E-commerce Conversion Optimisation: Product Page and Checkout Basics
When a store gets visitors but few orders, the instinct is to buy more traffic. Sometimes that is right. Often the cheaper move is to find the single step where the most shoppers leave and fix that first, because every visitor you add passes through the same leak.
The decision here is where to spend your next hour or your next budget: on more traffic, or on the page that converts it. A funnel read tells you.
Why the usual advice fails
Lists of conversion tips are everywhere: add reviews, change the button colour, add a countdown. They fail for a simple reason. They treat every store as having the same problem. A store whose product page is unclear and a store whose checkout asks for too much have different problems, and the same tip helps one and does nothing for the other.
Copying a competitor's layout has the same flaw. You see the page, not the audience, the price, or the traffic source that makes it work for them.
The model: measure the funnel, then form a hypothesis
The method has three moves. Measure the share of shoppers who pass each step. Find the step with the largest avoidable drop. Write one hypothesis for why, then change one thing.
Google's ecommerce documentation defines the standard events that map to these steps, including viewing an item, adding to cart, beginning checkout and purchase. If those events are not tracked correctly, your funnel is fiction, so verify them first with the tracking audit checklist and the GA4 e-commerce events guide.
A worked illustrative example
Illustrative example, not real results. In one month a store has 10,000 sessions. 6,000 view a product, 600 add to cart, 360 begin checkout and 108 buy.
Overall, 108 of 10,000 sessions buy, which is 1.08 percent.
The arithmetic: 6,000 of 10,000 is 60 percent, 600 of 6,000 is 10 percent, 360 of 600 is 60 percent, and 108 of 360 is 30 percent. The weakest step in relative terms is product view to cart. A hypothesis follows: shoppers reach the product but do not see enough reason to buy.
Now size the prize. If a product-page change lifted that step from 10 to 12 percent, with the later steps unchanged, there would be 720 carts, 432 checkouts and 129.6 orders, about 130 instead of 108. A checkout change that lifted the last step from 30 to 33 percent would give 118.8. These numbers are only a way to compare where effort pays most. They are not predictions.
Seven product-page checks
Clear offer
In five seconds, can a stranger say what it is, who it is for and the price?
Mobile first
Open it on a mid-range phone on mobile data. Is the buy button reachable without scrolling a wall of text?
Honest images
Do the images show the real product, size and use?
Trust cues
Return policy, contact details and payment options visible, without fake badges.
Delivery clarity
Cost and expected time shown before checkout, not discovered there.
Specific copy
Answers the questions people actually ask: size, material, what is included.
One main action
A single clear button, with variants and options easy to choose.
- Read it as a stranger
- On a phone first
Checkout basics
Checkout is where intent is highest and patience lowest. Review it with these questions.
- Fields. Remove every field you do not need to deliver the order.
- Guest purchase. Forcing account creation adds friction, so offer a guest route unless you truly need accounts.
- Surprises. Show shipping and fees early. A new cost at the last step feels like a trap.
- Payment options. Offer the methods your customers actually use. For cash-on-delivery stores, see the break-even guide before treating a placed order as a sale.
- Errors. Test what happens with a wrong phone number or a failed payment. Is the message clear, and is the entered data kept?
- Speed. A slow page loses people before they read it.
How to change things without fooling yourself
Change one thing at a time and write down the hypothesis, the date and the expected effect. If two changes ship together you will not know which one worked.
Small stores often do not have enough traffic to run a statistically sound A/B test. In that case a before-and-after comparison is weak evidence, since a promotion, season or traffic mix can move the rate on its own. Prefer larger, clearer fixes you can justify by logic, such as removing a field, and be sceptical of tiny differences.
When the recommendation changes
- Traffic is the problem. If most visitors are unqualified, a better page will not rescue them. Check the traffic source before the page.
- Very low volume. With a few hundred sessions a month, funnel percentages jump around. Use qualitative checks, such as watching real people use the page, instead of ratios.
- The offer is weak. No layout fixes a product nobody wants at that price.
What the data cannot tell you
A funnel shows where people leave, not why. It cannot see shoppers who compared you on another device, who left to ask on a messaging app, or who bought later by a different route. Treat each number as a clue and each hypothesis as a guess to test.
Your next action
Pull the last 30 days of funnel numbers, find the weakest relative step, and write one hypothesis with one change. If you want a structured review, the CRO & Funnel Audit covers the page, checkout and measurement together.