A store with no traffic has one problem. A store with traffic and no sales has four, and from the dashboard they look identical: sessions arrive, the conversion rate sits somewhere disappointing, and nobody knows which lever to pull. So the homepage gets redesigned. Then the button changes colour. Then a popup appears. Six weeks later the number hasn't moved, because the problem was never on the page that got redesigned.
The takeaway up front: only one of the four causes of "traffic but no sales" lives on your product page, and it is rarely the one people fix first. The other three are the traffic itself, the path to payment, and the offer. Each has a different symptom, a different fix, and a different cost — and you can tell them apart in an afternoon with data you already have.
Stop looking at your sitewide conversion rate
A sitewide conversion rate is an average across visitors who have nothing in common: someone who searched your brand name, someone who tapped a video ad out of boredom, someone returning to finish a purchase. Averaging them produces a number that describes nobody and points nowhere.
Open your analytics and split the same conversion rate four ways:
- By source — search, paid social, paid search, email, direct, referral.
- By landing page — where people arrive, not where they end up.
- By device — mobile against desktop, separately for each source.
- By funnel step — sessions, product page views, add-to-cart, checkout started, order placed.
Now read the split. This is the whole diagnosis:
- Low on one traffic source, healthy on the others → a traffic problem. The page works; the audience doesn't.
- Healthy add-to-cart, poor checkout completion → a friction problem. They wanted it and something stopped them.
- Plenty of product page views, very few add-to-carts → an unanswered objection. They're interested and something is missing.
- Low everywhere, including branded search and returning visitors → an offer problem. The people who already know you and came looking for you aren't buying either.
That last one is the important control. Someone who types your brand name into a search box arrived with maximum intent. If they don't convert, no amount of page work will help, because nothing about the traffic is wrong.
Cause 1: the traffic is wrong
This is the most common cause and the least popular diagnosis, because it means the cheap traffic you were pleased about is the problem.
It's tempting for an obvious reason: broad targeting and general-interest keywords are the easiest traffic to get and the cheapest per click. A post that travels widely, a broad-interest ad set, an article ranking for a term that sounds related to what you sell — all of it produces sessions, and sessions look like progress.
But intent doesn't survive the trip. Someone reading "how to clean suede" is not shopping for suede boots, and someone who watched an entertaining video is not in a buying frame of mind. They arrive, they look, they leave, and your conversion rate absorbs the damage.
What fixes it: changing what you buy or target, never the page. Move spend toward higher-intent placements, aim content at terms with a buying motive rather than a curious one, and judge each source on revenue per session rather than on clicks. You cannot fix a traffic-quality problem on the product page — and that's the trap, because the product page is the only thing you fully control, so it's what gets edited.
Before you conclude a channel is bad, check the economics honestly: a source can convert poorly and still pay, or convert well and still lose money once fulfilment is counted. True customer acquisition cost covers the arithmetic that decides it.
Cause 2: the page doesn't answer the objection
Symptom: people reach the product, spend real time there, and don't add to cart.
Every product has two or three questions that must be answered before anyone will pay, and they're specific to the category. Will it fit. What is it made of. When does it arrive. Does it work with the one I already own. What happens if it's wrong. Miss one and the visitor doesn't complain — they leave, and you record it as a low conversion rate.
It's tempting to write about how good the product is instead, because that's the enjoyable part and it feels like selling. But nobody is stuck on whether your product is good. They're stuck on a detail.
What fixes it: find the real objections instead of guessing them. Read your pre-sale messages and support inbox and count what people ask before buying — that list is your missing content, ranked by frequency, for free. Then answer each one on the page, in the order a buyer needs it, in plain language. Delivery time and returns belong near the buy button, not behind a footer link.
Cause 3: friction between wanting it and paying
Symptom: add-to-cart is healthy, completed orders are not. They decided to buy and something got in the way.
Friction is tempting to ignore because each individual piece of it seems reasonable: an account requirement that keeps your data tidy, a coupon field that pays for itself, a shipping cost revealed at the last step because it genuinely depends on the address. Each is defensible alone. Stacked, they are a wall.
What fixes it: walk your own checkout on your own phone, on mobile data rather than office Wi-Fi, as a brand-new customer with a real card. Time it. Count the steps and the form fields. The usual offenders are a forced account, a cost that appears only at the final step, a payment method your customers expect and you don't offer, and a form that fights the autofill keyboard.
This is the cheapest of the four to fix and the most binary — friction either exists or it doesn't. Once the checkout is clean, the people who still leave mid-purchase are a different problem, covered in how to recover abandoned carts.
Cause 4: the offer itself
Symptom: conversion is uniformly poor across every source, including branded search and returning visitors.
The offer is the whole package: the price, what shipping costs and how long it takes, the return terms, and whether a stranger has any reason to trust you with a card number. When people who deliberately came looking for you still don't buy, one of those is out of line with what they can get elsewhere.
This is the hardest diagnosis to accept, which is why it's usually reached last, after months of page edits. It also has the most substantial levers: repricing, restructuring shipping so the cost isn't a late surprise, or making the risk visible rather than hoping nobody thinks about it. How to price a product covers the first, and social proof for your online store covers the last — an unfamiliar store with no visible evidence that anyone has bought successfully is asking for a leap of faith at exactly the wrong moment.
The trade-off nobody mentions
Conversion rate improves when you remove bad traffic. That is a real gain in efficiency and it can still leave you with less money, because some poorly-converting traffic is profitable at volume.
So don't optimise the percentage. Optimise revenue per session, minus what that session cost to acquire, source by source. A channel converting at half the rate of another can be the better channel if it costs a quarter as much and the orders are the same size. Judging every source against one sitewide target will eventually lead you to switch off something that was paying.
What the dashboard can't tell you
Analytics tells you where people stop. It never tells you why. Two cheap habits close that gap and neither needs a tool budget:
- Watch five real sessions through whatever recording your platform offers, on mobile. You'll see the scroll that hunts for a detail that isn't there.
- Ask five recent buyers what nearly stopped them. People who bought will happily tell you what almost made them leave, and it's usually the same thing three times.
Five is a rule of thumb, not a sample size — it won't prove anything, but it reliably surfaces the obvious problem that no chart names.
Do this in order
- Segment conversion by source, landing page, device, and funnel step.
- Check branded and returning visitors first — if they don't convert, it's the offer, and nothing else matters yet.
- Walk your own checkout on a phone. Fix anything that annoys you.
- Count the pre-sale questions you actually receive and answer the top three on the page.
- Re-check the same segments after each change, one at a time. Change three things at once and you learn nothing.
The wider operational picture this sits inside — inventory, fulfilment, and the systems around all of it — is in the ecommerce operations guide.
FAQ
What is a good ecommerce conversion rate? The useful benchmark is your own rate by source, tracked over time. A single sitewide number compared against a published industry average tells you nothing actionable, because the traffic mix behind the two numbers is different.
How much traffic do I need before the numbers mean anything? Enough orders, not enough sessions. A handful of orders in a period can't distinguish a real change from ordinary variation. If a segment produces very few orders, treat it as a hint and confirm it with a second look before acting.
Should I redesign the site? Only if the diagnosis points at the page. A redesign changes everything at once, so if it works you won't know why, and if it doesn't you've spent the budget. Fix one identified cause and measure that.
My ads convert badly. Is it the ads or the site? Compare that source against your others. If search and email convert acceptably on the same pages, the site works and the mismatch is the ad audience or the promise the ad made.
Can slow loading be the whole problem? It can be a real contributor, especially on mobile data, and it shows up as people leaving before the page finishes. It's worth checking early because it's measurable — but treat it as one form of friction rather than the explanation for everything.
Diagnose before you redesign. Segment the conversion rate you already have, check whether the people who came looking for you specifically are buying, and fix the one cause the split actually points at. For more playbooks on growing an online store without guesswork, visit Rocket Maxx.
