Ford's $19.5B EV Write-Off and the $3M Question Nobody Asked

The technology was there. The investment was there. The customers weren't.

Ford wrote off roughly $19.5B on its EV strategy.

Estimated advisory fees for the consultancy later brought in to figure out what was holding EV buyers back: about $3M.

Sit with that ratio for a second. Those numbers are the reported and estimated ones, not audited figures from my desk. But the order of magnitude is the point, and the order of magnitude is not in dispute. The write-off is roughly six thousand times the fee for the work that asked why customers weren't buying.

Almost everyone reads that ratio as a story about consultants. It isn't. It's a story about sequence.

The cars were never the problem

Ford bet heavily on a future in which EV demand would keep accelerating. Given what the market looked like at the time, that was a defensible bet. Adoption curves were pointing up. Regulation was pointing up. Every competitor was moving in the same direction, which in a boardroom feels a lot like evidence.

And the company executed. It built the plants. It built the battery supply. It built vehicles that reviewers liked. The engineering was real.

The problem wasn't building the cars. It was understanding what would actually make people buy them.

That's a different discipline, run by different people, on a different budget, usually with a fraction of the rigour. A company will model tooling costs to two decimal places and then take the demand number from a slide.

"I would consider an EV" is not a purchase

Here is where the money went.

Customers can say they want EVs. Buying one is a different decision.

Ask a thousand drivers whether they would consider an electric vehicle for their next car, and you get a warm, agreeable, upward-trending number. Nobody is lying. Considering an EV is genuinely pleasant to imagine. It signals something the respondent likes about themselves. It costs nothing in the moment.

Then the same person stands in a dealership with a real price tag, a real charging situation at their real apartment building, a real question about what this thing is worth in four years, and a real memory of how the last car purchase went. A different mechanism takes over. It was always going to take over. It just never showed up in the study.

Billions are lost in the gap between what customers say and what actually drives them. That gap has a name, and it's the most expensive thing in marketing: the say-do gap.

Why a new category makes the gap wider, not narrower

In an established category the gap is often survivable. Ask someone which shampoo they'll buy and habit answers for them. The stated answer and the real answer are close, because both are being read off the same well-worn track.

A new category has no track.

So when you ask about EVs in 2021, the respondent has to build an opinion on the spot. They assemble it out of what they've read, what their neighbour said, what feels modern, and what they'd like to be true about their own future behaviour. That construction is fast, sincere, and disconnected from the machinery that will actually run the decision two years later. It is exactly why customers can't tell you why they buy.

Then the research industry hands that construction back to you as a demand forecast, with a confidence interval on it.

There's a second amplifier. Early adopters answer surveys about new categories with more enthusiasm than anyone else, and they buy first. So the first cohort of real sales confirms the forecast. The curve looks validated. Capacity gets expanded on the strength of a signal that only ever described the front 5% of the market, right at the moment the next 60% were about to behave completely differently.

What would have had to be measured instead

Two changes, and neither is a bigger sample.

1. Stop asking for reasons

Buying decisions are formed intuitively and justified afterwards. Decades of neuroscience say so, and every conventional questionnaire ignores it by design. Reaction-time based implicit measurement gets at what a category is actually linked to in a buyer's head before they have time to assemble an explanation for you. Fast responses signal intuitive certainty. Slow ones signal deliberation, which is not what a purchase runs on. That's the Frame, Measure, Infer sequence.

For an EV that means you stop measuring "purchase intent" and start measuring what the category is bound to. Freedom or constraint. Progress or compromise. Status or sacrifice. Those bindings predict what happens in the dealership. The intent score predicts what happens in the next survey.

2. Separate the drivers from the passengers

Correlation leads astray. In a real market, income, environmental attitude, urban living, charging access and openness to new technology all move together, and standard statistical models get biased by that multicollinearity. So you get a driver ranking that looks clean and is wrong.

Causal AI is what separates the factors that cause demand from the ones that merely travel alongside it. In the EV case that distinction is the whole ballgame. If the real bottleneck is perceived resale risk, no amount of range improvement fixes it. If it's charging anxiety, price cuts are money set on fire. Both hypotheses fit the correlation data. Only one of them is worth $19.5B.

This is what testing whether a product will sell before launch is supposed to do, and it is a different exercise from a concept test with a top-two-box score attached.

The $3M was never the risk

The consultancy was brought in to better understand what was holding EV buyers back and to help reshape the offer. That is genuinely the right question. It's the question the entire strategy rested on.

It just got asked after the capital was committed.

That's the pattern I keep seeing from the inside. The demand question is treated as a marketing question, so it gets a marketing budget and a marketing timeline, and it arrives after the industrial decision is already sunk. Then when the numbers miss, somebody spends real money to find out why. The diagnosis is competent. It's just a post-mortem.

Wrong remains wrong. Everything built on top of a wrong demand assumption inherits the error, and it compounds silently until an accounting line makes it visible. That's what a wrong market research insight actually costs, and it never gets booked against the research budget that produced it.

Roughly 5% of brands grow, 5% of products survive, 5% of campaigns work. The same 5% keeps showing up, and what they share isn't a better forecast. They stopped guessing about why people buy before they signed for the factory. The whole story is in THE TOP 5%.

Isn't it time to wake up?

Ford's EV write-off: frequently asked questions

Why did EV demand forecasts turn out to be so wrong?

Because most of them measured stated intent. Asking people whether they would consider an electric vehicle produces a warm, agreeable number, and that number was extrapolated into a demand curve. Buying one is a different decision, made against price, charging access, resale value and habit. Stated interest and actual purchase are two different variables, and only one of them fills a factory.

Was hiring a consultancy the mistake?

No. The reported advisory fee is a rounding error against a multi-billion write-off, and the diagnostic work of asking what holds buyers back is the right work. The problem is sequencing. That question was answered properly after the capital was committed, not before. The cost is never the advice. The cost is the assumption the advice arrived too late to test.

What is the say-do gap in a category launch?

It is the distance between what customers report about a new category and what governs their behaviour once real money and real inconvenience are involved. In an established category the gap is often small, because habit carries the answer. In a new category there is no habit to draw on, so respondents construct an opinion on the spot. That constructed opinion is what gets modelled.

How do you test demand for a new category before committing capital?

Measure implicitly rather than by asking for reasons, so you capture the associations that form before a respondent builds an explanation. Then separate causes from correlates, because in a real market price, availability, income and attitude all move together. The deliverable you want is not a forecast number. It is a named mechanism you can act on and stress-test.

Dr. Frank Buckler is the founder of SUPRA and a pioneer in Causal AI for marketing. He has applied implicit research methods across FMCG, pharma, financial services, and insurance for over 25 years. His current book is THE TOP 5%.