The Most Expensive Part of Your Ecommerce Operation Is the Part Nobody Owns

Every ecommerce business has a version of the same meeting. Someone brings up product data. Everyone agrees it needs work. The conversation ends without a clear owner, a clear timeline, or a clear plan. Three months later, the same meeting happens again.

Meanwhile, the problem compounds quietly — in search rankings that don't move, in conversion rates that plateau, in ad spend that works harder than it should for results that feel just out of reach.

Product data and merchandising are the unglamorous foundation of ecommerce performance. They're also, in most businesses, the thing nobody fully owns — and that gap is more expensive than most operators realize.

The Ownership Problem

Ask who owns product data in most ecommerce businesses and you'll get a complicated answer.

There's usually a merchandising team, or a product information team, that touches it most frequently. But the data lives somewhere in an IT system, which means IT has a stake in how it's structured and stored. Marketing is constantly pushing for better titles, better descriptions, better attributes — because they're the ones who feel the downstream consequences most acutely. Operations has opinions. Data science has opinions. And somewhere above all of them, an executive knows it's a problem but can't quite prioritize it over everything else.

The result is a lot of pressure and not much movement. Everyone knows the product data needs work. Nobody has clear authority to fix it. So it doesn't get fixed, or it gets fixed in patches, inconsistently, by whoever has bandwidth that week.

This isn't a people problem. It's a structural one. Product data sits at the intersection of too many functions to belong cleanly to any of them, which means it ends up belonging to none of them in the way that actually matters.

What "Needs Work" Actually Looks Like

The specifics vary by business type, but the patterns are consistent.

Smaller ecommerce retailers tend to have the basics covered — product titles, descriptions, maybe some images — and not much else. What they're missing is structured attribute data: the specific, queryable details about a product that tell a search engine, a shopping platform, or an AI agent exactly what the product is. Size, material, compatibility, specifications, use case. The information a customer needs to make a confident purchase decision.

Most smaller operators either don't know attributes matter or haven't acted on them. They have titles and descriptions. They don't have structure.

Larger ecommerce businesses usually have stronger taxonomy — their category architecture is reasonably organized — but attribute coverage is often uneven. Some product categories are well-documented. Others have almost nothing. And in both cases, the attribute data that does exist is frequently inconsistent: different values for the same attribute across similar products, missing fields, outdated specifications.

Manufacturers and distributors moving into ecommerce are in the most challenging position. Their product data was built for a different purpose — inventory systems, B2B catalogs, internal databases — and it wasn't designed to perform in a consumer search environment. The whole thing often needs to be reworked from the ground up.

What Your Customers Are Telling You

You don't have to take my word for it that this is a problem. Your own data is telling you.

Look at your product reviews and voice-of-customer reports. Look at your support tickets and live chat transcripts. You'll find a consistent thread: customers who can't figure out the details about a product before they buy. They're asking questions that should be answered on the product page. They're abandoning carts because they're not confident they're buying the right thing. They're returning products because what arrived didn't match what they expected.

That's a product data problem presenting as a customer experience problem. The information exists somewhere — in a spec sheet, in a manufacturer's database, in someone's head — it just hasn't been structured and surfaced where the customer can find it.

The cost of that gap shows up in conversion rate, in return rate, in support volume, and in customer trust. It's diffuse enough that it's easy to attribute to other causes. But the thread usually leads back to the same place.

The Search Consequence Nobody Talks About

Beyond the customer experience impact, there's a search consequence that's less visible but arguably more significant.

Most ecommerce platforms generate product titles and page content dynamically from the underlying product data. If the attribute data is sparse, the titles are generic. If the titles are generic, organic search performance suffers — because Google can't determine relevance for specific queries when the page doesn't contain specific information. The same dynamic plays out in Google Shopping, where attribute-rich product feeds consistently outperform thin ones.

The connection between bad product data and poor organic performance is real and direct. It's just slow enough that most businesses don't draw the line between cause and effect. They see rankings that aren't moving and assume the problem is their SEO strategy. The actual problem is upstream, in the data that the SEO depends on.

Agentic Commerce Is Making This More Urgent

Here's the part of this conversation that most ecommerce businesses haven't had yet.

Search is changing in a way that makes product attribute data more important than it's ever been, not less. AI-powered search agents, the kind that are increasingly mediating how consumers find and evaluate products, don't read product descriptions the way a human does. They query structured data. They're looking for specific, machine-readable attributes that let them match a product to a user's precise need.

A product with rich, well-structured attribute data is findable by these systems. A product with a title and a paragraph description is largely invisible to them.

This isn't a future concern. It's a present one. And the businesses that have their attribute data in order are going to have a meaningful advantage as agentic commerce matures, not because they did something sophisticated, but because they did the unglamorous foundational work that their competitors skipped.

Why This Is a Marketing Problem, Not a Data Problem

The reason product data and merchandising stay in the ownership limbo they typically occupy is that they're perceived as operational or technical problems. Data management. System architecture. IT territory.

They're not. They're marketing problems with operational dependencies.

The downstream consequences of poor product data are almost entirely marketing consequences: lower search rankings, higher cost-per-click in paid search, worse conversion rates, higher return rates, weaker customer confidence. The people who feel those consequences most acutely are in marketing. The people with the most to gain from fixing the problem are in marketing.

Which means marketing needs to own the problem — not the technical implementation, but the business case, the prioritization, and the accountability for outcomes. Without that ownership, the meeting happens again next quarter and nothing changes.

What Actually Fixing It Looks Like

Attribute extraction and taxonomy work have historically been expensive and slow for one simple reason: they require a human to look at each product, understand what it is, determine what attributes it should have, and populate the data. At any meaningful catalog scale, that's a weeks-long project that most businesses can't staff or afford.

That constraint has changed. The same work that used to require a dedicated team and a significant project budget can now be done faster and more consistently using AI — at a scale that makes it viable even for large catalogs. Ninety thousand SKUs. Full attribute population from existing product information. Two hours.

That's not a small efficiency gain. It's a different category of possibility. For the first time, businesses that have been living with inadequate product data because fixing it felt impossible can actually fix it — and then build on that foundation.

Better attribute data means SEO can build programmatic pages targeting specific product queries. Paid search can target more precisely and waste less budget on irrelevant clicks. Merchandising can organize products into coherent, navigable structures. Every downstream function performs better.

The product data isn't just a problem to solve. It's a foundation. And until it's right, everything built on top of it is working harder than it should.

The Question Worth Asking

If you run an ecommerce business, the honest question is: do you know the current state of your product attribute data? Not in the abstract — specifically. What percentage of your products have complete attribute coverage? Where are the gaps? What's the downstream impact on your search visibility and conversion rate?

Most operators don't have clear answers to those questions. Not because they don't care, but because the problem has never been scoped in a way that made it actionable.

That's where the work starts. Not with a technology investment or a system overhaul — with an honest look at what you have, what you're missing, and what it's actually costing you.

The unglamorous work is usually the most important work. In ecommerce, product data is almost always both.

-

Skuset works with ecommerce businesses to audit, extract, and optimize product attribute data — at a scale and speed that makes the project viable for the first time. Start with a discovery call to understand what your current data is costing you.

< Older Post Newer Post >

Why Your Ecommerce Channels Don't Talk to Each Other (And What It's Costing You)

By Will Eifert • May 1, 2026

Most ecommerce businesses are running multiple marketing channels simultaneously. SEO. Paid search. Content. Merchandising. In many cases, each of those channels is being managed by a different team, a different agency, or both — with different tools, different reporting, and different definitions of success. The channels are working. They're just not working together. That distinction matters more than most operators realize. A marketing operation where the channels are integrated — where what you learn in paid search changes what you do in SEO, where your product data informs your content strategy, where your organic performance shapes your advertising decisions — performs measurably better than one where each channel is optimized in isolation. Not because any individual channel is worse, but because the whole is less than the sum of its parts when the parts don't connect. This is the integration argument. It's less dramatic than a lot of marketing advice, and it doesn't require new technology or a complete overhaul of how you work. It just requires that the people and data on one side of your operation can see and use what's happening on the other. How Siloed Channels Actually Happen The fragmentation isn't usually a deliberate choice. It's the accumulation of reasonable decisions made at different times for different reasons. You hired an outside paid search agency because your internal team didn't have the expertise. That agency does good work on your campaigns, but they operate largely in their own world — they get a brief, they optimize toward their metrics, they send a monthly report. They don't have visibility into your organic performance, your content calendar, or your merchandising priorities. And because they're external, the natural friction of communication means the data sharing that should happen usually doesn't. Meanwhile, your SEO might be handled internally — or by a different external vendor — with its own tools, its own reporting cadence, and its own definition of what success looks like. If you have an enterprise operation, you might have an internal SEO team and an external SEO vendor running in parallel, which creates its own kind of chaos: overlapping priorities, conflicting recommendations, and a lot of meetings that don't produce clear decisions. Even when everything is handled in-house, the integration problem doesn't automatically go away. Internal teams develop their own processes, their own dashboards, their own ways of working. The paid search team knows their data. The SEO team knows theirs. But the insight that lives at the intersection — the thing you can only see when you look at both simultaneously — gets missed, because nobody's job is to look at both. The Most Expensive Version of This Problem There's a version of the siloed channel conversation that gets oversimplified in marketing discussions: the idea that you should never spend paid search budget on terms where you're already ranking organically. The math seems obvious — why pay for clicks you're already getting for free? The reality is more nuanced. For businesses with significant paid search budgets, the paid/organic overlap question is often less urgent than it appears. The economics of blended search presence — owning both the paid and organic positions for high-intent terms — can justify the spend. This isn't where most of the money is being left on the table. The real cost of siloed channels is subtler and, in aggregate, larger: it's the opportunities that never get identified because nobody is looking at the full picture. Here's what that actually looks like in practice. Your paid search team is running campaigns. They have access to search term reports — the actual queries that triggered your ads, what people clicked on, what converted. That data is a window into exactly what your potential customers are searching for, in their own words, at the moment they're ready to buy. It's some of the most valuable keyword intelligence available to any ecommerce business. Your SEO team is building out programmatic pages — product category pages, comparison pages, location pages — designed to capture organic search traffic at scale. They're working from keyword research, from competitor analysis, from their best judgment about what to target. But if the paid search data isn't flowing to the SEO team, those programmatic pages are being built on incomplete information. They exist. They might even rank for something. But they're not targeted with the precision that the paid search data would enable — the actual search terms, the actual conversion patterns, the actual language your customers use when they're close to a purchase. Tighten that targeting using the paid search data, and the same pages perform meaningfully better. Not because the pages changed, but because the strategy behind them finally reflected what was actually happening in the market. That's the missed opportunity that siloed channels produce. Not dramatic waste — quiet underperformance, at scale, across every channel that's operating without the full picture. The Moment It Usually Clicks When I work through this with ecommerce operators, the integration argument tends to land when I can point to something specific in their own business — not a theoretical problem but a real one that's already affecting their results. The most common version goes something like this: they've been creating programmatic SEO pages, which is the right instinct. But the pages were built without the search term data from paid campaigns, so the targeting is loose. Some pages perform well. Others sit at the bottom of page two and never move. The assumption is usually that the SEO strategy needs work, or the pages need better content, or the domain needs more authority. Sometimes those things are true. But often the real issue is upstream — the pages are targeting the wrong version of the right keyword, because the people who built them didn't have access to the data that would have told them exactly which version converts. When you show an operator that data and demonstrate how it would change what the pages target, the connection is immediate. The siloed channel problem stops being abstract. It becomes a specific thing that's happening in their business right now, with a specific fix that's within reach. SEO Informs Paid. Paid Informs SEO. Neither Informs Content Without Help. The integration argument isn't just about paid search and SEO. It runs through every channel. Strong organic rankings for a product category tell you something about where your authority is concentrated — which should influence where you focus paid search budget, because blended presence on high-authority terms is often worth more than isolated paid coverage on terms where you have no organic footprint. Paid search performance data reveals which product attributes matter most to customers at the point of purchase — information that should be feeding back into your merchandising team's decisions about how products are described and categorized. Your content strategy should reflect what your customers are actually searching for — which you can see most clearly in a combination of organic search data and paid search query reports, not in keyword research tools alone. Merchandising decisions about how products are organized and attributed affect both your organic search performance and the quality of your product feed in Google Shopping — which means the merchandising team's work is marketing work, whether it's thought of that way or not. None of these connections are complicated. They're just rarely made, because the people and data that need to be in conversation with each other are operating in separate rooms. What Integration Actually Requires The good news is that channel integration doesn't require a technology overhaul or a reorganization. It requires visibility and conversation — the right data in front of the right people at the right time, with someone responsible for seeing the connections and acting on them. That's harder than it sounds when the channels are managed by different external vendors with different reporting formats and different incentives. An agency optimizing your paid search campaigns has no particular reason to surface insights that are useful to your SEO team — especially if that team is a competitor's client or a separate internal function. The incentive structure doesn't support integration even when the data would enable it. It's also harder when the channels are internal but organized around functional silos — teams that are measured on their own metrics, reporting to different managers, working in different tools. The integration can happen, but it requires deliberate effort and someone with the authority and the context to pull the threads together. The businesses that get this right tend to have one of two things: either a genuinely integrated internal operation where data flows between functions by design, or a single external partner with visibility across all the channels simultaneously. The second is rarer than it should be — most agency relationships are channel-specific by design, which means the integration problem gets built into the engagement model before the work even starts. The Underperformance Is Quiet The thing about siloed ecommerce channels is that the cost doesn't show up as a line item. It shows up as performance that's good but not great — channels that are doing their jobs individually without producing the compounding effect that integration enables. The paid search team hits their ROAS target. The SEO team shows ranking improvements. The content team publishes on schedule. Everything looks fine in every individual report. What doesn't show up in any of those reports is the organic traffic that a better-targeted programmatic page would have captured. The conversion rate improvement that a more attribute-rich product page would have driven. The content that would have ranked and converted if it had been built around actual search term data instead of keyword research alone. That's the cost of siloed channels. Not dramatic. Not visible in any single dashboard. But real, and cumulative, and larger than most operators would guess if you added it up. The channels are working. They could be working together. That gap — between good individual performance and integrated performance — is where the real opportunity lives. - Skuset runs ecommerce marketing as a single integrated operation — paid search, SEO, merchandising, content, and product data working from the same data and toward the same outcomes. Start with a discovery call to see what that looks like for your business.

What "We Use AI" Actually Means at Most Agencies (And What It Should Mean)

By Will Eifert • May 1, 2026

If you've talked to a marketing agency in the last eighteen months, you've heard it. Somewhere in the pitch, usually after the case studies and before the pricing, someone says it: "We use AI." You nod. They move on. And you're left with the same question you had before they said it: what does that actually mean for me? The honest answer is that it depends entirely on which agency you're talking to — and most of them are still figuring it out. What Most Agencies Mean When They Say It The majority of agencies using AI right now are using it in ways that are genuinely useful but fundamentally limited. Content drafts. Code generation. Pulling together summaries from analytics data. Running queries against a dataset to surface insights faster than a human could manually. None of that is bad. It's just small. The thinking behind it is small too. An agency that uses AI to generate a first draft of a blog post or pull together a monthly performance summary has made their existing workflow a little faster. They've reduced the time it takes to do things they were already doing. That's an efficiency gain — but it's an efficiency gain they're almost certainly keeping for themselves, not passing to you in the form of better work or lower cost. What you're getting in that scenario is the same agency model, slightly accelerated. A junior person is still doing the execution. A senior person is still checking in monthly. The work is still siloed between the team that handles your SEO and the team that handles your ads. The AI made things faster at the margins. The fundamental structure didn't change. The Question Nobody Is Asking The right question isn't "do you use AI?" Every agency is going to say yes to that question for the next decade regardless of what they actually do with it. The right question is: how has AI changed the way you think about the work — not just the speed at which you do it? There's a meaningful difference between using AI as a tool you pick up when it's convenient and building your operation around what AI actually makes possible. Most agencies are doing the former. The latter requires a different kind of thinking — and a willingness to rebuild how you work rather than just add a new tool to an existing process. What Deliberate AI Architecture Actually Looks Like Here's the distinction that matters in practice. Popping a question into an AI chat and taking the answer at face value is not AI integration. It's asking a very capable tool a decontextualized question and hoping the output is good enough. Sometimes it is. Often it isn't. And the person doing it usually can't tell the difference, because they haven't given the AI enough context to work with and they don't have enough expertise to evaluate what comes back. What actually works is almost the opposite of that. Before you ask AI to do anything, you have to establish the rules — your judgment, your tendencies, your standards, the things you know from experience that a model trained on the entire internet wouldn't automatically know. You encode your expertise into the system before the system does anything. At Skuset, that looks like fifteen years of ecommerce marketing judgment — what moves revenue, what's performance theater, how to evaluate a keyword strategy, what a well-structured campaign actually looks like — codified into a framework that sits at the center of every engagement. The AI doesn't operate on its own instincts. It operates within mine. The difference in output between those two approaches is significant. An AI operating with limited context and no established framework produces generic, average work. An AI operating within a well-constructed framework built around specific expertise produces something much closer to what a senior practitioner would produce — and it produces it at a scale and speed that a senior practitioner alone couldn't match. That's the distinction between incidental AI use and deliberate AI architecture. One makes things faster. The other makes things fundamentally different. Why This Matters for Your Business If you're an ecommerce business evaluating marketing partners right now, the AI question is worth taking seriously — not because AI is magic, and not because you should be impressed every time someone claims to use it, but because the gap between agencies that have thought carefully about AI integration and agencies that haven't is growing fast. The agencies that are thinking carefully about it are able to do things that weren't previously possible at reasonable cost. A full technical SEO audit in twenty minutes. A complete paid search campaign buildout in thirty. Ninety thousand product attribute extractions in two hours. These aren't claims about AI being powerful in the abstract — they're specific capabilities that translate directly into what a client gets for their investment. The agencies that aren't thinking carefully about it are charging you 2023 rates for work that now takes a fraction of the time it used to. They may be using AI, but they're using it the way someone uses a calculator — to do faster what they were already doing, without reconsidering what they should be doing in the first place. The Noise and the Signal The honest reality for most business owners is that AI has become a kind of background noise. Everyone uses it, everyone claims it, and after a while the claims stop meaning anything. That's understandable. The hype has been relentless. But underneath the noise there's a genuine signal worth paying attention to: AI hasn't replaced expertise, but it has dramatically multiplied what expertise can produce. The question is whether the people you're working with have done the work to actually capture that multiplier — for themselves and, more importantly, for you. When you hear "we use AI" from your next agency pitch, the follow-up question is simple: show me what that actually means for the work you do for my business. If the answer is "we generate content faster" or "our reporting is more automated," that's fine as far as it goes. But it doesn't go very far. If the answer is a specific, concrete explanation of how AI is integrated into their operating model — what it changes about the scope of work they can deliver, the speed at which they can deliver it, and what that means for your investment — that's worth paying attention to. That second conversation is rarer than it should be. But it's the one that actually matters. - Skuset is a full-stack ecommerce marketing operation built on fifteen years of practitioner judgment and deliberate AI integration. If you want to understand what that looks like in practice, start with a discovery call .