For a long time, the retail industry ran on a shared piece of fiction. Retailers knew their stores. Brands knew their products. Both wrote each other reports based on partial information, and both pretended they could see the shelf. They couldn't. And neither could anyone else.
That fiction is finally ending. The shift won't be loud - there will be no single, headline-grabbing moment when the industry "switches over". But over the next decade, retail will quietly become an industry that operates on a single, verified view of the shelf. And when it does, it'll be more profitable, less wasteful, and far less adversarial. Everyone wins. I don't say that lightly.
Why now
Three forces have lined up.
First, retail has digitised at the point of sale. Modern POS providers ship real-time event streams as a feature, not a custom build. The transaction is becoming a first-class digital object - geotagged, line-itemised, available the second it happens. Even five years ago this was rare. Today, across the partner networks tapestry® works with, it's the default.
Second, the cost of being late has gone up. Margins are thinner. Promotional mistakes are more visible. Stockouts don't just lose a transaction any more - they push shoppers to a competing channel that they may not come back from. The penalty for a slow signal has compounded into a real number on the P&L.
Third - and this is the part most people miss - AI has become useful for the retail middle. Not the headline AI of the consumer apps, but the operational AI of the everyday: ranked lists, evidence-backed recommendations, plain-language summaries against trusted data. The shape of work in retail is changing because the assistant in the corner of the screen finally has something useful to say.
The penalty for a slow signal has compounded into a real number on the P&L.
The system of trust
The thing that's missing in most of the industry's conversations about AI isn't model quality. It's provenance. The question retailers and brands ask - quietly, in every meeting - is: whose data is this, and what does it actually say?
For decades, the answer was different depending on who you asked. The retailer's number, the brand's number, the panel data number, the syndicated number - four different reports, four different definitions, four different decisions. The data became a thing to argue about, not a thing to act on.
What's changing now is that the shelf itself - verified, real-time, line-itemised - becomes the single source. Both sides see it. Both sides agree it's correct. And the conversation moves from "is this data right" to "what should we do about it". That's the system of trust. It sounds boring, and it's the most important thing happening in retail technology.
The three properties of trustworthy retail data
- Verified at the source. POS-grounded, line-itemised, captured at the moment of sale. No re-keying, no estimation, no reconciliation.
- Governed in the share. Aggregation rules, anonymity controls, audit trails - every share logged, every payment traceable.
- Bounded in the question. AI that answers analyst-grade questions inside its competence - and refuses, gracefully, the questions it shouldn't.
What changes for retailers
The shorthand: better decisions, faster.
The longhand is more interesting. Three things change in a retailer's day-to-day when they move onto shared shelf truth.
The first is that exception management compresses from weeks to hours. A range gap that used to surface at quarter-review now surfaces as a notification on Tuesday morning, with a recommended order top-up attached. The signal-to-action loop shrinks dramatically, and so does the leakage.
The second is that supplier conversations get a lot shorter - and a lot better for the retailer. When both sides walk into a category review looking at the same numbers, the meeting stops being a negotiation about data and starts being a decision about deals. Most retailers I've talked to report that the quality of those conversations improves immediately. The numbers don't lie, on either side.
The third is the part that surprises people: reporting becomes a revenue line. When governed shelf data is something brands actively want to pay for, the retailer's reporting infrastructure flips from cost centre to contribution centre. It's modest at first - but it compounds, and it changes the strategic conversation about what retailers are actually selling.
What changes for brands
Brands stop building reports. That's the bluntest way to put it.
For a long time, brand teams have spent enormous amounts of time stitching together panel data, retailer exports, syndicated reports and their own internal sell-in numbers - trying to assemble a picture of the shelf that was always partial, always a few weeks late, and always slightly different from what the retailer said. The opportunity cost has been crushing.
Shared shelf truth solves that. The brand's analyst suddenly spends 80% less time assembling and 80% more time deciding. The category manager's monthly review becomes a daily one - not because anyone made them, but because the data is available and the work is now possible.
It also reorders the brand's commercial conversations. When you can show a retailer the impact of your last promotional cycle - net margin, halo, cannibalisation - instead of arguing for it, you stop being a vendor and start being a partner.
The promo retune, in real time
A practical example. A supplier ran a multipack promo across one of our retail partners this past quarter. Sell-through looked fine on the headline. The classic story: sales up, ROI claimed, everyone happy.
The shared data told a different story. The promo had cannibalised the supplier's higher-margin single-unit SKU at a rate that erased most of the margin uplift. Halo was modest. Net contribution was barely positive. Without shared shelf truth, that promo would have been repeated next year. With it, the supplier and retailer sat down and redesigned the mechanic together - switching to a multi-buy across two SKUs that lifted both. Margin up, no more cannibalisation, supplier happy, retailer happy.
That conversation, ten years ago, was impossible. Today, it's an afternoon meeting.
The brand's analyst suddenly spends 80% less time assembling and 80% more time deciding.
Where HANK fits
I've written about HANK before, so I'll keep this short. HANK isn't AI in the consumer sense of the word. He's an Intelligent Assistant - bounded to the analyst-retrieval space, action-attached, confidence-scored. He doesn't have an opinion on what colour to paint your store. He has a very strong, very evidenced opinion on which fourteen SKUs are about to run out, ranked by exposure, with a draft order top-up waiting for your approval.
The reason that matters in the context of this piece is that HANK only works because the data underneath him is trusted. An assistant on top of untrusted data is a hallucinator. An assistant on top of verified shelf truth is an operator's analyst. The difference is everything.
And here's the part people don't expect: HANK works across the retailer–brand boundary. The same assistant, looking at the same data, can prepare the buyer's brief for a category review and the supplier's response - in the same window, in the same terms. That's structurally new. It's why the conversations get shorter.
Closing
The story I've been telling for the last few years isn't really about AI. It's about trust.
AI is the accelerant. Verified shelf data is the foundation. Governance is the constraint that keeps the system honest. Put those three together, give them five years to compound, and retail will be unrecognisable - not in the surface aesthetics, but in the underlying operating model. Less arguing about data. More acting on it. Fewer surprises. Tighter relationships between retailers and brands. Better profit on both sides. Real money for shoppers, in the form of better-stocked shelves and sharper pricing.
None of that's a moonshot. It just requires the industry to commit to one shared idea - that the shelf is the truth - and to build the rails to share it, governed and audited, between the people who already touch it every day.
That's the work. We think it's worth doing.
