Is your website losing sales? A timed one-hour check with GA4, Microsoft Clarity and PageSpeed Insights finds the biggest leaks and prices each one in dollars.
Your store is probably losing sales at one or two steps, and one hour with GA4, Microsoft Clarity and PageSpeed Insights shows which. Contentsquare's 2026 benchmark found 35.2% of sessions affected by friction, across 99 billion sessions on 6,500+ sites. The question for your store is which leak is biggest in dollars, and that takes a funnel, a device split and a few recordings.
This is the do-it-yourself version of the free teardown we record for stores. It runs on a timer because the point is a ranked list of three leaks by the end of the hour, priced in dollars a month. Each step says what to open, what to write down and what the numbers mean.
How can you tell if your website is losing sales?
Your website is losing sales when one step of the path to purchase loses far more shoppers than the same step on another device, another traffic source or last year's version of your store. A low overall conversion rate is a symptom. The step-by-device comparison tells you where to look.
Signs worth the hour:
- Traffic is flat or rising and orders are flat or falling over the same weeks, compared with the same weeks last year.
- Phone shoppers convert at less than half the desktop rate. On Contentsquare's benchmark, mobile converts about 41% below desktop (2.0% against 3.4%), so a phone rate under half of desktop is a wider gap than average.
- Shoppers add to cart and stop. Baymard's survey puts the top reason for abandoning a checkout at extra costs shown too late (40% of US shoppers who abandoned one).
- GA4 records noticeably fewer purchases than Shopify has online orders, or the gap between the two changed suddenly.
- A theme update, new app or checkout change went live and conversion moved the same week.
Any one of these justifies the hour. If none applies, the hour still gives you a baseline to compare against next quarter.
What do you need before you start?
You need admin access to Shopify and GA4, Microsoft Clarity installed for at least a week, and a spreadsheet. GA4 should be connected through the Google & YouTube app, which Google calls the strongly preferred setup for Shopify.
Clarity needs data before it can show you anything. If it isn't installed yet, install it today (the Microsoft Clarity: AI Insights app takes a few minutes and costs nothing) and run this check in a week. Our Clarity on Shopify guide covers setup, masking and consent.
Pick one date range and use it everywhere: the last 28 complete days. That matches the window PageSpeed Insights uses for its field data, and it avoids comparing today's half-processed numbers. Google says GA4 can take 24 to 48 hours to process data.
Minutes 0 to 10: is your tracking telling the truth?
Your tracking is good enough for this check if GA4's purchases sit a little below Shopify's online store orders and that gap is steady from week to week. If GA4 shows more purchases than Shopify, or far fewer, fix tracking before you trust any funnel.
- In Shopify, open Analytics, then Reports, and note online store orders for the 28 days. Leave out point-of-sale and draft orders, which never pass through a browser checkout.
- In GA4, open Reports, then Monetisation, then Ecommerce purchases, and note purchases for the same dates.
- Divide GA4 purchases by Shopify orders. Write the ratio down. Repeat for each of the four weeks.
Read the ratio this way:
| What you see | What it usually means | Next step |
|---|---|---|
| GA4 a little below Shopify, steady each week | Normal browser-side loss from ad blockers, declined cookies and pages that didn't load | Carry on with the check |
| GA4 above Shopify | A purchase tag firing twice, usually old code left in the theme next to the app | Fix before reading the funnel |
| GA4 far below Shopify, or the ratio jumped in one week | A tag removed, a consent setting changed or a checkout change broke the event | Fix before reading the funnel |
| GA4 revenue lower than Shopify total sales, orders close | The app's purchase value excludes shipping and tax, per Google's event reference | Compare orders, not revenue |
Neither Shopify nor Google publishes a normal size for the gap. Shopify's help centre says only that the two services use different tracking mechanisms and won't record visitors equally. The useful test is stability: a ratio that holds is the cost of browser tracking, and a ratio that moves means something broke. Our GA4 on Shopify guide has the fix for each pattern.
Minutes 10 to 25: where do shoppers drop off, and on which device?
GA4's purchase journey report shows where shoppers drop off: it counts users at session start, product view, add to cart, begin checkout and purchase, and it splits each step by device category by default (Google's report guide). Open Reports, then Monetisation, then Purchase journey, set the 28 days, and copy the device rows into your spreadsheet.
Work out the step rates for each device: the users at each step divided by the users at the step before. A worked example for a store doing about AU$2M a year, with figures we made up to show the method:
| Step | Desktop users | Desktop step rate | Mobile users | Mobile step rate |
|---|---|---|---|---|
| Session start | 20,000 | 46,000 | ||
| View product | 11,000 | 55% | 23,000 | 50% |
| Add to cart | 1,650 | 15% | 2,990 | 13% |
| Begin checkout | 1,155 | 70% | 1,346 | 45% |
| Purchase | 693 | 60% | 646 | 48% |
In this example the phone leak sits between cart and checkout: 45% of phone carts reach checkout against 70% on desktop. The product page step differs by a few points, which is normal. The two big gaps are cart to checkout, then checkout to purchase. If the product page step is your gap, start with our product page guide.
Settings that change what the report shows:
- The report is a closed funnel: a user counts only if they passed through the steps in order. Google's toggle for an open funnel helps when shoppers land straight on a product page from an ad.
- Counts are users, while Shopify counts sessions, so GA4 and Shopify step rates won't match. Our post on Shopify conversion rate explains both definitions.
- The checkout journey report (also under Monetisation) splits checkout into shipping, payment and purchase. Google's own page says it needs the add_shipping_info event; Google's newer Shopify reference says the app sends it only on checkouts using Checkout Extensibility. Check the event appears in your property before you rely on that report.
Shopify's own conversion rate breakdown report (Analytics, Reports, Behaviour) gives a second view in sessions: sessions with cart additions, sessions that reached checkout and sessions that completed it. If both tools show the same weak step, you've found the leak.
Write down the worst step on the worst device. That is the one you watch in Clarity next.
Minutes 25 to 45: what are shoppers doing on the worst step?
Clarity recordings show what shoppers did on the step that loses them, and Clarity's filters find the sessions with frustration signals so you don't watch at random. Open Recordings, filter by device (mobile, in the example), by visited URL (the cart page, in the example) and by a frustration signal.
Microsoft defines the signals in its Clarity documentation:
| Signal | Clarity's definition | What it often means on a store |
|---|---|---|
| Rage click | Several clicks in a small area in quick succession | A button that looks clickable but responds slowly or not at all |
| Dead click | A click with no visual change and no navigation | A product image that doesn't zoom, an underlined label that isn't a link |
| Quick back | Leaving a page for the previous one faster than the usual dwell time | A product page that didn't match what the ad or collection promised |
| Excessive scrolling | More vertical scrolling than the expected average | Shoppers hunting for size, delivery or price information |
| JavaScript error | An error while the browser ran a script | A broken app or theme script; click errors are errors that follow a click |
Watch 10 to 15 recordings of the worst step with 20 minutes on the clock. For each one, write a single line: where the shopper stopped and what they did immediately before. When the same field, button or hesitation over a shipping price appears in three sessions, you have a pattern worth pricing.
Contentsquare's retail data gives a sense of scale: rage clicks appeared in 5.3% of retail sessions in 2025, and JavaScript errors caused frustration in 17.8%. If your filtered set shows far more, the signal is the leak.
Limits that change what you conclude:
- Clarity records checkout pages only on Shopify Plus. On other plans the recording ends when the shopper enters checkout, so a checkout leak shows up as a funnel drop in GA4 with no recording behind it.
- A recording is rebuilt from page data, not filmed from the shopper's screen. One strange replay may be a reconstruction fault; the same failure on the same device in five sessions is a finding.
Minutes 45 to 55: is speed costing you sales?
Speed is costing you sales if PageSpeed Insights shows your product and collection pages failing Core Web Vitals for real phone users. Run the two templates that carry most of your traffic through PageSpeed Insights on mobile and read the top section, which is field data from real Chrome users over the previous 28 days, reported at the 75th percentile.
| Metric | Good, per Google | What a fail looks like to a shopper |
|---|---|---|
| Largest Contentful Paint (LCP) | 2.5 seconds or less | The product image arrives late |
| Interaction to Next Paint (INP) | 200 milliseconds or less | Tapping a size or add to cart feels stuck |
| Cumulative Layout Shift (CLS) | 0.1 or less | The page jumps and the tap lands on the wrong thing |
Shopify's 2026 analysis of its own stores found that for every 100 milliseconds slower a store loads, conversion tends to be about 3.5% lower, and that stores at 2.5 seconds LCP reported roughly 30% lower conversion than stores at 1.5 seconds. Shopify calls it a correlation, not proof of cause, so use it to rank speed against the other leaks rather than to forecast a lift. If speed does rank first, our guide to speeding up a slow website covers the fixes.
Layout shift is the metric owners miss. When we rebuilt The Flower Room's storefront, the share of phone page loads with a stable layout went from 31% to 93% (Chrome's field data, February to April 2026 on the old platform against September to November 2026). Their load speed was already good on both platforms, so layout stability was the speed problem worth fixing, and a load-time check alone would have missed it.
If the field data section says there isn't enough data, the page has too little Chrome traffic for Google to report. Use the origin result for the whole site, which PageSpeed Insights shows when it has it.
Minutes 55 to 60: which three leaks cost the most?
The leak that costs the most is the one with the largest sessions × missing conversion × average order value. Write each leak you found as one line in that formula, then sort by dollars.
Monthly cost of a leak = sessions reaching the leaky step × the conversion rate you are missing × average order value
GA4's funnel counts users rather than sessions; either works if you use the same unit for every leak. Continuing the worked example, where every input is an assumption chosen to show the arithmetic:
- The store has 20,000 desktop users a month converting at 3.5% (693 purchases) and 46,000 mobile users at 1.4% (646 purchases). Average order value is AU$130, so the store turns over about AU$2M a year.
- First, size the whole phone gap. At Contentsquare's average distance below desktop, mobile would convert at 3.5% ÷ 1.74 = 2.0%. The missing rate is 2.0% − 1.4% = 0.6 points, and 46,000 × 0.6% × AU$130 = AU$35,880 a month. That is the outer limit for every phone leak combined.
- Leak 1, cart to checkout on mobile. 2,990 phone carts, 45% reaching checkout against 70% on desktop: 25 points missing. Those extra checkouts would complete at the phone checkout rate of 48%. 2,990 × 25% × 48% × AU$130 = AU$46,644 a month if phones matched desktop at this step. Matching desktop here would put phones above Contentsquare's average gap, so cap the leak at the whole phone gap, AU$35,880.
- Leak 2, a dead click on the product gallery seen in 4 of 12 recordings. Without a measured conversion gap, size it from the affected users: if a third of 23,000 phone product viewers hit it and 1 in 200 of them would otherwise have bought, 7,667 × 0.5% × AU$130 = AU$4,983 a month.
Count overlapping leaks once, in the place where you'd fix them. Each figure is a ceiling: no change closes a gap completely, and you should expect to win back part of it.
| Rank | Leak | Upper-bound cost a month | Where you saw it |
|---|---|---|---|
| 1 | Phone cart to checkout | AU$35,880 (capped at the whole phone gap) | GA4 purchase journey, Clarity rage clicks on the cart |
| 2 | Phone product gallery dead click | AU$4,983 | Clarity dead clicks on product pages |
| 3 | Phone LCP of 3.4 seconds on collection pages (assumed for the example) | Not priced: correlation only | PageSpeed Insights field data |
You now have a ranked list, which is what the hour was for. Keep the spreadsheet: the ratios and step rates you wrote down are the baseline any fix gets measured against.
Which leaks are most common on Australian stores?
Delivery and checkout length are the leaks Australian research points to most. Check these first when your worst step is the cart or the checkout.
- Delivery information. Australia Post's 2026 eCommerce Report says 70% of Australians are less likely to buy when delivery communication at checkout is poor (Omnibus Survey, November 2025). A cart that shows no delivery estimate or cost is the first suspect.
- Delivery time. In the same report, THE ICONIC says it observes "a 3-10% decrease in conversion for every additional day added to the delivery time shown at checkout", depending on product and day of the week.
- Checkout length. The same report cites Shopify research finding 52% of Australians abandoned a cart because checkout was too long or complicated, rising to 67% for higher spenders.
- Costs revealed late. Extra costs top Baymard's list of reasons for abandoning a checkout at 40%, followed by slow delivery at 20% and not trusting the site with card details at 19% (Baymard).
Our guide to why people add to cart but don't buy works through each of these causes and how to confirm it on your own store.
Where does the one-hour check stop working?
The one-hour check stops working at the fix. It tells you which step leaks and roughly what that costs; it doesn't tell you which change will close the gap, and it can't prove a change worked.
- Reading 15 recordings finds a pattern. Reading 200 and turning them into ranked, built fixes takes days, and most of that time is the build.
- Fixes on the cart, product page or checkout need design and theme code. On Basic, Grow and Advanced plans, apps can change only the thank-you and order status pages of checkout; changes to the information, shipping and payment steps need Shopify Plus.
- A fix needs a baseline written down before it ships and the same weeks compared afterwards, so a seasonal rise doesn't get mistaken for the fix working.
- If the tracking check failed, every number after it is suspect until the tags are repaired.
On Paperlake, a store we run, the leak sat between choosing a book and paying for it. The fix was a rebuilt order flow that opens payment on the price screen, and Stripe checkout sessions that paid went from 25% to 50% (23 March to 18 June against 1 October to 17 November 2026; case study). Seeing the leak was quick; the rebuild and the measurement took weeks.
Stores under about AU$1.5M a year online, or under 20,000 visits a month, usually do best running this check themselves each quarter and fixing the top leak in-house. Our homepage says the same: at that size you need more traffic before fixes can be measured.
Sources
- Contentsquare 2026 Digital Experience Benchmark: frustration - 35.2% of sessions affected by friction; 99 billion sessions, 6,500+ sites, Q4 2024 to Q4 2025
- Contentsquare 2026 Digital Experience Benchmark: conversions - desktop conversion 74% higher than mobile; mobile 69.9% of traffic
- Contentsquare retail digital experience: frustration - rage clicks in 5.3% of retail sessions, JavaScript errors in 17.8% (2025)
- Google Analytics Help: purchase journey report - steps, closed funnel, device category breakdown
- Google Analytics Help: Shopify setup - the Google & YouTube app as the preferred setup
- Google tag platform: Shopify event parameters - purchase value excludes shipping and tax; add_shipping_info on Checkout Extensibility checkouts
- Google Analytics Help: data freshness - 24 to 48 hours processing
- Shopify Help Center: analytics discrepancies - why Shopify and third-party numbers differ
- Microsoft Clarity: semantic metrics - definitions of rage clicks, dead clicks, quick backs, excessive scrolling, errors
- Microsoft Clarity: Shopify integration - checkout tracking on Shopify Plus
- PageSpeed Insights: about and web.dev: Web Vitals - 28-day field data, 75th percentile, thresholds
- Shopify: Store speed and conversion (2026) - 100 ms slower, about 3.5% lower conversion; correlation
- Australia Post eCommerce Report 2026 - delivery communication 70%; checkout length 52% and 67%
- Baymard Institute: cart abandonment rate statistics - reasons for abandoning a checkout
- Shopify Help Center: customising checkout with apps - checkout step apps on Shopify Plus only
- The Flower Room case study and Paperlake case study - client figures with their windows
Questions
How do I know if my online store is losing sales?
Compare each step of your purchase funnel by device. In GA4's purchase journey report, a step where phone shoppers drop away much faster than desktop shoppers marks a leak. As a yardstick, Contentsquare's 2026 benchmark of 99 billion sessions found desktop converting 74% better than mobile; a much wider gap on your store points to a phone problem.
Why does my website get traffic but no sales?
Either the traffic doesn't intend to buy, or something on the site stops people who do. Split conversion by traffic source first: a single campaign of low-intent visitors can drag the whole store's rate down. If buyers from search and email also stall, find the step where they leave in GA4 and watch Microsoft Clarity recordings of that step. Contentsquare's 2026 benchmark found 35.2% of sessions hit friction such as rage clicks, slow pages or errors.
What free tools show where customers drop off?
GA4's purchase journey and checkout journey reports show which step loses shoppers, split by device. Microsoft Clarity, which is free with no traffic limits, shows what they did on that step through recordings, rage clicks, dead clicks and quick backs. PageSpeed Insights shows how fast the page loaded for real Chrome users over the last 28 days.
How do I work out how much a leak costs?
Multiply the sessions that reach the leaking step by the conversion rate you are missing and by your average order value. For example, 46,000 phone sessions a month converting at 1.4% instead of 2.0%, at a AU$130 average order, is up to AU$35,880 a month. Treat that as a ceiling, since no fix closes a gap completely.
When should I stop diagnosing and get help?
When the leak needs a design and build change you can't ship yourself, when your tracking is too broken to trust the funnel, or when you have read enough recordings to know there is a problem but not which fix to try first. Finding a leak takes an hour; fixing it and proving the fix worked takes a baseline, a build and several weeks of measurement.



