Your PIM is a Jenga tower nobody wants to touch. Why customer-driven product data needs agility your stack never had, and three actions for directors.
Why your product data is the most important system you’re afraid to touch - and what to do about it.
A buyer I’ve been speaking to recently sent me an email.
He’d had a diagnostic done on his product data model. The conclusion was that they needed to make their PIM more agile.
His question to me was simple. Which attributes are prime candidates for optimisation, and which definitely aren’t?
It’s a fair question. It’s also the wrong one.
Because what he was really being asked to do was put his fingers in the fire on the one system in the business that nobody wants to touch.
It’s Not a PIM. It’s a Jenga Tower.
Think about how your Product Information Management system actually came to be.
It started life as one piece. Just a place to hold product data. Brand, colour, size, weight. The basics.
Then you stacked things on top of it. Store inventory. ERP. Suppliers. The web catalogue. The mobile app. The marketplace feed.
Every team in the business now relies on it. It’s become the oracle. The single source of truth for how products move through your operation.
And now somebody’s asking you to wiggle the piece at the bottom.
Nobody wants to play PIM Jenga. Because everyone knows what happens when the tower falls.
That’s the real reason these projects stall. It’s not because retailers don’t see the problem. They see it clearly.
It’s because the cost of getting it wrong is the cost of getting everything wrong, all at once.
Two Types of Product Data, One Big Problem.
There’s a distinction most retailers haven’t made yet, and it changes everything once you see it.
Your product data isn’t one thing. It’s two things, doing two very different jobs.
First, you’ve got operational data. Weight. Dimensions. Packaging. SKU count. Warehouse location. Whether it ships to France or only the UK. Whether you're selling it or a marketplace partner is, or both.
This stuff is evergreen. A 50kg bag of cement is going to stay a 50kg bag of cement. It only changes if the manufacturer changes the spec, or if you change how your business operates. And when that happens, you do one big update, and everyone moves on.
Then you’ve got descriptive data. Colour. Style. Occasion. Material. Mood. The stuff your customer actually shops by.
That’s what I’ve started calling the sales surface. It’s the data that constructs the experience. It’s how products get found, get understood, and get bought.
Operational data is how your business runs. Descriptive data is how your products sell. They are not the same thing – and they don’t need the same speed.
Treat them like they’re the same thing, and you end up with a single rigid system trying to be two opposite things at once.
One needs stability. The other needs agility.
And right now, almost nobody has built for that distinction.
This Isn’t Coming From the Top. It’s Coming From the Customer.
Normally, big changes to product data come from a business decision. Management decides to enter a new category. To launch a marketplace. To sell internationally.
Everyone gets the memo. The stack changes. The work happens. Done.
This time is different. The pressure isn’t coming from upstairs. It’s coming from outside.
Customers are searching in full sentences now. They’re asking AI agents to find them a wedding-guest dress for an outdoor September wedding in Cornwall. They’re shopping by mood, by mission, by moment.
They don’t care that your product data says “red, cotton, size 12.” They want to know if it works for the occasion in their head.
That’s a market force. And market forces don’t wait for your roadmap.
The realisation that how customers shop isn’t being met by how systems are delivering – it’s been slow to bubble up. But now it’s starting to hit home.
The Agility Nobody Built For.
So, what does descriptive agility actually look like? Let me get specific, because this is where it gets real.
You need to be able to rename an attribute. “Length by width” becomes “Dimensions” because that’s how a customer would say it.
You need to be able to consolidate values. Twelve shades of red collapse into one. “Crimson,” “scarlet,” “cherry” all become “red.”
You need to be able to add brand new attribute groups. Wallpaper didn’t need “colour palette” last year. This year, with atmospheric and pastel trending, it does.
You need to be able to add new values inside existing groups. Microwaves never used to be on your site. Now they are. Suddenly, you need wattage, capacity, finish, and smart control.
These changes aren’t edge cases. They’re the everyday reality of intent-led retail.
And in most current systems, each is a project.
And Now It Gets Worse.
A retailer told me recently that their PIM has a hard cap of 200 attributes. Total. Globally. Across every product in every category.
Two hundred.
Now think about what intent-led retail actually requires. Around 30% of your attributes will be global – colour, brand, price. The other 70% are category-specific. Wallpaper has “pattern” and “palette.” Microwaves have “wattage” and “capacity.” Dresses have “occasion” and “neckline.”
You can blow through 200 in three categories.
That’s not a strategy problem. That’s a system limitation. The kind of limitation that quietly forces a re-platform conversation nobody planned for.
This isn’t about deciding which attributes to optimise. It’s about whether your system can hold the attributes your customer needs in the first place.
The Ghosts in the Stack.
Now for the bit that should genuinely worry you.
I’ve seen this play out a few times. A retailer realises they need to make a change to a core system. They go to do it – and discover the only person who understood how it worked left the business three years ago.
One client had to bring back their old IT director – the one they’d made redundant – on a very expensive contract, just to explain how their systems talked to each other.
Another business had two retired consultants who came in occasionally to keep an ancient platform alive. One of them sadly passed away. The business genuinely had to ask: What do we do if the other one goes too?
These aren’t horror stories. They’re what happens when you build a Jenga tower for fifteen years and never label the pieces.
And it’s the reason most PIM modernisation projects either don’t start or end up costing a small fortune to a global consultancy.
A Source of Truth That Refuses to Sit Still.
So we land at the contradiction every retailer is going to have to make peace with.
You need a product source of truth. One place where the right version of the product lives, so every channel inherits the same thing – site, app, in-store kiosk, marketplace, agent.
But there’s a catch.
Truth is an evolving concept.
Because customer intent moves. New ranges arrive. New behaviours emerge. New ways of describing products show up that didn’t exist eighteen months ago.
A source of truth that can’t evolve isn’t a source of truth. It’s a fossil.
You need a product source of truth. But you also need to remember that truth itself is an evolving concept.
That’s the design brief almost no PIM was built for.
And it’s why this is the conversation senior commerce leaders need to start having now – not when the next taxonomy project lands on someone’s desk in 2028.
Why This Matters Right Now.
If your descriptive data is rigid, you can’t respond to trends. You discount instead. You watch the AOV slide.
If you can’t add new attributes when you launch new ranges, those ranges become dark aisles. Customers can’t find them. They convert at a fraction of the rest of the site.
And if your data can’t speak the language of agentic discovery – long-form, conversational, intent-rich – then the AI agents that are about to shop on behalf of your customers will quietly skip past you.
This is a quiet crisis. And it doesn’t announce itself in your trading meetings. It shows up as a slow leak in conversion. A creeping rise in bounce rate. A discount habit you can’t seem to break.
Three Actions for E-commerce Directors.
So if any of this lands, this is where I’d start. None of these are rip-and-replace. All three you can begin this quarter.
1. Plan for descriptive agility – even if you don’t know the answer yet.
Agility in this domain is specific. It means being able to rename attributes, consolidate values, retire categories, and add whole new groups without a five-year taxonomy project. It means wiggling one piece of the Jenga tower without bringing the house down. The ‘how’ will be different for every business. But the goal is the same: get this domain out of the rigid stack and into something that can keep up with the market. This is exactly where layering an intent-led enrichment system over your existing PIM – rather than rebuilding it – earns its keep.
2. Build for evolving truth, not static truth.
Stop thinking about your product data as something you set and forget. The next generation of commerce – agentic, conversational, intent-led – is going to demand product data that’s richer, deeper, and continuously evolving. Q&A. Highlights. Good-to-know. The stuff Google’s new attribute system is already making space for. If your product source of truth can’t evolve, it’ll quietly stop being the truth. And the channels that need it most will start working around it.
3. Demark your operational and descriptive data.
Take a long look at your product data model and draw a line down the middle. Operational on one side: availability, fulfilment, ratings and sizing. Descriptive on the other: colour, occasion, style, mood. Until you’ve made that distinction explicit, you can’t treat them differently. And until you treat them differently, you’ll keep applying operational governance to a sales surface that needs to move at the speed of customer intent. And that’s extremely fast-moving!
None of this is easy. I’m not pretending otherwise.
But doing nothing isn’t a strategy anymore. It’s just a slower way of losing.
You can’t replatform your way out of this. You can’t consult your way out of it either. You have to start by separating the two jobs your product data is doing – and giving each of them the system they actually need.
That’s the work. And it starts with one honest conversation about which Jenga pieces are load-bearing, and which ones were just there to make the tower look taller.
Until Tuesday after next,
Warren
About FoundIt!
FoundIt! is the intent-led commerce platform that delivers +4-6% sitewide revenue for multi-category retailers, including M&S, B&Q, John Lewis, Saks and Neiman Marcus. We close the gap between how your catalogue is organised and how your customers actually shop – without a rip-and-replace.