Catalogs
in the Wild.

Something quiet has happened in the way modern businesses sell. Not all at once, and not anywhere it could be easily seen. But sometime in the last few years, the product catalog stopped being the thing that supported every revenue channel and became the thing every revenue channel runs on.

It happens through every order placed by a B2B buyer who searched a product number. Through every marketplace listing approved or rejected. Through every partner integration that did or did not pass syndication. Through every AI agent making a purchase decision based on what it could see in a structured attribute. Through every retail-media impression that earned its share of voice. The catalog is the substrate. Everything sits on it.

For finance leaders, that makes the catalog a P&L lever. For commercial leaders, the leverage point for every channel they sell through. For the people who run product information, an asset they have been asked to scale faster than any team could keep up with.

And the demands keep growing. A modern catalog now feeds twenty or more endpoints at once. Each one wants the same truth in a different shape. Each one breaks differently when the data drifts. The teams running these catalogs were not given new tools to match the new reality. They were given the same PIM, the same spreadsheets, and a quiet mandate to keep up.

This report is about the companies that have stopped trying to keep up the old way, and started running their catalogs as something closer to infrastructure. Names removed. Patterns preserved. The work, finally, made visible.

Catalog quality is not an operational task.
It is a P&L lever.

Catalog drift is silent. It doesn’t trigger alerts. It doesn’t appear on a dashboard. It compounds across millions of records, across dozens of channels, across quarters, until a buyer can’t find a product, a marketplace rejects a feed, or a partner publishing job fails for the third time this month. By then, the cost has been paid for months. Sometimes for years.

Two percent drift on a 4-million-SKU catalog
is 80,000 records bleeding revenue
every day no one is watching.

The product data stack most companies still run was designed for a much simpler world. A PIM held the master record. A DAM held the assets. A handful of channels — a website, a few major marketplaces, a partner network — pulled from those systems on a weekly schedule. Distribution was linear. Standards held. The systems worked, mostly because the demands on them were modest.

That world is gone. A modern catalog now has to live in twenty or more endpoints simultaneously. ERPs and PIMs. Dozens of distributor and retailer partners. Marketplaces with their own format requirements. Retail media networks. AI search. Agentic discovery layers. Channels that didn’t exist eighteen months ago and probably won’t exist in their current form eighteen months from now. Each one demands a different shape of the same data. Each one fails differently when the data drifts.

Legacy product data tools weren’t built for this kind of distribution. They were built to be a source of truth, and they remain genuinely valuable in that role. The problem is the work that has to happen on top of them. Translating a single source of truth into the dozens of shapes modern commerce demands. Keeping all of those shapes current as feeds change. Catching the drift before it surfaces somewhere expensive.

Today, that work happens downstream. Manually. In spreadsheets, one-off integrations, and services engagements that take quarters to deliver and break the moment a vendor changes a feed format. The gap between the source of truth and the channels that depend on it is where catalog drift accumulates. It is also, increasingly, where the operating cost of running a modern catalog actually lives.

The systems built to manage product data
were never designed to distribute it.

EKOM is not a faster PIM. It is not a better DAM. It is not a syndication platform with AI bolted on. It is a different way of thinking about product data altogether, built for how commerce actually operates now. The contrast below is what most teams describe when they tell us about the work they used to do, and what happens when they stop doing it that way.

EKOM is infrastructure, not a project.
Continuous, not a quarterly cleanup.

For finance leaders, this collapses catalog operations from a recurring services line into a continuous platform. For commercial leaders, it means every channel runs on consistently clean data without quarterly cleanup cycles. For product information teams, it means the catalog stops being a permanent emergency and starts behaving like the asset it was always supposed to be.

EKOM sits between every source of product data your business has and every channel that sells. It applies the standards you have already encoded (taxonomy, brand voice, channel rules) across three layers, continuously, at machine speed. The work the platform does on the next layer happens automatically once the previous one is in place. This report focuses on the first layer.

Normalize.

The first layer, and the one this report shows. EKOM ingests from any source — PIMs, DAMs, ERPs, supplier feeds, manufacturer files, scraped data, partner APIs. It profiles every column, classifies what belongs and what doesn’t, maps non-obvious fields to vertical taxonomy, and recommends fixes within your existing schema. Approved corrections get applied. Drift gets surfaced. Standards hold.

Enrich.

Once a catalog is normalized, the gaps become obvious. EKOM’s enrichment layer fills them using the customer’s own brand voice, category rules, and channel-specific requirements. Titles, descriptions, attributes, structured content. Every output operates under governance. Encoded standards in, approved content out.

Distribute.

A normalized and enriched catalog still has to live everywhere the business sells. EKOM delivers clean, format-ready data to every endpoint that matters — commerce platforms, ERPs, PIMs, retailer partners, marketplaces, retail media, AI search — in the exact shape each one requires, on schedule, by API, continuously.

What follows is a cross-section of findings from four production catalog analyses. Names removed. Patterns preserved. A few things to know before you read them.

These four are the focus. The platform serves many more.

These four show how different catalogs really are — different schemas, different vocabularies, different regulatory pressures. EKOM operates across many more: apparel, beauty, building products, food and beverage, sporting goods, medical supply, hardware.

The samples are deliberately small. The catalogs are not.

The findings represent a manageable slice of a much larger analysis. The catalogs run from tens of thousands to several million SKUs. The patterns compound proportionally at scale.

Each sample is paired with the work behind it.

Every vertical takes two pages. A findings page that shows what the platform identified and the action it took. A methodology page that shows how. The findings pages are where the consequences land. The methodology pages are where the analytical depth becomes visible.

In automotive, fitment data locked in ID strings. In foodservice, workflow artifacts hiding as product attributes. In industrial, nine fuse SKUs with amperage ratings three other fields contradicted. In furniture, 203 columns concealing eight standard concepts.

What’s printed here is illustrative.
The catalogs themselves are not.

Before any action is taken, EKOM profiles the catalog. The platform combines large-context-window AI with vertical-specific heuristics to evaluate the data at three levels: column, row, and cross-record. Unlike rule-based tools that check one field against a schema, EKOM reads the relationship between fields and resolves conflicts — surfacing errors that look valid in isolation.

Customer in production: a North American automotive aftermarket distributor focused on turbochargers, engine internals, and powersports.

Cross-section of findings. Customer in production. Sanitized for publication.

Automotive aftermarket catalogs live or die by their fitment data. Year, make, model, engine. Search depends on it. Partner publishing depends on it. AI-driven purchasing won’t even see a product without it. When that data is locked inside opaque vehicle ID lists, the whole catalog becomes unfilterable.

Customer in production: a foodservice disposables manufacturer supplying restaurants, bakeries, and hospitality.

Cross-section of findings. Customer in production. Sanitized for publication.

Catalogs accumulate columns the way old houses accumulate cabinets. Some started as legitimate product attributes. Others started as internal workflow flags that drifted into the schema and stayed there, with professional-sounding names and just enough data to look real.

Column coverage sample — how EKOM classified them

Customer in production: an industrial electrical distributor supplying contractors, OEMs, and MRO customers.

Cross-section of findings. Customer in production. Sanitized for publication.

Industrial electrical catalogs depend on precise handling of units of measure and class designations. A breaker rated 100 amps and one rated 100 A are the same product. But if filters and partner publishing systems treat them as different values, the catalog fragments and search returns half of what it should.

Input variants from real records — normalized output

Customer in production: a consumer furniture and home retailer publishing across multiple commerce channels.

Cross-section of findings. Customer in production. Sanitized for publication.

Customer catalogs grow organically. Column names get coined by whoever added them in the moment they were needed, often years apart, often by people no longer at the company. The result is a schema that contains the right data under the wrong names.

Customer column names resolved to standard concepts

The product information leaders we work with describe the same daily reality before EKOM. Every record reviewed manually. Every change tracked in a spreadsheet. Every quarter, another massive cleanup project that fixes what should have been fixed continuously. Every team drowning in catalog work that has no clear end state. The catalog is never done.

With EKOM, the model inverts. The platform handles the volume. Normalization, classification, continuous standards application — all happening automatically within encoded governance. What surfaces to the operator gets tranched intentionally. Approve a pattern in one click. Approve a category. Approve a batch. The platform is intuitive about what actually needs human judgment, and never asks anyone to read every flag.

The product information leader can finally report with confidence: the data powering every commercial channel is working for the business, not against it. Standards encoded once apply everywhere. Drift gets surfaced as it appears, not discovered three quarters too late. And EKOM fits the existing workflow rather than replacing it.

Scale is not the problem.
10,000 products or 10 million,
the platform performs the same.

The scale that breaks manual catalog operations is exactly where EKOM becomes most leveraged. A 50,000-SKU catalog and a 5,000,000-SKU catalog run on the same platform, with the same governance, with the same continuous quality. Headcount doesn’t have to scale with SKU count. Operating cost doesn’t balloon with channel count. The leverage compounds the larger the business gets.

The findings on the previous pages come from EKOM’s normalization layer. They show the platform reading a catalog, recognizing what’s there, recommending the right corrections, and applying them within the customer’s schema. That work is foundational. It’s also one third of what the platform does. The two layers that follow are where the work translates into commercial advantage.

Enrichment.

A normalized catalog reveals exactly what’s missing. EKOM’s enrichment layer fills those gaps using the customer’s own brand voice, category rules, and channel-specific requirements. Titles. Descriptions. Attributes. Structured content for every record that needs it. Enrichment runs under the same governance the rest of the platform does. Encoded standards in, approved content out. The catalog ends up ready to compete in search, win on marketplaces, and pass partner publishing requirements on the first try.

Distribution.

A catalog that’s been normalized and enriched still has to live everywhere the business sells. EKOM’s distribution layer publishes to every endpoint that matters — commerce platforms, ERPs, PIMs, distributor and retailer partners, marketplaces, retail media, AI search — in the exact shape each one requires, on schedule, by API, continuously.

One retailer in this report had been failing partner publishing requirements for multiple quarters. After normalization, the same catalog passed on the first submission. The feed hadn’t changed. The catalog had.

The business case.

Together, the three layers collapse a workflow that traditionally spans a PIM, a DAM, a syndication network, multiple integration projects, and ongoing services engagements into a single continuous operation. Faster time-to-channel. Fewer partner publishing failures. Better channel readiness across search and marketplaces. Lower compliance risk. A catalog that doesn’t drift.

The catalog is the asset.
EKOM is what runs it.

A summary of findings across four production catalog analyses. Each sample is a cross-section of a larger catalog. The patterns shown at sample scale compound proportionally across the full data set. All customer names removed. Patterns preserved.

The four catalogs in this report are not outliers. The patterns — miscategorizations, safety spec conflicts, regulatory exposure, workflow artifacts leaking into product schema, fitment data locked in opaque strings — are the operating reality of running a modern catalog at scale. They appear in every vertical EKOM serves, regardless of how mature the team or how disciplined the process.

The findings above are from samples. The catalogs behind them run from tens of thousands to several million SKUs. The scale of the underlying problem is proportional. The platform that found these issues in a sample will find them everywhere else in the data — and fix them continuously, without adding headcount, without quarterly cleanups, without waiting for a marketplace to reject a feed before anyone notices.

The catalogs profiled in this report are not edge cases. The patterns we showed — miscategorizations, attribute drift, schema fragmentation, vertical-typical gaps, workflow artifacts inside product fields, regulatory exposure, contradictory data across columns — are the operating reality of running a modern catalog at scale. They appear in every vertical EKOM serves, regardless of how mature the team or how disciplined the process.

If your catalog reflects what you’ve seen on the previous pages, or you suspect it might, EKOM runs scoped analyses for prospective customers. The work takes hours, not weeks. The output is a sanitized version of exactly what’s printed here, applied to your data, sized to your scale, focused on the catalog you actually operate. No commitment. No services engagement.

We’re happy to walk through findings with finance, commercial, and product information leaders together. The catalog touches all three, and most of the conversations we have go better when all three are in the room.