Ecommerce conversion rate optimization: a method to sell more from the same traffic
A method to make your store convert more: A/B tests, pages and offers.

Conversion and A/B testing guides
Start here

Ecommerce conversion rate: what's good, how to measure it and how to raise it
The conversion rate formula, public benchmarks by category, the most common measurement mistakes and the levers to pull at each funnel stage.
Ecommerce conversion rate optimization (CRO) is the work of getting a larger share of your visitors to buy without adding traffic. The method has five steps: measure the funnel, find the stage that loses the most shoppers, write a hypothesis, A/B test the change, and keep only what wins on paid orders, not clicks.
This is the main guide for the topic on the Horsee blog. It explains why conversion matters so much when you buy traffic, how to read your store's funnel, where problems usually hide, how to prioritize what to change and how to run an A/B test that gives you a trustworthy answer. The detailed guides for each part are listed at the end.
Why does conversion matter so much when you buy traffic?
Simplified, a campaign's cost per sale is the cost per click divided by the conversion rate. That means conversion moves the cost of every order as much as the price of the ad does. An illustrative example: with $1.00 clicks and a 1% conversion rate, each sale costs $100 in media. Raise conversion to 1.5% and the same sale costs about $67, with no change to the campaign.
There is a second effect. A store that converts better can afford to pay more per click and stay profitable. That opens room to scale spend and compete for audiences that did not pencil out before. For paid-traffic operators, improving the store is often the cheapest lever available.
What counts as a "good" rate and how to calculate yours is covered in our guide to ecommerce conversion rate. This page focuses on the method for raising it.
How do you read your store's funnel?
Your overall conversion rate tells you how much the store sells, not where it leaks. For that, split the journey into stages and measure how many people make it from one to the next:
- Visit: the shopper lands on the store (homepage, product page or landing page).
- Product: they view a product page.
- Cart: they add to cart.
- Checkout: they start checkout.
- Purchase: they complete and pay.
The stage with the biggest drop, compared with what is normal for your niche and your own history, is where to start. Break the funnel down by source too (each campaign brings a different audience) and by device (most paid traffic arrives on mobile, and a page that works on desktop can fall apart on a phone).
In Horsee, UTM Analytics shows the funnel by source, medium and campaign, with visits, cart, checkout and purchase columns and the revenue for each segment. The general analytics dashboard also covers cart abandonment. The Horsee Help Center explains both screens.
Before trusting the numbers, make sure your tracking is right. Duplicated or missing events distort the whole funnel. Our guide to tracking and attribution shows how to check.
What usually hurts conversion at each stage?
The table maps funnel symptoms to their most common causes and to what is worth testing.
| Symptom | Likely cause | What to test |
|---|---|---|
| Many visitors leave without viewing a product | The ad promises one thing and the page shows another; slow page | Send the ad straight to the product page; match the headline to the ad; lighter images |
| They view the product but do not add to cart | Unclear offer, weak photos, no proof, price without context | Main benefit above the fold, in-use photos, reviews, visible guarantee, bundles |
| They add to cart but do not start checkout | Shipping costs revealed late, confusing cart | Show shipping and delivery time earlier, one clear checkout button |
| They start checkout but do not pay | Long form, forced account, too few payment options, low trust | Fewer fields, guest checkout, cards and wallets on the same step, visible policies and badges |
| They try to pay, but no sale shows up | Declined card or unpaid order | Recovery messages for declined and pending payments |
The last two rows overlap with other topics. Payment friction is covered in our guide to checkout and payments, and what to do about shoppers who leave is covered in recovering lost sales with email, WhatsApp and a CRM.
What is the conversion optimization method, step by step?
- Measure. Funnel by stage, source and device over at least a few weeks. Write down conversion rate, average order value and revenue per visitor as your baseline.
- Diagnose. Pick the stage that loses the most. Look at the page as a shopper would, on a phone, coming from the ad. Read the questions your support team gets: they reveal what the page does not answer.
- Write a hypothesis. Use the format "if we change X, we expect Y, because Z". Example: "if we show the delivery estimate next to the buy button, more people will add to cart, because delivery time is the most common pre-sale question in our inbox".
- Test. When traffic allows, run an A/B test with both versions live at the same time. When it does not, change and compare periods carefully (see below).
- Decide and log. Keep the winner, drop the loser and write down the result, including what failed. A test log stops you from repeating ideas that already lost.
Where do good hypotheses come from?
The funnel shows where you lose people, not why. For the why, bring in qualitative sources:
- Support conversations: pre-sale questions on email and chat are the doubts your page leaves open. If everyone asks about delivery time, delivery time belongs on the page.
- Post-purchase survey: one question in the confirmation email ("what almost stopped you from buying?") surfaces objections you cannot see from the inside.
- Session recordings and heatmaps: tools like these show where people stop scrolling, where they click on things that do nothing and where they bounce. Watch a handful of sessions from shoppers who left at checkout.
- A walkthrough as a customer: open the store on your phone from the ad and note every moment you would hesitate. Ask someone outside the team to do the same.
- Reviews and returns: what customers praise becomes copy; what drives returns may point to a promise the page explains badly.
Conversion rate, AOV or revenue per visitor: what should you optimize?
Conversion rate is the headline metric, but it is not the only one that matters. A change can lift conversion and lower average order value, or the other way around. Watch three numbers together:
| Metric | How to calculate | What it shows |
|---|---|---|
| Conversion rate | Orders ÷ sessions (or visitors) | How many people buy |
| Average order value (AOV) | Revenue ÷ orders | How much each order is worth |
| Revenue per visitor | Revenue ÷ visitors | The combined effect of both |
Revenue per visitor is the best metric for offer tests, such as bundles and tiered discounts, because it accounts for both effects. For page-clarity tests, conversion rate is usually enough, as long as AOV holds steady.
How do you prioritize what to test?
You will always have more ideas than traffic. A simple way to choose is to score each idea from 1 to 5 on three criteria and start with the highest totals:
| Criterion | Question |
|---|---|
| Impact | How many visitors see this part of the page, and how much could the change sway their decision? |
| Confidence | Is there evidence for it (funnel data, support questions, recordings, surveys)? |
| Ease | How cheap is the variant to build? (high score for easy) |
Big, visible changes tend to return an answer faster than small tweaks. Changing the offer, the hero image or the order of page sections usually moves results more than changing a button color.
How do you turn CRO into a routine?
Conversion optimization works as an ongoing process, not a one-off project. A simple cadence for a small team:
- Every week: check the funnel by campaign and note unusual drops. A sudden drop is usually technical (page down, payments failing, a broken tracking event), not a change in behavior.
- Every two to four weeks: close the running test, log the result and start the next one on your prioritized list.
- Every month: review your hypothesis list with what support and recordings revealed, and re-rank by impact, confidence and ease.
- With every new product or campaign: check that the landing page still matches the ad's promise.
Above all, always have a test running on your highest-traffic pages. A page with no test running is traffic that is not teaching you anything.
What is an A/B test, and when is it worth it?
In an A/B test, traffic is split between two versions of the same page: the current one (control) and the new one (variant). Both run during the same period, for the same kind of visitor, so any difference in results can be credited to the change rather than the week, a holiday or a campaign.
It pays off when the page gets enough traffic to collect conversions on both versions in a reasonable time. With only a few sales a week, a test can take months to give an answer. In that case, ship the obvious fixes directly (slow page, missing information, broken mobile checkout) and save A/B tests for decisions where there is real doubt.
Rules for a trustworthy A/B test
- One hypothesis per test. If the variant changes the offer, the photos and the copy at once, you will know it won but not why. Sometimes that is fine; just be clear about what you are testing.
- Measure sales, not clicks. The primary metric is paid orders, or revenue per visitor. More button clicks without more orders do not help.
- Run full weeks. Behavior shifts between weekdays and weekends. At least 7 days keeps one odd day from deciding the test.
- Wait for enough conversions on both sides. With few purchases, the difference may be noise.
- Do not stop when the chart looks good. Checking daily and stopping at the first spike is the most common way to crown a false winner.
- Do not touch a running test. Changing the traffic split or editing the variant mid-test contaminates the data.
How A/B testing works in Horsee
In Horsee, an A/B test starts from a store page (homepage, product page or a custom page). In Sell → Pages, create the test from the page's three-dot menu. You set the traffic split, and variant B is created as a copy of the page for you to edit. Both versions are served at the same URL, with the split decided at the edge before the page loads.
The test dashboard shows unique visitors, conversions, revenue, conversion rate, revenue per visitor and average order value for each version, plus the variant's lift, significance and a 95% confidence interval. Conversions are counted on paid orders. By default, the verdict waits for at least 7 days and 30 conversions per version before naming a winner. When you have a result, Promote and Publish puts the winning version live at the original URL and ends the test. All of your store's tests live under Marketing → A/B Testing.
What should an ecommerce store test first?
Some hypotheses come up in almost every store that sells with paid traffic. Use them as a starting point, not a recipe:
- Top of the product page: the ad's promise repeated in the headline, the main benefit and the price with context (bundle, installments, shipping).
- Hero image: the product in use versus the product on its own, or the reverse.
- Social proof near the button: reviews and customer photos before the long description.
- Guarantee and returns in view: the policy summed up in one line near the buy button.
- Offer: a single unit versus a bundle with a tiered discount.
- Section order: FAQs higher up for products that raise lots of questions.
- Landing page: sending the ad to the product page versus a longer-form landing page.
What if your traffic is too small for A/B tests?
Use a disciplined before-and-after comparison:
- Record conversion rate and revenue per visitor for the last two to four weeks.
- Make a single change and leave everything else alone (including campaigns, if you can).
- Compare a period of the same length, with the same days of the week.
- Be skeptical of small differences: seasonality and ad changes can explain them entirely.
It is less reliable than an A/B test, but far better than changing five things at once and never knowing what worked.
Speed, mobile and trust: the basics before testing
Some problems do not need a test; they need a fix. Before spending traffic on A/B tests, check:
- Mobile speed: heavy images and third-party scripts are the usual culprits. Open the store on an ordinary phone, on a cellular connection, from the ad.
- Mobile layout: buy button visible, text readable without zooming, nothing cut off.
- Essential information: price, shipping, delivery time, returns, payment options and contact details should be easy to find.
- A checkout that works: place a test order with every payment method you offer.
- The payment options your market expects: cards and wallets such as Apple Pay and Google Pay, PayPal for shoppers who prefer it, and local methods in the countries you sell to. A missing favorite method shows up as checkout drop-off, not as a complaint.
One full cycle, from data to decision (illustrative example)
To see the whole method at work, picture a pet accessories store selling through Instagram ads. The numbers here are made up for the example.
- Measure: the funnel for the last four weeks shows plenty of people reaching the product page, but few adding to cart on mobile. On desktop, the step-through is much better.
- Diagnose: on a phone, the buy button only appears after three scrolls, below a large gallery. In the inbox, the most common question is about sizing.
- Hypothesis: "if we move the size selector and the button right under the first photo, with a size chart next to it, more mobile visitors will add to cart, because today they cannot see the button and are unsure about size".
- Test: variant B with the new order, a 50/50 split, running for two full weeks.
- Decide: the variant brings in more paid orders without a drop in AOV; the result clears the minimum days and conversions and shows as significant. The store promotes the variant and logs the lesson: on mobile, the button belongs on the first screen.
Common conversion optimization mistakes
- Copying another store's "winning test". Their audience and offer are different. Use it as an idea and validate it on your own traffic.
- Watching conversion rate alone. A variant can sell more orders of lower value. Check revenue per visitor and AOV too.
- Ending tests early. A few days and a few sales produce winners that vanish later.
- Testing with broken tracking. If the purchase event does not arrive, the test measures noise.
- Forgetting the people who already left. Optimizing the page without recovering carts leaves money on the table.
- Not logging what you tested. Six months later, nobody remembers why the page looks the way it does.
Guides in this topic
- Ecommerce conversion rate: what's good, how to measure it and how to raise it: the formula, public benchmarks and the levers at each funnel stage.
Upcoming guides in this topic will cover A/B testing an online store in practice, the anatomy of a product page that converts, store speed and statistical significance explained for store owners.
Related topics
- Cart recovery and CRM: how to bring back shoppers who almost bought.
- Checkout and payments: the last step of the funnel, where sales slip away most often.
- Tracking and attribution: reliable data to know whether a change worked.
Test, measure and scale on one platform
In Horsee, A/B testing, the funnel by campaign, checkout and cart recovery live on the same platform, and each version's conversion is measured on paid orders, with no external testing tool. See the details on the Horsee growth page: A/B testing, affiliates and email, and compare plans on the pricing page.
From the guide to a live store.
Build your store with AI in minutes, with checkout, payments and server-side tracking ready.
