A store at $40K a month needs about seven weeks of traffic to detect a 20 percent lift in purchase rate, and roughly nine in ten tests don’t beat the original anyway.
Now, let’s say you sell home goods on Shopify. You’re doing about $60K a month, your dashboard says 2.0 percent conversion, and every app pitch in your inbox promises to lift it. You have a list of ideas: a new checkout block, a bundle widget, a spin to win wheel, a headline test. Which one do you run?
I’d argue none of them yet. That 2.0 percent is the last number in a chain of four, and until you know which link is weak you’re guessing with your own traffic. This guide shows how to read the chain, how much traffic a test really needs at your stage, and what the podcast guests who do this work for a living check first.
If shoppers aren’t reaching your site, or they click a search result and land on the wrong page, start upstream with twelve conversion fixes measured in the search result and on category pages. This piece picks up once the visitor has arrived.
A single conversion rate can’t tell you what to fix because it blends four separate decisions into one number: whether the shopper looks at a product, adds it to the cart, starts checkout and pays. A drop in any one of them shows up the same way on your dashboard.
To split it, measure all four steps in one analytics tool so every step counts the same visitors. GA4 is the practical choice because Shopify’s integration sends GA4’s standard ecommerce events: view_item, add_to_cart, begin_checkout and purchase. Count sessions that contain each event, then divide each step by the one before it. Check first that every event actually fires on a test order, because a broken event invents a bottleneck that isn’t there.
Use Shopify as the cross-check. The Conversion rate breakdown report in Shopify Analytics shows sessions with cart additions, sessions that reached checkout and sessions that completed checkout, but it calculates every step against total sessions, and by default it’s an open funnel that counts shoppers who skip steps. Switch the Open Funnel toggle off to see only sessions that went cart, then checkout, then purchase in order. Don’t mix its session counts with GA4’s in one calculation; the two tools define and count sessions differently. One more wrinkle: shoppers who tap an accelerated checkout button on the product page skip the cart, so a store that leans on those buttons will show a lower cart rate than its real buying intent.
Anthony Morgan, founder and CEO of the CRO agency Enavi, calls this the intrasite funnel, and on episode 440 he made the case that the headline rate is close to a vanity metric. If all you know is that conversion is 0.9 percent, you have no idea where the bottleneck is. If you know product page visitors add to cart at 6 percent, you know which page to open.
The four steps worth tracking are landing to product view, product view to add to cart, add to cart to checkout, and checkout to purchase, and each has a rough target to compare against before you compare against anyone’s headline conversion rate.
Anthony shared Enavi’s targets, drawn from its own client data, on episodes 440 and 468. For landing to product view, “the target baseline is 50 to 70%.” For product view to add to cart, “12 to 18% is where we want to be.” For add to cart to checkout, “we want 40% of users that add to cart are checking out.” For checkout to purchase, Enavi shoots for a minimum of 45 percent. Treat these as one agency’s benchmarks, not industry law; your own trend from last quarter is the other comparison that matters.
The checkout number surprised me most. Anthony’s view is that most Shopify brands clear 45 percent, because shoppers already know the Shopify checkout. “Honestly, the juice isn’t worth the squeeze oftentimes on the checkout,” he said. The brands that do struggle there, in his experience, are the ones springing unexpected shipping costs, unclear delivery times, or duties and taxes on international orders at the last step.
That lines up with Baymard Institute’s survey of why US shoppers abandon checkout, updated October 2026. Leaving out people who were just browsing, 40 percent said extra costs were too high, 20 percent said delivery was too slow, and 18 percent said the site wanted an account. Some of that is disclosure: showing the shipping cost and a delivery date next to the add to cart button stops the surprise at checkout. Some of it is the offer itself, and no wording fixes a shipping charge that’s too high or a delivery that’s too slow. How you hit a faster delivery date belongs to your fulfillment setup, covered stage by stage in the ecommerce fulfillment guide. If your checkout really is under 45 percent, the guide to which Shopify checkout changes are worth making covers what you can change and what to leave.
Reading the funnel means pulling the four step rates, finding the one furthest below target, then splitting it by device and traffic source until the problem points at one page on one screen size. Here is how that looks on one store.
This is a labelled composite, built from the funnel patterns Anthony described and the testing math later in this guide. It is not a real store, and every figure is illustrative.
The store sells home goods at a $75 average order. Last month GA4 recorded 40,000 sessions and 800 purchases, so 2.0 percent conversion and $60K in revenue. Of those sessions, 24,000 viewed a product, 2,400 added to cart and 1,440 began checkout.
Divide each number by the one above it. Landing to product view is 60 percent, inside Anthony’s range. Product view to add to cart is 10 percent, below his 12 percent floor. Cart to checkout is 60 percent against a 40 percent target, and checkout to purchase is about 56 percent against 45. One step is on fire, and it isn’t the checkout.
Split that step by device. Mobile carries about 72 percent of product viewing sessions and adds to cart at about 7 percent. Desktop adds to cart at about 18 percent. So this isn’t a product page problem. It’s a mobile product page problem.
Open the best selling product on a phone and scroll. On this composite store the price is there, but the shipping cost, the delivery estimate and the dimensions sit below a reviews widget and two app blocks, three screens down. A shopper on a phone has to hunt for the exact answers that, in Baymard’s survey, send people away.
The fix list is mostly moving things rather than adding them. Put the delivery estimate and shipping cost directly under the add to cart button. Move dimensions into the first screen. Pull a one line review summary up near the price. Then the subtraction: take out the upsell block that pushes the button down, and turn off the entry overlay on mobile product pages. Moving mobile add to cart from 7 percent to 10 percent is a 43 percent relative change on the step that feeds everything after it. Leave the checkout alone.
An A/B test on purchase conversion needs more traffic than most Shopify stores get in a month: at a 2 percent conversion rate, detecting a 20 percent lift with 80 percent power takes about 42,000 sessions split across both versions. With less, a real 20 percent lift will often be missed.
That figure comes from the standard sample size formula at 95 percent confidence and 80 percent power with a 50/50 split, the same math behind Evan Miller’s free sample size calculator. Run it for a few store sizes and the pattern is clear. The table assumes an $80 average order, 2 percent purchase conversion and a 10 percent add to cart rate, and shows the smallest lift one month of traffic can reliably detect. Revenue is only a stand-in for sessions here; your own session count is what matters.
Two things stand out. A store at $50K a month can only detect a purchase lift of about 23 percent in one month, and routine page changes rarely move purchase rate that much. And the add to cart column is far more forgiving, because the rate is five times higher, so testing the step you’re fixing gets you an answer sooner. The catch is that more carts don’t guarantee more orders, so keep purchase rate as the check.
Then add the win rate. Jeffrey Lin, founder and CEO of ABConvert, told me on episode 493 that for brands testing consistently, leaving out tests stopped early, “in general, I would say 5 to 10%, but mostly 10%, means you will have nine failure tests out of the 10 tests you’re running.”
Put those together and here’s my view. Below about $100K a month, a one month purchase rate test on a routine page change mostly spends a month learning nothing. The exception is a big decision you can afford to run longer: a $60K store that holds a price or shipping test for five to six weeks can detect a 20 percent lift. That’s a real test. A headline tweak run for two weeks isn’t.
When your traffic is too thin to test, ask the customers who just bought what nearly stopped them, fix the obvious answer, and measure the step before and after. Some of the biggest wins I’ve heard on the show came from exactly that.
Matt Stafford, managing partner at Build Grow Scale, explained on episode 411 that the first thing his team does with every brand is put one question on the thank you page: “What was the one thing that almost made you not buy?” He said “probably six out of our 10 biggest wins have come from that question on the thank you page.”
His best example was a brand doing about $60 million a year that sold off a long video sales letter. The question collected 520 answers in seven days, and 460 of them said the video was too long. The owner had already tried shortening it and lost conversions every time. So the team added a playback speed control instead. “They got a 10% lift,” Matt said, on traffic the brand was already paying for. That’s his reported result, not an audited one, but the method costs nothing to copy.
Here’s how a small store uses it. Read the answers once a week. Pair them with free session recordings from Microsoft Clarity on the page your funnel flagged, and watch ten mobile sessions that viewed a product and left. When the same objection shows up in both places, fix it, then compare the step rate for the four weeks before and after. That comparison isn’t proof the way a controlled test is, because seasons and promotions move numbers too, so change one thing at a time and note anything else that changed that month.
If your list of test ideas keeps growing and nothing ships, the problem is usually ownership, not capacity. The piece on why the “we should test that” list never gets shorter covers giving each idea an owner and a date.
Shopify Rollouts is enough for most stores that want to test a theme, checkout, discount or catalog change, while a dedicated testing app earns its fee once you want shipping rate tests, offer tests or detailed revenue per visitor reporting across many variants.
Per Shopify’s Rollouts requirements, checked November 2026, rollouts are available on the Basic plan or higher, and Experiments, the version that splits traffic between a control and a treatment, need the Grow plan or higher. Experiments can include catalog changes, which means you can test prices natively; Shopify warns that during the test your ads, product feeds and other channels may show a different price from the one a visitor sees. Rollouts can’t change Liquid templates or run on vintage themes, and their analytics count only your online store, cart and checkout. If you’re deciding whether Experiments alone justifies an upgrade, the Shopify pricing guide lays out when each plan upgrade pays for itself.
Jeffrey was generous about the native option. “I think it’s a great tool,” he said, describing Rollouts as built around releasing a change safely, while ABConvert is built for running testing as an ongoing program. That’s a fair way to split the decision. All prices below are in US dollars from each app’s Shopify App Store listing in November 2026.
Choose Shopify Rollouts if you’re on Grow or higher, under about $250K a month, and the change you want to test is a theme, a discount, a checkout configuration or a catalog price. It costs nothing extra and keeps you from adding another app.
Choose ABConvert if you’re past roughly $100K a month and want price, shipping and offer tests on one platform. Its $99 Starter plan covers theme, template, URL and checkout tests; price, shipping and offer tests start on the $199 Growth plan with 5,000 test orders, and the listing notes shipping tests need a store with Carrier Service API access.
Choose Intelligems if you want to test one lever at a time and pay for that lever only. Content testing starts at $69 a month, discount and bundle testing or free shipping threshold testing at $112.10 a month each, and price testing sits in the $349 Unlimited plan. Pricing scales with your order volume.
Choose Shoplift if your tests are mostly page and theme layouts, such as product pages, collections and landing pages. Its Core plan lists at $99 a month, priced by monthly unique visitors.
Skip all three if you’re under about $50K a month. Look back at the table: the app would be measuring changes your traffic can’t detect.
Shipping tests are worth running before design tweaks because shipping cost is a real money decision for the shopper, and Jeffrey’s experience across his customers is that “generally things related to shipping have a higher success rate.” Judge the result on basket size, purchase rate and margin together, not on one of them.
I’ve felt this as a customer. On episode 493 I described buying bar soap from Duke Cannon here in Canada: products around $7 to $9 each, and $19 to ship. That’s two or three bars of soap just in shipping, so I add eight or nine to the cart to get the shipping waived. “Do I really want nine bars of soap? Probably not.” But the math in my head says it beats paying $19 to ship three.
Jeffrey’s advice was to set the threshold from your own orders, not a competitor’s. Export your orders from Shopify and look at how basket sizes cluster. If your best seller is $60 and nothing in the catalog costs under $20, a $75 threshold and a $100 threshold both force a second item, so testing one against the other tells you nothing. Pick the threshold that separates your one item buyers from your two item buyers, then test it.
Before you raise a threshold, check what each extra order earns after shipping and fees. The guide to ecommerce accounting on Shopify covers contribution margin per order if you haven’t worked it out yet.
What to change depends on your traffic: Starting stores fix obvious leaks from customer feedback, Growing stores fix the weakest step and test add to cart, Scaling stores add a testing program, and Leading stores test prices and personalization. At every stage, take something out before you put something in.
Starting, under about $10K a month. Add the thank you page question this week. Put shipping cost, delivery date and returns next to the add to cart button. Don’t install a testing app or a paid heatmap tool; Clarity is free.
Growing, about $10K to $100K a month. Pull the four step rates every month and work only on the step furthest below target. Test add to cart with Rollouts Experiments if you’re on Grow. Match paid traffic to its landing page, because the paid traffic landing page audit shows how often ad clicks land on a page making a different promise. Before you scale the spend behind those clicks, the guide to where your next ad dollar should go sets the most a new customer can cost you.
Scaling, about $100K to $500K a month. This is where a dedicated testing app makes sense, starting with shipping and threshold tests. Give the testing calendar one owner.
Leading, $500K a month and up. You can detect single digit lifts, so price tests and personalized merchandising become fair game. Personalized product ranking is its own decision, covered in what Shopify already does for AI personalization before you buy a tool.
The subtraction pass applies at every stage. The pattern I watched most often during my years at Shopify was premature complexity: stores between $500K and $2M a year stacking signup overlays, upsell widgets and badges on the product page until the add to cart button sat below the fold on a phone. Every block on that page should answer a question a shopper actually asks. If it doesn’t, remove it and watch the add to cart rate. A lot of conversion work is deleting things.
Jeffrey Lin, Anthony Morgan and Matt Stafford were guests on the eCommerce Fastlane podcast, and ABConvert is Jeffrey’s company. Names on this page may link through /recommends/ links that can earn a commission.
A good Shopify conversion rate is one that is rising against your own last quarter, because published averages mix stores with very different prices, categories and traffic. A $30 impulse product and a $900 sofa will never convert alike. A more useful check is the four step rates behind the headline number. Enavi’s targets, which Anthony Morgan shared on the eCommerce Fastlane podcast, are 50 to 70 percent of visitors viewing a product, 12 to 18 percent of product viewers adding to cart, 40 percent of carts starting checkout, and at least 45 percent of checkouts completing. If one step sits well below its target, that step is where your conversion rate is being lost.
Find out why a Shopify store has traffic but no sales by splitting the funnel into four steps and checking which one breaks. In GA4, count sessions with a product view, an add to cart, a begin checkout and a purchase, then divide each by the step before it. Then split the weakest step by device and traffic source. If visitors never reach a product page, the problem is the landing page or the traffic itself. If they reach product pages and don’t add to cart, the product page is missing an answer, often shipping cost, delivery time or sizing. Ask recent buyers what nearly stopped them to confirm what that missing answer is.
You need roughly 42,000 sessions to detect a 20 percent lift in a 2 percent purchase conversion rate at 95 percent confidence and 80 percent power, split evenly between two versions. That is a little over a month of traffic for a store at $50K to $60K a month with an $80 average order. Smaller lifts need far more: detecting a 10 percent lift takes about 160,000 sessions. Testing add to cart rate instead needs much less traffic because the rate is higher, around 7,700 sessions for a 20 percent lift on a 10 percent rate. Use a sample size calculator before you launch, and never stop a test early because one version looks ahead.
Shopify Rollouts is good enough for testing theme, checkout, discount and catalog changes, including catalog prices, for most stores on the Grow plan or higher, which is where its Experiments feature becomes available. It splits visitors between a control and a treatment and reports the comparison in rollout analytics. It does not change Liquid templates, does not run on vintage themes, and only measures your online store, cart and checkout. A dedicated app such as ABConvert, Intelligems or Shoplift makes more sense once you are past roughly $100K a month and want shipping rate, offer or many variant tests reported against revenue per visitor.
Most Shopify stores should not start with the checkout, because the Shopify checkout already converts well for shoppers who reach it. Check the step first: divide sessions that completed checkout by sessions that began checkout. If at least 45 percent finish, put your effort into product pages and the cart. If fewer finish, the usual causes are surprise shipping costs, unclear delivery dates, and duties or taxes added at the last step for international orders. Those are mostly fixed before checkout, by showing total cost and delivery dates on the product page and cart, and by collecting duties up front for cross-border orders.