Fix Product Catalog Data Quality Across Retail Channels
Poor product catalog data leads to lost sales, customer frustration, and operational chaos across retail channels. This article presents four proven strategies to maintain clean, accurate product information, backed by insights from industry experts who have solved these challenges at scale. These practical approaches help retailers eliminate data conflicts and ensure consistency whether customers shop online, in-store, or through mobile apps.
Establish a Golden Record in PIM
Look, if you're stuck in a loop of fixing product data errors after they happen, you're already behind. You've got to stop treating data cleanup like it's a constant game of whack-a-mole. You need to shift from being reactive to being an upstream gatekeeper.
The single biggest mistake I see? Letting ecommerce or store teams make manual overrides in their own systems. That's how you end up with fragmented silos. Once that data starts diverging, you're just inviting customer friction. It's a mess.
The fix is a 'Golden Record' hierarchy, usually inside a PIM or MDM system. You have to strip the power away from those channel-specific departments. Centralize it. Create a dedicated data stewardship team-they're your final arbiters.
And for the suppliers? Stop letting them dump raw, messy spreadsheets on you. Force them to map their data to your schema. If a new entry conflicts with your master record-say, for a material or a size-the system should automatically kill it. It shouldn't get anywhere near your storefront until it's reconciled.
The real game-changer is supplier-side accountability. If you make them validate their data against your master schema through a structured portal, you stop the bad data from entering your ecosystem in the first place. It's simple, really. You turn your internal team's job from manual cleanup into exception management. You define the truth, you verify it, and you do it before a single customer ever sees it.
Cultivate Trust With Primary Suppliers
Generally speaking, we don't really have issues with incorrect delivery notes coming from manufacturers other than China. Unfortunately, with small orders from China (e.g., 20-30 items), an item is often missing, and I suspect this is intentional to avoid challenges. We have very good relationships with our main suppliers, who are large UK suppliers, and it works both ways. Occasionally we receive extras and we'll alert them. Other times, if items are missing, we'll alert them, and it works on an honesty factor where both parties have traded for a long time. We know what works and we know we can both be honest with each other. We do occasionally take photographs if an item arrives in the box badly damaged. We may possibly open it under camera just in case there is a discrepancy, but such an event is very, very rare now, and generally we work well with our UK suppliers.
Designate One Decider for Conflicts
The data conflicts we were dealing with weren't exotic. Same product listed as "navy" from one supplier and "midnight blue" from another. Sizes that mapped differently depending on which spreadsheet you opened. Materials were described in ways that were technically accurate but inconsistent enough to cause real customer service problems. We tried solving it with better tooling first, which was the wrong instinct. The problem wasn't that we lacked a system for storing clean data. It was that nobody owned the decision of which version was correct when versions conflicted.
The change that actually reduced mismatches wasn't a software implementation. We designated one person on the merchandising team as the final arbiter for any conflicting product attributes. Not a committee, not a process requiring sign-off from multiple departments. One person with the authority to make a call and move on. It felt like an oversimplification when we proposed it, and some people pushed back because it seemed to concentrate too much decision-making in one role.
In practice, it worked because the bottleneck had never been the decision itself. It was the ambiguity about whose decision it was. Once that was clear, conflicts got resolved in hours rather than sitting in someone's inbox for two weeks. The remaining problem is supplier onboarding. New suppliers still come in with their own naming conventions, and the cleanup still happens manually. We've automated parts of it, but not in a way I'd call solved.
Assign Attribute Owners and Block Shopify Edits
We are a Shopify-only fragrance retailer, no physical stores, but supplier-data conflict is just as ugly when the storefront is your entire surface area as it is when you are syncing to retail.
I run PerfumeM (perfumem.com), an independent fragrance retailer since 2017. We carry product from over a hundred fragrance houses, each of which sends its own data feed, its own product imagery, and its own spelling conventions. Same scent, three different names. Same bottle, two different milliliter sizes. Same year of release, conflicting dates.
The process change that fixed the worst of it for us was simple to describe and unglamorous to enforce.
One owner per attribute. A single person on the team owns product name spellings, another owns size and SKU, another owns imagery, another owns brand metadata. Suppliers do not write directly to our Shopify catalog. They write to a master internal spreadsheet, and the attribute owner is the only one who can resolve conflicts before the row gets pushed live. That single rule eliminated about ninety percent of the duplicate-listing problem we had in 2020.
Master spreadsheet is the source of truth, Shopify is downstream. We push from the spreadsheet to Shopify daily via a small import job. Shopify is never edited directly. If a customer service person spots a wrong spec on a product page, they file it against the spreadsheet, not against Shopify, and the next push corrects it. That single-direction flow is what made the policy stick.
For imagery specifically. We treat the manufacturer reference shot as the visual source of truth and document every secondary image as a derivative of it. When suppliers send conflicting product photography we always default to the manufacturer set, which makes the conflict-resolution call obvious instead of a daily judgment call.
The ownership model is the lever that actually mattered. Tooling change without ownership change just gives you cleaner conflicts. Ownership change without tooling is also fine. Tooling plus ownership is how you stop relitigating the same data fights every quarter.
Ahmad Khan, founder of PerfumeM (perfumem.com)
