Insights

e-commerce

You can’t scale chaos. Why clean product data matters more than your ad budget right now

Oct 20, 2026 · 9 min read · Human

[post image]

Online stores grow for years without a plan. Then every new channel multiplies the work instead of the sales. What product data chaos looks like, what it costs and where to start cleaning it up.

Most of the stores that come to us have the same problem, even though each describes it differently. “Ads keep getting more expensive.” “We can’t keep up with adding products.” “Merchant Center rejects half of our feed.” “Our descriptions are copied from the manufacturer.” “Nobody knows how many variants of the same dog bed we have.” These aren’t four problems. They’re one: the catalog grew faster than anyone could keep it in order.

The thing about chaos is that it doesn’t hurt until you have to multiply it. With one store and one channel, you can live with inconsistent names and missing attributes. But in 2026, one store isn’t enough. Products need to be in Google Merchant Center, in the Facebook and Instagram catalog, in TikTok Shop and on marketplaces, and more and more often they need to make sense to AI assistants that compare offers on the customer’s behalf. Every one of those places reads the same data. If the data is bad, the error repeats everywhere, and the cost of fixing it grows with every channel.

What chaos looks like once you put numbers on it

We ran this exercise with several clients, and the results were similar. We counted how many hours a month the team spends on work it shouldn’t be doing by hand: fixing rejected products in feeds, rewriting descriptions for yet another channel, adding missing EAN codes, hunting for photos in someone’s inbox, answering customer questions about sizing because the size isn’t on the product page. For a store with a few thousand SKUs, it came to one or two full-time jobs. Not jobs that grow the business. Jobs that put out fires.

The other side of the equation is less visible but more expensive. A product without attributes doesn’t show up when customers filter. A product with the wrong EAN never makes it into Google Shopping. A product with the manufacturer’s description doesn’t stand out in search, so the campaign has to pay extra to push it. A product without proper structured data doesn’t exist for an AI model putting together an answer to “what dog bed works in a small apartment.” Every one of those gaps is a sale that didn’t happen, and nobody will ever see it in a report.

Why your ad budget won’t fix your catalog

The most common reaction to falling sales is simple: put more into Google Ads or Meta. It works briefly and expensively. Ads send traffic to a product page that doesn’t convince, so conversion drops and acquisition costs rise. Ad algorithms learn from feed data, so they get the same thing the customer gets: an incomplete picture. From the outside it looks like the market got tougher. Up close, you can see you’re simply pouring money into a leaky bucket.

That’s why the order in our method is non-negotiable: order first, then scale, then the future. Not because order is noble, but because every zloty you spend after that works harder.

Where stores that do this well start cleaning up

  1. One source of truth for every product. Name, description, features, variants, photos, EAN and channel-specific versions in one place, not scattered across the store, a spreadsheet, an inbox and one person’s head.
  2. Rules instead of manual work. A naming pattern, a tone for descriptions, a photo order and required attributes defined once and applied to the whole catalog. This is a task AI is genuinely good at, as long as a human writes the rules.
  3. Duplicates and variants. Merging the same products from several stores into one master product, fixing option groups, cleaning up sizes and colors. Boring, but this is where most of the manual work disappears.
  4. Quality control before publishing. A product doesn’t go out to a channel unless it meets that channel’s requirements. Better to stop it in the system than get a rejection from Merchant Center three days later.
  5. Measurement. How many products have complete data, how many were rejected, how many hours catalog management took this month. Without those numbers, “order” is an opinion, not a fact.

Small stores need this too

There’s a lingering belief that product data management is only for stores with tens of thousands of SKUs. Our experience says otherwise. With fifty products, order gives you consistency and saves time for an owner who’s doing everything alone anyway. With five thousand, it’s the only way to avoid hiring another person to fix spreadsheets. The difference is scale, not principle.

There’s one more reason order has become more urgent than it was a year or two ago. Customers increasingly ask AI assistants about products, and assistants don’t read product pages the way people do. They read data. A store with a chaotic catalog is invisible to them or, worse, visible with the wrong information. We cover this in a separate article on GEO. Here, one sentence is enough: you can’t scale chaos, and in 2026 you can’t hide it either.

Running an online store?Let’s start with a 30-minute conversation about your catalog, campaigns and margin.Explore AI eCommerce

Contact

Want to earn your audience’s trust?