Most Shopify stores doing real traffic don’t have a traffic problem. They have a conversion problem they can’t see — because nobody is measuring the right thing at the right step. Here are the five leaks I find most often, and how to close them without redesigning the whole store.
1. Hiding cost until the last possible moment
The single most reliable way to lose a buyer is to surprise them with a number at checkout. Shipping cost revealed only on the final step is the classic offender. By then the buyer has committed emotionally, sees an unexpected total, and feels tricked — so they leave.
The fix: surface shipping logic early. A free-shipping progress bar in the cart drawer does two jobs at once — it removes the surprise and it nudges average order value upward. Show the cost, or show the path to free shipping, before the buyer reaches payment.
2. Treating the mobile PDP like the desktop one, shrunk
Seventy per cent of your traffic is on mobile, and the mobile product page is usually an afterthought — the desktop layout reflowed into a single column with the buy box pushed below a wall of marketing copy.
The fix: on mobile, the primary action belongs above the fold. Lead with the product, the price, and the add-to-cart. The story, the specs, the reviews — those go below, for the people who scroll. Design the mobile page first, then scale up.
3. No proof at the decision point
Reviews at the bottom of the page are reviews most buyers never see. The moment of decision is at the buy box, and that’s exactly where most stores have nothing — no guarantee, no social proof, no reassurance.
The fix: cluster trust signals near the add-to-cart. A specific guarantee, a verified-review count, a genuine low-stock signal where it’s true. Not a wall of badges — two or three honest signals at the point of hesitation.
4. Optimising by opinion instead of by test
The most expensive CRO mistake isn’t a bad button colour. It’s shipping changes based on whoever argued loudest in the meeting, with no control and no measurement. You can’t learn from a change you didn’t measure.
The fix: every meaningful change is a hypothesis with a control and a pre-registered success metric. You will be wrong about a third of the time — that’s normal and it’s the point. A real programme keeps what wins and reverts what doesn’t, and compounds the learning.
5. Stopping at the first win
A brand runs one good test, gets a lift, and declares CRO “done.” But conversion optimisation isn’t a project with an end date. It’s a capability. The brands that pull away from their competitors are the ones running tests continuously, building a library of what works for their audience.
The fix: treat CRO as an ongoing cadence, not a one-off engagement. Weekly is ideal: review, ship the next test, read the result, decide. The compounding is where the real money is.
None of this is exotic. It’s discipline applied to the parts of the funnel where money actually leaks. If your store is doing the traffic and not the revenue, the answer is almost never “more traffic.” It’s closing the leaks you can’t currently see.