Episode 4: Why Supplier Data is Never Usable - and Three Leavers to Fix it

Buy a PIM. Build a supplier template. Set up a portal. Mandate the fields. Wait for clean data to arrive.

This is the plan most businesses run. And it doesn't work. Not because the tools are bad, but because the plan contains a hidden assumption that turns out to be wrong every single time: that supplier data will arrive in a state you can actually use.

It doesn't. It never has. And the gap between what suppliers send and what your systems need is growing, not shrinking.

The hidden cost that everyone ignores

When supplier data doesn't arrive in a usable state, someone has to fix it. In most businesses, that someone is a team of people sitting between the supplier and the system, manually transposing, reformatting, and filling in gaps.

They are human middleware. And the cost of running them is almost entirely invisible, because it's just how it's always been done.

The numbers are striking when you actually surface them. On average, launching a single supplier SKU involves 10 to 15 separate interactions between supplier and buyer. Verifying the data takes two to five hours per SKU. From receiving the initial product introduction to going live can take up to a month.

That's per SKU. Across a catalog of tens of thousands of products, with hundreds of suppliers, it adds up to a genuinely massive operational drag. Most leadership teams have no idea it's happening at this scale because it lives in the day-to-day work of the data team rather than appearing as a line item anywhere.

Why the problem is getting worse

The structural issue has always been there. But three things are making it harder.

First, the volume of data required from suppliers keeps increasing. Compliance requirements are pushing businesses to collect more technical attributes than ever before. Digital product passports are on the horizon. Multichannel distribution means the same product needs to be described in more ways, for more contexts, at more levels of detail.

Second, many suppliers, particularly in B2B distribution and manufacturing, are not digitally mature. Small teams, data locked in PDFs, no structured product information management of their own. Expecting them to fill in a 200-attribute spreadsheet with 30 tabs and turn it around in two weeks is not a realistic process. It just generates emails, errors, and delays.

Third, the way data gets consumed has changed. It used to be that humans could fill in the gaps. A buyer could look at an incomplete product page and make a judgement call. Now the consumers of product data are increasingly machines. Search algorithms, recommendation engines, AI-powered agents making buying decisions on behalf of customers. Machines cannot infer what isn't there. If the data doesn't exist in structured form, the product doesn't get surfaced.

The problem also runs in both directions. Suppliers aren't just dealing with your template. They're dealing with 15 different templates from 15 different customers, all formatted differently, all with different mandatory fields. The frustration on the supplier side is real, and it directly affects how much effort they put into each submission.

The three levers

Most businesses try to solve this by buying a tool. The tool assumes the process is fine and that automation will smooth over the gaps. It rarely does. There's a more reliable sequence.

1. Structure

The goal with structure is not to force suppliers to reshape their data before they send it. It's to design an intake process that accepts data in the shape suppliers already have it, and then reshapes it on your side.

These are two different things: the onboarding structure and the presentational structure. Decoupling them means suppliers can submit their line card in a format that's natural for them, and the transformation into your taxonomy, attribute model, and channel requirements happens downstream. You stop blocking the flow of data at intake, and you stop creating a process that's painful enough for suppliers to deprioritise.

A 100-attribute mandatory submission form is not a structure. It's a bottleneck.

2. Usability

Once the structural foundation is in place, usability is about removing the back-and-forth. Every field a supplier is asked to provide should have a clear name, a clear definition, and clear guidance on how to fill it in. The format of submission should be flexible enough to meet suppliers where they are, whether that's a spreadsheet, a PDF upload, a form, or a bulk file.

The goal is to reduce friction without reducing quality. These are not the same thing. When suppliers understand exactly what's being asked and have an easy way to provide it, fill rates go up significantly. Moving a supplier from 20% fill rate to 80% is not unusual once the process is actually usable. The data gets better not because you've demanded more but because you've made it easier to give.

What usability is not: a supplier portal that requires six emails to figure out how to log in, with no guidance on what any of the fields mean. That's unfortunately closer to what most suppliers encounter.

3. Automation

Automation should come last, not first. This is where most businesses get the order wrong.

Automation in this context means things like template mapping, AI extraction from unstructured sources, auto-normalization of units and formats, and consistency checks. It's not about removing humans from the process. It's about removing the repetitive, low-value manual tasks that humans currently have to do on every submission.

The problem with jumping to automation before structure and usability are in place is that you end up automating a mess. The tools handle 20% of cases cleanly and leave the other 80% still requiring a person to reconcile and transform. Most supplier portals on the market today are, to put it plainly, glorified Excel templates inside a login screen. They don't automate anything meaningful. They just move the pain somewhere slightly different.

When automation is built on top of a clean structure with usable intake, the results are transformative. When it's bolted on top of a broken process, it's just a more expensive version of the same problem.

Where to start

This doesn't have to be a massive program of work. The way to approach it is to pick one lever and one supplier.

Run an MVP with a single supplier in a single category. Sit down with them and ask what's difficult about your current process. Suppliers have a lot of insight into where the friction is, and almost nobody asks them. Use that conversation to make one specific improvement, and then measure what changes.

There are a few places this tends to work well as a first move. Adding clear guidance notes to your existing submission template. Separating mandatory fields from optional ones and reducing the mandatory set to what you actually need to launch a product. Removing one step from the verification loop. None of these require a new tool.

The other option is to pick a category where you know the data gap is costing you and trace the problem back through the onboarding process. Where exactly does data quality break down? Is it that suppliers aren't sending it? That it's arriving in the wrong format? That the verification process is too manual? Each of those has a different fix.

The takeaway

Supplier data arrives in an unusable state because the process assumes it won't. Fixing that starts with structure, then usability, then automation, in that order.

Most businesses have this backwards. They invest in tools before they've defined what usable data actually looks like, and before they've made it genuinely easy for suppliers to provide it. The result is expensive, manual, and slow.

One supplier. One category. One lever. That's a practical place to start.

‍