How to Find Your Best-Selling and Worst-Reviewed Products

Two different signals, not one

How much a product sells tells you about demand, visibility, pricing, and how well it's merchandised. How it's reviewed tells you about the actual experience of owning it. Those are genuinely different things, and a product can score high on one and low on the other without anyone noticing, because most stores only look at one signal at a time — a sales report that doesn't mention ratings, and a review feed that doesn't mention volume.

The fix isn't picking one signal over the other. It's plotting every product against both at once, which turns two separate lists into a single simple framework: four quadrants, each one calling for a different action.

Why most stores don't already do this

The framework itself isn't complicated. The obstacle is almost never the analysis — it's that sales data and review data usually live in two different systems that were never built to talk to each other. Sales numbers sit in Shopify or WooCommerce, broken down by SKU. Reviews sit in a separate app or on Google, broken down by nothing in particular, since most review tools have no concept of which specific product an order contained. Building the four-quadrant view by hand means manually cross-referencing a sales export against a reviews export, one product at a time, which is exactly tedious enough that most stores never actually do it and fall back on checking each signal separately instead.

Tailwind's Orders app is built specifically to close that gap. It connects to a Shopify or WooCommerce store, imports the order history and full product catalog including line items, and ties every order to the specific products it contained — so you can find top products and the ones customers complain about in one place instead of reconciling two spreadsheets by hand. Once sales and review sentiment are reported against the same product list, the four-quadrant split takes minutes to build instead of an afternoon, and it stays current automatically as new orders and reviews come in.

The four quadrants

Hits: high sales, good reviews

This is the quadrant every store wants every product in. High volume and a strong rating mean the product is doing exactly what it's supposed to — reinforcing itself, generating the reviews that help the next shopper decide, and earning its spot on the homepage or the ad budget.

The action here is mostly protective: don't change what's working. Keep an eye on whether the rating holds as volume grows — a product that was excellent at 50 orders a month can start slipping at 500 if a supplier swaps materials, a packaging line changes, or quality control hasn't scaled with demand. A hit that's starting to drift toward "problem hit" is worth catching while it's still a small drift, not once the rating has already moved enough to show up in the aggregate.

It's also worth asking what's actually driving a hit's success, so you can repeat it elsewhere. Sometimes it's the product itself; sometimes it's a product page that happens to answer the exact questions customers have before anyone else's does. If it's the latter, that page is a template worth copying across other listings, not just a one-off success.

Problem hits: high sales, bad reviews

This is the quadrant that needs attention fastest, and it's also the one that's easiest to miss if you're only looking at a store-wide average. High sales volume means a lot of customers are going through the same bad experience, which is a bigger and more urgent problem than the same rating on a product nobody buys — the damage compounds with every order, not just sits there.

A problem hit usually traces back to something specific and fixable: a sizing chart that doesn't match the actual garment, a product photo that oversells the real color or material, a packaging choice that lets items arrive damaged, a description that sets an expectation the product doesn't meet. The sales volume is actually an advantage here — you have enough review text to find the pattern quickly, instead of guessing from three or four data points.

Treat this quadrant as a short list, not a long one. If more than a handful of products land here, the problem is probably not product-specific — it's something systemic, like a shipping carrier that's started damaging packages across the board, or a supplier change that touched more than one line at once.

Quiet stars: low sales, good reviews

A quiet star is the opposite kind of surprise — a product nobody's paying attention to that's quietly earning excellent reviews from the small number of people who do buy it. These are usually catalog items that launched without much marketing push, got buried under newer products, or never had a clear merchandising slot.

The action here is promotion, not fixing. A quiet star is evidence that demand would likely grow if more people saw the product — feature it on the homepage, add it to a bundle with a hit product, mention the review highlights directly on the product page, or include it in an email to existing customers who bought something similar. The risk with this quadrant isn't that it's broken. It's that it's invisible, and invisible products stay invisible until someone deliberately decides to surface them.

Quiet stars are also a useful check on your ad spend. If you're paying to acquire traffic for a mediocre hit while a genuinely well-reviewed product sits two pages deep in your catalog with no promotion at all, that's a budget allocation problem hiding behind a sales report that only shows volume, not potential.

Candidates to discontinue or fix: low sales, bad reviews

Low volume and a bad rating is the lowest-urgency quadrant in terms of damage being done right now, simply because few people are buying the product to begin with. But it's still worth a decision, because a product sitting here is quietly taking up a product page, inventory space, and search visibility for no return.

There are two honest paths out of this quadrant: fix it, or drop it. Fixing makes sense if the underlying issue is cheap and specific — a product description rewrite, a size chart correction, a packaging tweak. Dropping makes sense if the low sales reflect weak demand independent of the review problem, in which case fixing the reviews wouldn't move much anyway. What doesn't make sense is leaving it exactly as is — a stagnant product with a bad rating isn't neutral; it's a small, ongoing drag on anyone who does stumble onto it.

One more reason to resolve this quadrant deliberately: a product with a weak rating sitting in your catalog can drag down how your store reads as a whole to a shopper who happens to click into it after browsing a hit elsewhere. A single bad product page is rarely fatal to a sale on its own, but it chips away at the trust a strong catalog is otherwise building.

A note on how the review data gets there

It's worth being explicit about one thing: every quadrant above only means something if the underlying reviews came from asking every customer the same way, regardless of what you expect them to say. Filtering who gets asked for a review based on predicted satisfaction — skipping customers who returned an item, or only sending requests to people who didn't file a support ticket — would quietly erase the "problem hits" and "candidates to fix" quadrants entirely, not by fixing the underlying issues, but by making sure you never hear about them. The whole value of this framework depends on the review data being honest, which means every buyer gets asked, the same way, after delivery.

Putting it into practice

Start by listing every product with two columns next to it: units sold in the last quarter, and average review rating over the same window. Sort by volume first to find your hits and problem hits, then sort by rating to catch the quiet stars and the candidates to fix. Most stores find their problem hits within the first pass — it's usually a short, uncomfortable list that's been hiding in plain sight.

Revisit the table quarterly rather than constantly. Review ratings and sales volume both move slowly enough that a monthly rebuild is mostly noise, but a product can genuinely shift quadrants over a few months as a fix takes hold or a new product cannibalizes an old one's sales.

Keep the thresholds simple rather than precise. You don't need a formal cutoff for "high" versus "low" sales or "good" versus "bad" reviews — ranking products against each other and looking at the ones clustered at the extremes is usually obvious without needing a statistical definition of the boundary. The framework is meant to focus attention, not produce a perfectly defensible score.

This same approach pairs well with the native analytics already sitting in your store — see Shopify analytics for small stores for the sales-side reports worth checking regularly before you ever get to the review side of the table.

If you're ready to stop reconciling spreadsheets to build this view, Tailwind's Orders app reports sales and reviews by product automatically once it's connected to your store. Tools start at $5 a month, the Full plan is $50 a month, and the first month is free with no credit card and no contract.