"John Lewis Is Marketing to Machines Now"
September 9, 2026
Why one of Britain's best-known retailers is creating content for people, and the machines helping them decide what to buy.
Welcome to this week's Takumi AI Brief.
John Lewis & Partners has spent nearly two decades making one of the most anticipated Christmas ads in Britain.
Now it's built a content studio inside its Oxford Street flagship to produce videos, creator content and conversations all year round.
The reason is worth sitting with.
AI is starting to send people shopping.
John Lewis says product searches coming through AI agents have grown from 0.3% to 2.5% in a year.
That's more than eight times higher.
The absolute number is still small.
The direction isn't.
01 · The Trend That Actually Matters
When most marketers hear AI search, the conversation goes technical fast.
Schema. Product feeds. GEO. Structured data.
All of that matters.
But John Lewis is doing something more interesting.
It's making more content.
The retailer has opened The Box Studio inside its Oxford Street store, where creators produce content around products and trends daily. It's also launching original programming, including a social-first series hosted by Angela Scanlon.
John Lewis managing director Peter Ruis described the growth in AI-driven product searches as "exponential," happening across every age group.
This isn't just a traffic story.
It tells us something about how AI is learning which brands and products deserve to be recommended.
And the answer may be much broader than optimising a product page.
02 · The AI Shift to Watch
For years, marketers created content for two reasons.
To attract attention. To persuade people.
Now there's a third job.
Helping machines understand the brand well enough to recommend it.
Think about how someone actually shops through ChatGPT or Gemini.
They're unlikely to search "John Lewis toaster."
They're more likely to ask what's a good toaster for a small kitchen, what to buy their parents for their anniversary, which coffee machine works for someone who hates complicated technology.
The AI has to work out which products belong in that answer.
To do that, it needs context. Product information helps. Reviews help. But so do demonstrations, expert opinions, creator conversations, editorial content, and the wider reputation surrounding the brand.
Which leads to a slightly counterintuitive conclusion.
The answer to AI search may be more brand marketing, not less.
03 · If I Were in the CMO Chair
There's been a temptation to treat GEO as the next version of SEO.
Find what the algorithm likes. Optimise for it. Move up the rankings.
I think that view is already too narrow.
We saw recently how fast ChatGPT 's citation behaviour around Reddit, Inc. could change, an 86% collapse in a matter of days. Optimising around one source or one model's current preference is fragile by design.
John Lewis is taking a more interesting route.
It's building a richer ecosystem of information around the brand.
Call this Ambient Authority. Not authority engineered for one algorithm, but authority that exists everywhere at once, so it survives no matter which source a given AI system happens to trust this month.
Some of that content reaches customers directly. Some travels through creators and social platforms. Some becomes evidence AI systems use when deciding what to recommend.
The lines between brand, content, search, commerce and GEO are disappearing.
Which has a much bigger implication for CMOs.
AI visibility can't sit in a corner of the SEO team anymore.
04 · If I Were in the CMO Chair
I'd ask a deceptively simple question.
If an AI knew everything publicly available about our brand, would it know enough to recommend us?
Not simply find us. Recommend us.
→ Would it understand what makes the product different?
→ Would it know who it's right for, and when it should be considered?
→ Would it find independent evidence supporting what we claim?
→ Would it encounter genuinely useful customer conversations, not just our own advertising?
That's where the next competitive advantage may sit.
AI is very good at summarising what already exists.
It's considerably harder for it to invent authority that isn't there.
The CMO Takeaway
The most interesting thing about John Lewis building a creator studio isn't the studio.
It's who the content is now being created for.
The customer still matters. Of course they do.
But between the brand and that customer, another influence has appeared. An AI that searches, compares, filters, and increasingly recommends.
For years, we worried about whether our brands were searchable. Then whether they were shareable.
The next question is harder.
Are they recommendable?
John Lewis seems to understand the answer requires more than technical optimisation.
It requires giving both humans and machines something actually worth talking about.
Which might be the real irony of AI search.
The machines may end up making good, old-fashioned brand building more important, not less.
Mohit Lodha, Founder, Takumi AI