a bolder look.

Six months later, the report is green again. Sales are still flat.

It's easy to read that as a talent problem. I don't think it is.

I think there's a machine between the idea and the customer. It decides what gets planned, measured and rewarded, and most of it never looks at the customer.

I call it the Perception Engine. It has five parts, and none of them needs anyone to be lazy or dishonest.

Every person involved does what makes sense from where they sit. That's what makes it hard to see.

Liquid Death is useful here because it ran without one. Put each part next to Cessario's path, and you can see what the part does.

The five parts

1. The Empathy Inversion

Most campaigns get planned from the inside. The team sits down with the product, the price, the channels and the brand guidelines. They talk about what makes the product good and what the company stands for.

All of that is real. None of it is what the customer is thinking about.

Cessario started from the other end. He didn't start from water.

He started from something people already did: musicians carrying water in an energy drink can. The need was there before the brand was.

Aristotle noticed something about persuasion a long time ago. The speaker never supplies the whole argument. The listener fills in the missing piece from what they already believe.

A message built in a conference room leaves a gap, and the customer fills it with their own day. If their day has nothing to do with your mission statement, the piece never fits.

That's the Empathy Inversion. The message gets built for the room it was written in.

2. The Research Ceiling

So the campaign underperforms, and the next step is to ask customers what they want.

Ask people what they want from bottled water, and my guess is you'd hear the sensible answers, like clean and healthy. That's close to what the water aisle already said, with its mountains and springs.

Gerald Zaltman at Harvard has estimated that around 95 percent of what shapes a buying decision happens below conscious thought. Other researchers put the share lower.

They agree on the direction: much of what drives a choice never makes it into the answer someone gives you.

Itamar Simonson's research adds a twist. The act of asking can shape the preference people report, so part of the answer gets made in the moment of asking.

Cessario didn't ask anyone whether they wanted water that looked like beer. He put the can in front of strangers and watched what they did.

That's the Research Ceiling. A survey reaches the sensible layer, and the sensible layer usually sounds a lot like the brief.

3. The Measurement Cartel

Now the campaign is live, and someone has to report on it.

In 2014, the Media Rating Council set the industry standard for counting a display ad as viewable. Half of the ad has to be on screen, in a browser tab that's open, for one second. Nobody has to look at it.

Karen Nelson-Field has spent years measuring what people's eyes actually do with ads. Her work suggests an ad starts to stick in memory at around two and a half seconds of real attention. Plenty of ads that count as viewed never get close.

So why does everyone keep the number?

The platform gets to count more of its inventory as delivered. The agency gets to report that it hit the plan. The marketing manager gets a clean report to take upstairs.

Nobody has to coordinate, and nobody has to lie. Each one has a good reason to keep the number, and none can change it without taking the hit first.

I call that the Measurement Cartel, though there's no meeting and no conspiracy. It's three sets of incentives that point the same way, and none of them point at the customer.

Look at what Cessario's test counted instead. Views, yes.

But the reaction that raised his money was strangers asking where to buy a product that didn't exist. No delivery report has a line for that.

4. The Lag Effect

Say a team gets past all of that and builds something that connects. You'd think the results would show up soon after.

They usually don't. Most companies review marketing every quarter, and brand doesn't move on that schedule.

When Les Binet and Peter Field went through the IPA Databank's campaign records, they found that brand-building effects build over many months. Short-term promotions show up in days.

A brand campaign launches in January. By the end of March the numbers haven't moved much, so it looks like it isn't working.

Budget shifts to search ads and discounts, which show results fast, and the next quarter looks better. What gets missed is that the fast stuff may be spending down awareness the company built earlier.

Liquid Death's first year of sales was $2.8 million. On a large company's quarterly plan, that's a rounding error. It took four more years to reach $263 million.

Every person in the January story made a reasonable call. They made it on a clock too short to see the result.

5. The Position Lock

By now, some people inside the company know something is off. So why doesn't anything change?

Because changing direction costs more, right now, than staying put.

Michael Porter's idea of mobility barriers fits here: the more you've built around a position, the more it costs to leave it. Robert Cialdini's work on commitment points the same way. Once people have defended a position in public, they tend to hold on to it.

Coca-Cola owns Dasani and Smartwater. PepsiCo owns Aquafina. Either could have put water in a tallboy can with a skull on it.

Cessario doesn't sound worried. "It's really hard to replicate the marketing," he told CNBC.

I don't know what happens inside those companies. I do know what the engine predicts.

Their water brands were built for the aisle as it was, along with the shelf deals, the research and the reports that go with them. Leaving that costs a lot, right now, and the cost lands on specific people. Staying costs a little at a time, spread across everyone.

That's the Position Lock. For each person who would have to sign off on leaving, staying is the safer move.

How the parts feed each other

Put the five parts back inside one company's year.

The plan is last year's plan, because questioning it means questioning the people who approved it. The survey checks the plan and comes back sensible, so it agrees with the brief. The campaign launches, and the report counts what was delivered, so it comes back green.

By the second quarter, sales haven't moved much. The delivery numbers say the media did its job, and the brand work hasn't had time to show up.

So brand looks like the weak spot, budget moves to promotions, and the next report looks better.

When the year ends flat, the strategy can't be the problem, because it was approved. The media can't be the problem, because its numbers were green. That leaves the creative.

New agency. New tagline. Same year, starting over.

That's why fixing one part fails. Braver creative still goes through the same survey, the same report and the same quarterly clock.

The case against my example

There's a strong objection to all of this, and it uses my own example.

For every Liquid Death, how many loud little brands burned through someone's savings and vanished? We hear about the one that worked.

The 2014 client may also have been right. A public service campaign may be the wrong place to launch a product.

I agree with both points. The engine does real work. It stops a lot of bad ideas, and most ideas are bad.

My problem with it is narrower. None of the five parts looks at the customer, so the engine can't tell a bad idea from a good one that's simply strange. It stops both.

Cessario's test could tell them apart, because the one thing it measured was whether strangers wanted the thing.

The second objection is that Liquid Death worked because there was no company yet. There was no sales team, no brand guidelines and no board.

That's true, and it marks a limit of the framework. A day-one startup has no engine to fight.

But you don't need to quit your job to borrow what Cessario had. You need to change some of what the engine rewards.

What you can change

If you run marketing, you don't control the whole machine. You do control the brief, the questions you ask customers and some of what goes into the report.

That's enough to start. Here are five places, one for each part.

Start the brief from the customer's day. Before you write anything, write down what the customer was doing right before they'd see your message. Cessario's version was a band on a hot stage, holding a can that wasn't theirs.

Watch what people do, not only what they say. Surveys have their place. But search terms, the pages people visit, where deals stall and why customers leave will tell you things a questionnaire can't.

Keep one number no platform or agency can move on its own. It might be branded search over time, direct traffic, or how many new prospects bring up your point of view. Liquid Death's was strangers asking where to buy.

None of these is clean, so watch more than one. Decide before launch what result would make you stop.

Put brand and promotion on different clocks. Judge short-term campaigns on short-term results. Give brand work a budget that's protected for a year or more, and judge it on things that move slowly.

Put a price on standing still. Add up what it costs to stay where you are, such as the extra ad spend to make up for weak word of mouth. Give that number an owner, and put it next to the cost of changing in the meeting where next year's strategy gets approved.

Next Monday

Next time you open a performance report, read it the way someone outside the company would.

What does each number measure? Who picked it? What would happen to them if it came back red?

You might find the report is right about everything it measures. Then look for the line Cessario's test had and your report doesn't. Did anyone want the thing?

The can on that stage said Monster. Inside was water.

A report can be accurate about the can.

~ Charles