McLaren dropped papaya at home.

When McLaren swapped papaya for the white-and-green Gemini livery at Silverstone, fans watching on Reddit spent practice trying to find the car and couldn't. One called it #HAASILLACMARTIN. Another just wrote that papaya is much better.

That's the whole argument in one complaint. A fan who can't find the car is a distinctive asset doing its job by not being there. A special livery can signal something. It can't build anything. Those are two different jobs, and teams keep selling the first as if it were the second.

I went to the McLaren and Google Gemini launch in London a few days before the race, and I've been chewing on this ever since.

There are two ledgers here, and the confusion comes from letting them blur into one.

On the team's side, a full-livery takeover is a good product. It's a new piece of inventory that didn't exist last season, it's easy to sell, and it gives the partner something clear to point at: we took over the car for the weekend, we had our guests, we brought our creators. That's all real, and it's all worth money. No complaint from me there.

The brand side is where it falls down. For long-term memory, a livery change does almost nothing. The problem isn't selling the takeover. It's the second claim that comes with it - that the same weekend also built the brand. That part has very little to show for it.

Distinctive assets are built by not changing them. Papaya is worth something to McLaren because they've run it for years, until the colour brings the team to mind before the name does. I call that return on memory, ROM - my own shorthand for it, and it sits on top of what Ehrenberg-Bass call mental availability: how easily a brand comes to mind, earned through consistency. Every special livery is a small withdrawal from that. You're spending the one thing that built the asset.

So the honest answer to "Does a livery change build memory?" is no. It works against the way memory is built.

McLaren gets away with it, and not because the change helps them. Papaya is banked so deep that taking it away reads as a deliberate move rather than a mistake.

There's a decade of orange for the white-and-green to play off, so people file it as McLaren doing something.

The reason points inward too: the M2B, the car that started the team, and 1966, the year Bruce McLaren scored their first point at the '66 British GP at Silverstone. When the reason is your own story, the change enhances the brand rather than interrupting it.

It's still a one-weekend thing. A nice way to say you've been here a long time and you honour where you came from, and then the papaya's back. And even the team that cleared the bar paid for it - fans lost the car on track, and the verdict from the stands was that the real one is better.

One thing got buried that shouldn't have. The 1,000th Grand Prix is the most ownable signal McLaren has had this year because no one else on the grid can claim it, and it's been overshadowed by the Gemini takeover and the Bruce McLaren story.

Same weekend, opposite result, which is what makes Cadillac a useful comparison.

Their look is the black-and-white split - black down one side, white down the other - and it's about eight races old.

Season one is when you'd hammer that home until people learn it. Instead, they painted over it in race nine with a full red, white and blue for the Fourth of July.

There's nothing banked yet for the change to play off, so it doesn't read as Cadillac doing something. It just reads as a different car.

The reason points outward, too. It borrows a national holiday that happens to land on someone else's home weekend - Silverstone isn't Cadillac's home race, even though it’s the team's UK base - so it builds the Fourth of July, not Cadillac.

Not every team could pull on an American flag; Cadillac can, as the American team on the grid. But they're borrowing it, not building something that's theirs.

A flag is shared by everyone flying it, especially in America's 250th year, so leaning on it makes them look like every other American thing that week.

I'll be fair to them. They can wear that flag more credibly than anyone else out there, and as a way to get noticed when almost no one can pick you out yet, it isn't stupid.

It's just answering the wrong question. Look at Lewis Hamilton's iconic yellow helmet - he can go purple for a weekend, and you still read Hamilton, because the yellow is banked deep enough to snap back to him.

That's permission earned over years. Cadillac are reaching for permission they haven't earned, over the top of the one thing they'd started to make their own.

The bigger nitpick I have with the Google takeover is that it doesn’t land anywhere.

I got to the launch in East London late - a delayed layover meant I missed Norris and Piastri pulling the covers off.

The car sat on a stage, with Gemini-generated backgrounds moving behind it. Three of Piastri's helmets to one side, a giant helmet you could design yourself, some sim racing, a photo booth at the back, they're planning to shrink down for the paddock.

Every part of it was fine. None of it connected. You go and try this, then you go and try that. It was a floor plan, not a story.

Letting a fan design the livery after the fact is the giveaway. They are being shown what it could do.

Compare that to what the building looks like. The thing I'll remember from the 1,000th is the drivers and old cars lined up together at Monaco - one human moment that threw off a week of content on its own.

Last year, McLaren took over Trafalgar Square with Hilton, and that was a room you could walk through. The wrap is a signal you pay for; the feeling is what you're actually buying. At the launch, the feeling wasn't there.

Gemini needs one thing more than most brands on that grid: for people to feel warmer about AI. The activations didn't do that. They showed what the tool can do - design a livery, ask it what time the race starts, get it to work out how many points McLaren has in the championship.

That's capability, and capability gets people to consider a product; warmth is a different thing, built on how something makes you feel. An AI totting up championship points is about as cold and transactional as it gets, and showing that it can design your livery for you pokes the exact worry people already have.

The giveaway is that the one warm moment I can name without thinking is James Coker being funny and human about it - a creator doing creator things, the only part of the whole thing that wasn't a demo.

The feeling got outsourced to talent because it wasn't built into the room. And from Ben Gosling, who runs the partnership, this is one of the few full takeovers Google have done - which makes the gap between the size of the bet and how little of it was aimed at feeling all the more surprising.

For contrast, look at Anthropic and Williams. On the mechanics, they're almost the same as Gemini and McLaren: branding on the car, an AI worked into the team's operations, their own special home livery at Silverstone, even a launch film.

What's different is the idea about AI that each one sells. Claude's "thinking partner" line is about augmentation — humans and the model working together - which is an argument about how to feel about AI.

Gemini is a demonstration of what the tool does. I wouldn't call either a win; both are unproven, and both launched a story before any results are in. But one of them made the case for a feeling, and the other showed off a feature.

I don't really understand why livery changes have become such a thing these last few years. It feels like one or two teams turn up with a special livery almost every race now. Each one is fine on its own; some are nice. But too much of anything stops it from being special.

Being the odd one out only works when the rest of the grid is consistent, and consistency is a shared resource. When teams keep changing, they wear down the very thing that makes anyone's one-off stand out. It cancels itself.

And I've come round to thinking McLaren's call isn't really a misjudgement - it's a trap.

Silverstone is the loudest, most chaotic weekend of the year, and a historic British team feels the pressure to do something big hardest of all, on track and off it.

So the pull toward a splash is strongest at the exact point a splash is least likely to stick. The livery starts to look like something that had to be done, not something anyone set out to do.

The takeover is a good commercial product that keeps getting sold as a brand exercise.

It signals; it doesn't build. It earns its place only when three things are true at once: the asset is already banked, the reason points inward at your own story, and the change is wrapped in something that makes people feel.

McLaren cleared the first two and missed the third. Gemini showed off a tool when the job was feeling. Cadillac shouldn't be doing this for another two seasons.

So the question to sit with before you sign off on a takeover is a dull one. Name the outcome it's meant to serve, and name the thing you'll measure it against.

If Gemini's goal is a warmer feeling about AI, there's brand tracking that should move - trust, favourability - and you check against it.

If you can't name the measure, you didn't buy brand-building. You bought a splash and called it strategy. Sponsorship is one channel, a big one in F1, and it answers to the same scoreboard as everything else.