What Actually Makes a Product Successful?
Walk into most ecommerce admin panels and you'll find a "Best Sellers" report sorted by units sold or revenue, usually within the last thirty days. Merchandising teams use it to decide what goes on the homepage. Marketing teams use it to decide what to feature in the next email. Nobody questions it, because the report feels objective. The numbers are real. The product sold, or it didn't.
That objectivity is exactly the problem. A single ranking metric, however accurate it is at measuring the thing it measures, is a poor foundation for a decision as consequential as which products get the store's most valuable real estate. This article makes the case that a product success score worth trusting has to look at more than one number, and walks through why each of the common shortcuts — revenue, views, discounts, margin, staff intuition — breaks down under scrutiny.
The Comfortable Default: Sort by Sales
Sorting by sales is the default for a good reason: it's easy to compute, hard to argue with in a meeting, and grounded in something that actually happened. But "the product that sold the most" answers a narrower question than most people assume. It tells you what happened last month. It says very little about whether that product deserves the store's best placement going forward, or whether it sold well for reasons that have nothing to do with genuine, durable demand.
Consider two products with identical trailing-30-day revenue.
| Metric | Product A | Product B |
|---|---|---|
| Revenue (30 days) | $18,400 | $18,400 |
| Discount applied | 70% off | No discount |
| Units sold | 460 | 92 |
| Gross margin | 4% | 52% |
| Repeat purchase rate | 3% | 27% |
By a revenue-sorted best-seller list, these two products are tied. By almost any other measure that matters to the business, they are not remotely comparable. Product A is manufacturing revenue by giving away margin. Product B is quietly building a base of customers who come back. A ranking system that can't tell these two products apart isn't measuring success — it's measuring transaction volume, which is a different and much shallower thing.
Is a Product Successful Because It Sells the Most?
High sales volume is evidence of something, but it's rarely evidence of only one thing. A product can sell in large volume because customers genuinely want it. It can also sell in large volume because it was placed at the top of a category page, because a competitor briefly went out of stock, because a discount code made it artificially cheap, or because it happens to be the default option shown to every visitor regardless of fit. Volume describes an outcome. It doesn't explain the cause, and the cause is what determines whether that volume will continue.
Or Because It Receives the Most Traffic?
Traffic has the same problem in a different direction. A product can accumulate enormous view counts purely through position — first item in a grid, top of a search result, linked from a high-traffic blog post — independent of whether visitors who see it actually want it. High traffic with weak conversion isn't a success signal. It's often the opposite: evidence that a lot of attention is being spent on something that isn't converting it into revenue, while a better-fitting product further down the page never gets the exposure to prove itself.
Product performance has to account for what happens after the view, not just the view itself.
What If It Only Sells Because It's Discounted 70%?
This is one of the more uncomfortable questions in merchandising, because discounted products genuinely do move volume — that's the entire point of a discount. The question worth asking isn't whether the discount worked. It's what happens to demand the moment the discount is removed. A product that only sells at 70% off is telling you its price-to-perceived-value ratio is off by a wide margin at full price. Treating its discounted sales figures as a general success signal, and featuring it accordingly, tends to produce an unpleasant surprise the next time the price returns to normal and sales collapse.
A product propped up entirely by discount depth isn't demonstrating demand. It's demonstrating price sensitivity, which is a different and much less durable thing.
What If It Has Almost No Inventory Left?
Here's a scenario every merchandiser eventually runs into: a genuinely strong-selling product is down to its last dozen units, with the next restock weeks away. It's tempting to keep it front and center on the homepage — it's earned the spot, historically. But featuring a product that's about to sell out converts high-intent traffic into frustration. A customer clicks through excited, finds their size gone, and that disappointment gets associated with the brand, not just the product.
Inventory optimization and merchandising placement are rarely thought of as the same discipline, but they need to be. A ranking that ignores stock position is optimizing for clicks that a meaningful share of the time won't convert into a completed order.
What If Demand Exists But Inventory Can't Support It?
This is a distinct and arguably more damaging version of the same problem. It's one thing for a product to sell out occasionally. It's another for a product to sell out repeatedly, every restock cycle, because demand consistently outpaces supply. Continuing to promote it heavily under those conditions creates a pattern of avoidable stockouts — a self-inflicted wound where the merchandising system keeps directing traffic toward disappointment, cycle after cycle, instead of adjusting.
Can Popularity Become Harmful?
It's worth sitting with this directly: yes. A product that is popular enough to chronically outstrip supply isn't purely a merchandising win. Every stockout it causes is a customer who arrived ready to buy and left without completing the purchase — sometimes to a competitor, sometimes just gone. A merchandising strategy that keeps pushing traffic toward a product with a track record of unavailability is, in a very literal sense, spending marketing effort to generate frustration.
What If Another Product Has Lower Sales but Higher Future Potential?
Historical sales data, by definition, describes the past. A product launched three weeks ago with modest sales but unusually strong early engagement — high add-to-cart rate, strong repeat visits, positive early reviews — may be a better bet for future performance than a mature product whose sales have quietly been declining for two months even though its trailing totals still look respectable. A ranking system anchored entirely on historical volume will always favor the second product over the first, right up until the first one becomes obviously undeniable, by which point the head start has already been lost.
This is the central limitation of any purely backward-looking metric: it's accurate about what happened and silent about what's about to happen. Product ranking that only rewards history will always be structurally late to reward trajectory.
Should Homepage Ranking Only Reward Historical Performance?
If the honest answer to the previous section is no, then homepage and category ranking logic that relies purely on trailing sales or trailing revenue has a structural blind spot built into it. It will reliably surface yesterday's winners. It will just as reliably miss tomorrow's, because tomorrow's winners, almost by definition, don't yet have the sales history to compete with an established product on a metric built entirely from history.
Is Conversion Rate Enough?
Conversion rate solves some of the problems above but introduces its own. A product converting at 8% sounds excellent — until you notice it only had twenty visits last month. A conversion rate calculated from a small sample is closer to noise than signal, and treating it with the same confidence as a rate calculated from ten thousand visits will systematically overrate obscure, low-traffic products that got lucky and underrate consistent performers whose rate is slightly lower but backed by real volume.
Conversion rate also says nothing about order value, margin, or what happens after the sale. A product converting extremely well at a heavily discounted price isn't necessarily healthier than one converting moderately well at full price.
Is Revenue Enough? Is Profit Enough? Is Velocity Enough?
Each of these is a legitimate and useful number. None of them, alone, is sufficient.
- Revenue ignores the cost of getting there — discounting, returns, and margin erosion can sit invisibly underneath a healthy top-line number.
- Profit ignores growth trajectory and can quietly favor mature, slow-moving products over younger ones still building momentum.
- Velocity — how fast a product sells relative to its stock — ignores whether that speed is sustainable or whether it's simply draining a small batch of inventory that took months to plan and can't be replenished quickly.
Every one of these metrics is a legitimate lens. None of them, viewed alone, is the full picture, and a ranking system built around any single one will reliably optimize for that one dimension at the expense of the others.
A Field Guide to Misleading "Best Sellers"
It helps to name the failure patterns directly, because most merchandisers have seen every one of these without necessarily having language for it.
High Sales, Permanent Stock-Outs
A product that reliably sells out every cycle looks like a hero on a sales report and behaves like a liability on the storefront — constantly directing interested traffic toward an "unavailable" badge.
High Traffic, Poor Conversion
A product with excellent visibility and weak purchase intent is absorbing merchandising real estate that a better-converting, lower-traffic product could use more productively.
Low Sales, Extremely High Engagement
A product with modest sales but unusually long view durations, high scroll depth, and repeated return visits may be sitting one price adjustment, one better product photo, or one piece of missing information away from a genuine breakout — evidence a sales-only ranking can't see at all.
New Products Without Historical Data
Every product starts with zero sales history. A ranking system that requires a sales track record to earn visibility guarantees new products stay invisible long enough that they never get the chance to build one.
Discount-Boosted Products
Strong performance that evaporates the moment the discount is removed was never really performance — it was price elasticity, temporarily disguised as demand.
Products That Damage Profitability
A product that sells consistently at a loss, or at a margin thin enough that returns and payment processing eat the difference, can look successful on a revenue or units-sold report while quietly costing the business money every time it sells.
Strong Repeat-Purchase Potential
A product with a modest first sale but an unusually high rate of customers returning to buy it again — or buying complementary items afterward — carries a kind of long-term value that a single-month snapshot will never capture.
Seasonal Products
A product that's genuinely excellent in December and nearly irrelevant in June needs a ranking approach that understands timing, not a static score computed once and left unchanged.
Evergreen Products
A steady, unglamorous performer that sells consistently all year, without spikes, can be undervalued next to seasonal stars purely because its month-over-month numbers never look dramatic — even though its cumulative contribution may be larger.
Toward a More Honest Definition of Product Success
None of this is an argument that sales data doesn't matter. It's an argument that sales data, alone, was never sufficient to answer the question "should this product get more visibility?" A more honest framework treats product success as multidimensional, evaluating several categories of evidence together rather than picking one and discarding the rest.
| Dimension | What it captures | What it misses alone |
|---|---|---|
| Commercial performance | Sales, revenue, order volume | Margin, discount dependency, sustainability |
| Customer interest | Views, engagement, return visits | Whether interest converts to purchase |
| Conversion efficiency | Purchase rate relative to traffic | Sample size and statistical reliability |
| Inventory health | Stock position relative to demand | Underlying customer desirability |
| Demand sustainability | Performance without discount support | Short-term promotional spikes |
| Merchandising stability | Consistency of performance over time | Emerging, high-potential products |
| Future selling potential | Early trajectory, repeat-purchase signals | Proven, mature track record |
Each row is a legitimate lens. Each row, used alone, distorts the picture in a specific and predictable direction. A product success score worth relying on for merchandising decisions has to combine several of these dimensions rather than defaulting to whichever one is easiest to pull from a sales report.
Why This Requires Judgment, Not Just Data
Combining these dimensions isn't a matter of averaging a few numbers together. Different dimensions carry different reliability depending on how much data supports them — a conversion rate from ten thousand sessions deserves more trust than one from twenty. Different dimensions matter more or less depending on business context — a store trying to clear seasonal inventory before it becomes dead stock has different priorities than one trying to build a loyal repeat customer base. Historical performance and forward-looking potential need to be weighed against each other, not simply added.
This is, at its core, closer to a forecasting problem than a reporting problem. A best-seller list reports what happened. A genuinely useful product scoring approach estimates, from everything currently known, what deserves visibility next — which is a fundamentally different exercise, and one that benefits from combining many weaker signals rather than trusting any single strong-looking one completely.
How This Shows Up in Practice
Without walking through the mechanics of any specific system — that's genuinely proprietary territory for platforms that specialize in it — it's useful to describe the shape of what a more complete evaluation tends to draw on simultaneously: commercial performance measured over a meaningful window rather than a single snapshot; customer engagement that distinguishes genuine interest from passive traffic; inventory position that accounts for whether visibility can actually be fulfilled; and some accounting for how much confidence a given data point deserves, so a handful of lucky early sales on a new product isn't mistaken for a proven trend.
Some ecommerce platforms — Peloran among them — have moved in this direction, building ecommerce merchandising tools around multidimensional scoring rather than a single sortable column, precisely because a single column was never built to answer a question this layered. The point isn't the specific mechanics of any one system. It's the underlying shift: treating product success as something evaluated across several dimensions continuously, rather than reported from one number periodically.
A More Holistic Way to Evaluate Product Success
None of the individual metrics discussed here — revenue, traffic, conversion rate, margin, velocity — is wrong. Each one is answering a real, narrow question accurately. The mistake is treating any one of them as a stand-in for the much broader question a merchandising decision actually needs answered: does this product deserve more of the store's attention right now, given everything currently known about how it performs, how customers respond to it, whether the business can actually fulfill the demand it generates, and whether its current numbers reflect durable interest or a temporary distortion like a steep discount.
A genuinely holistic approach doesn't discard best-seller data, traffic data, or margin data. It refuses to let any single one of them make the decision alone, and it stays willing to update its judgment as new evidence comes in — a product's position on a homepage isn't a reward earned once and kept forever, it's an ongoing bet that deserves to be reassessed as the underlying signals shift.
The next time a "Best Sellers" report gets pulled up before a merchandising decision, it's worth asking a harder question first: successful by which measure, over what window, at what cost, and for how much longer? Most of the time, the honest answer requires more than one column of data — and merchants who start asking that question tend to find they've been mistaking volume for success for longer than they realized.