Ecommerce Conversion Rate Optimisation That Shows in ROAS

To raise your conversion rate from Google Ads, fix what happens after the click, in this order. First, check that conversions and order amounts are real. Then find out whether you lose sales on your site or in the auction, and compare your rate with stores in your category. Only then fix pages and checkout, starting with your top products by ad spend, and judge each change on equal periods. In GetProfit data, ROAS moved with conversion rate and average order value far more than with cost per click.

What is conversion rate optimisation for a store on Google Ads?

Conversion rate optimisation (CRO) is work on your site that gets more of the shoppers who arrive from ads to place an order. For a store on Google Ads, the number to move is the conversion rate: the share of ad clicks that end in an order.

Google Ads divides conversions by the ad interactions it can track to a conversion over the same period. In Google’s example, 50 conversions from 1,000 interactions make a conversion rate of 5% (Conversion rate: Definition). The same page notes that the rate can go above 100% if you track several conversion actions or count every conversion.

Shopify’s guide uses another base: it divides orders by visits to your website (Ecommerce Conversion Rate: Benchmarks & Tips). That is why your store’s own figure and your Google Ads figure rarely match. The conversion rate formula shows which one to use and what one extra percentage point is worth.

Conversion rate is one of three numbers behind ROAS. The other two are average order value and cost per click:

ROAS = conversion rate × average order value ÷ cost per click

Example store, not client data.

A tableware store gets 8,000 clicks a month at 5 per click. It converts 3.75% of them into 300 orders, with an average order value of 600. Its ROAS is 3.75% × 600 ÷ 5 = 4.5. Lift the conversion rate to 4.25% on the same clicks, and the store gets 340 orders, 204,000 in revenue and a ROAS of 5.1. Or it can keep ROAS at 4.5 and pay up to 5.67 per click, which leaves room to bid for more traffic.

So a better page raises both ROAS and the cost per click you can afford. Your margin sets the ROAS you need to break even: see ecommerce unit economics.

In our data, ROAS falls with conversion rate and average order value, not with cost per click

We looked at 1,360 store-months of GetProfit data from June 2025 to June 2026. We measured each store’s monthly change against the median store in the same month, which removes the effect of season and holidays. Within a store, ROAS moved with conversion rate (correlation +0.527) and with average order value (+0.540). It barely moved with cost per click (−0.100).

The bad months show the same picture. We took the 285 months in which a store’s ROAS fell by more than 20% against other stores. In those months, the median change against other stores was −19.5% in conversion rate and −19.4% in average order value. Cost per click rose a median of only 3.4%, and spend barely changed (−1.3%).

These are observations, not an experiment: they show what moves together, not what causes what. Our data comes from Google Ads: clicks, orders and order amounts. What shoppers do on your pages lies outside it, so our advice on pages relies on Google’s requirements and on published usability research.

So when ROAS drops, look first at what happens after the click. To decide whether to work on average order value or conversion rate first, see how to increase ROAS.

In what order should you work?

Page and checkout fixes matter. But when you pay for traffic, start with the numbers your ads learn on.

Step 1: can you trust the conversion rate you are optimising?

Your conversion rate in Google Ads is only as accurate as your conversion tracking. Google splits conversion actions into primary and secondary ones. A primary conversion action appears in the “Conversions” column, and Google uses it for bidding as long as its goal is used for bidding. Secondary actions are for observation only (About primary and secondary conversion actions). Google warns that setting these up wrongly can stop Smart Bidding from optimising effectively.

If an add-to-cart or a page view is primary, your conversion rate looks high, and the campaigns learn to find carts instead of orders. If two goals count the same order, the rate comes out too high. If every order carries one fixed amount, the conversion value and ROAS mean nothing, whatever the page does.

The portal’s Conversion check looks at four things:

Two of the rules behind these checks are simple. If one product’s conversion value in one day is above 30 times the account’s typical order value, the portal treats it as an anomaly. If a campaign with 10 or more conversions has an average value per conversion between 0.8 and 1.2, the portal treats the event as broken. That pattern usually means a micro-conversion or a placeholder value of 1, not an order. The conversion tracking audit walks through these checks step by step.

Step 2: is the problem on your site or in the auction?

ROAS splits into three numbers, so a drop has three possible causes. Compare each one with your own previous months.

Keep in mind where Performance Max sends shoppers. By default, final URL expansion is on in every Performance Max campaign. It can replace your final URL with a landing page Google judges more relevant (Performance Max Optimizations). So part of your paid traffic can land on pages you never set as final URLs.

To tell the site from the auction in one pass, see ROAS dropped: ads or website. If clicks keep coming but orders do not, start with traffic but no sales.

Step 3: is your conversion rate low for your category?

Across 114 stores with at least eight months of GetProfit data (June 2025 – June 2026), the conversion rate from product ad clicks was 1.9% in the bottom quarter, 2.7% at the median and 4.0% in the top quarter.

A higher conversion rate does not guarantee a higher ROAS. When we split the same stores by ROAS, the top quarter converted at 2.75%, below the second quarter’s 3.13%. Those stores got their ROAS from something other than conversion rate. To check whether your rate is normal, see average ecommerce conversion rate.

Category matters more than any global figure. In our study of 1.4 million products, the conversion rate ran from 1.21% for furniture and 1.40% for car parts to 4.93% for electronics and 5.78% for pet supplies. The study used a separate sample of 130+ stores over 13 months, so its median store converts at a different rate too: 2.29%. Find your category in conversion rate by category.

Read your own rate over three months or more. In a small store, one week holds too few orders to tell a trend from chance.

Step 4: which pages should you fix first?

Fix the pages that take most of your ad spend. In the median store of our study of 1.4 million products, the top 1% of products took 27.7% of ad spend and the top 10% took 68.1%. Improve those product pages, and you touch most of your ad spend at once. A page that gets a few clicks a month barely registers in ROAS, however well you polish it.

Start with a list of products sorted by spend over the last three months. Then add any other pages Performance Max sends traffic to. To rank hundreds or thousands of pages by the money at stake, see CRO priorities by ad spend.

Step 5: what is broken on your top pages and in checkout?

On the pages that take most of your spend, check three places in turn: the product page, the path from cart to order, and loading speed.

What does a product page need for a shopper from a Shopping ad?

First, it has to match the ad. Merchant Center checks your product data against the landing page. Pages that miss the minimum requirements can get your products or your whole account disapproved (About landing page requirements). The same page says that shoppers are more likely to leave without buying when a landing page does not match the ad.

Then the page has to answer the shopper’s questions. Baymard Institute benchmarked the product pages of 343 leading US and European ecommerce sites and rated 51% of them “mediocre” or worse (Product Details Page UX Research Studies). Its guidelines cover images, the buy section, shipping and returns, variations, descriptions, specifications and reviews. Baymard sells access to this research, so read it as a vendor’s benchmark. Product page optimisation covers what a page needs beyond the feed and which parts Google already reads from it.

Across a large catalogue, descriptions are hard to keep up by hand. For templates that cover thousands of products, see how to write product descriptions.

Where do shoppers drop off between the cart and the order?

Across 50 studies, Baymard puts the average cart abandonment rate at 70.22% (50 Cart Abandonment Rate Statistics 2026). In its survey of US online shoppers, 42% had abandoned a cart because they were just browsing. The public page does not state the sample size. These are the most common other reasons:

Even large sites have room to improve checkout. In Baymard’s benchmark of the checkout flows of 343 top-grossing US and EU sites, 65% performed “mediocre” or worse (E-Commerce Cart & Checkout Usability Research). Its abandonment page adds that the average US checkout shows 23.48 form elements by default, while an ideal flow can be as short as 12–14. Baymard estimates that the average large ecommerce site could gain a 35.26% higher conversion rate from better checkout design. That is an estimate for large US and EU sites, not a promise for yours.

To see the step where your own shoppers leave, measure the funnel in GA4. Google’s guide to ecommerce measurement lists events such as view_item, add_to_cart, begin_checkout, add_shipping_info, add_payment_info and purchase (Measure ecommerce). For the checkout fixes worth doing first, see cart abandonment.

Does site speed matter for sales from ads?

Speed affects whether a shopper waits for the page you paid to show them. Google’s Search Central guidance gives three Core Web Vitals targets (Understanding Core Web Vitals and Google search results):

• Largest Contentful Paint within the first 2.5 seconds of loading;
• Interaction to Next Paint below 200 milliseconds;
• Cumulative Layout Shift below 0.1.

Merchant Center adds its own reason: redirects add time between the click and the page, and long loading can stop Google from reading your product information. To find which speed problems cost you sales from ads, see site speed for online stores.

Step 6: how do you judge a change without an A/B test?

A small store rarely has enough traffic to split it in two. Its own numbers also swing widely from month to month. In GetProfit data on 110 stores with at least eight months of history (June 2025 – June 2026), a store’s monthly ROAS typically sat 16% away from its own median. At the median, a store’s best and worst months were 3.1× apart.

So a 10% change after a page update can be ordinary noise. Compare equal periods before and after the change, on the products the change touched. Read conversion rate and average order value separately, and avoid windows that cross a seasonal peak. To find out what a small store can test, and how to judge a change without a test, see A/B testing for small stores.

How do you keep the ads working through site changes?

A theme update, a new plugin or a rollback can quietly break the tag that sends orders to Google Ads, change how prices show or hide stock status. Merchant Center then sees a page that no longer matches the feed. Its landing page requirements recommend putting price and availability in the page’s initial HTML with schema.org structured data. They warn that markup generated by JavaScript can make Shopping crawls less frequent and less reliable.

After every update, run the checks in the website update checklist. A move to a new platform or a redesign needs its own plan: replatforming without losing Shopping.

Before the first paid click: is your site ready?

If your store is new or about to start ads, you can check the site in an afternoon:

1. Place a test order with a real value. One purchase should arrive in Google Ads exactly once, with the real order amount.
2. Open five product links from your feed. Each should land on the product, with the same price, currency, availability and pre-selected variant.
3. Show the delivery cost and time on the product page. Extra costs are the top reason shoppers abandon carts in Baymard’s survey.
4. Allow checkout without an account. Forced sign-up is a reason for 18% of shoppers in the same survey.
5. Complete a purchase on a phone. Errors and crashes are a reason for 17%.
6. Check the speed of your top product pages against the three Core Web Vitals targets.

A conversion rate optimisation checklist for this month

1. Audit the conversions. One primary purchase goal, the real order amount, one goal per order, no amounts many times above your average order value.
2. Write down three numbers for the last three months: conversion rate, average order value and cost per click. Mark which one moved.
3. Compare your conversion rate with stores in your category, not with a global average.
4. List the products that took most of the spend and the pages Performance Max sends traffic to.
5. Check those pages against Merchant Center’s landing page requirements and best practices: a link to the product page itself, matching price and availability, a pre-selected variant, nothing covering the buy button.
6. Fix the checkout problems you can see: delivery cost up front, guest checkout, fewer fields, a test order on a phone.
7. Judge each change on equal periods, by conversion rate and average order value separately, and keep a log of what changed and when.

Which part of your ad account loses points?

The account score in the portal shows which part of the account costs you points. It covers four areas: the data Google receives, products, campaign structure and the history of changes. The data area is worth 17 of the 100 points. When conversion tracking is broken, the portal also sets the parts of the score that depend on conversions to zero, so the score drops instead of flattering the account.

What state your ad account is in. One score instead of a dozen tabs of figures — plus a breakdown of exactly where the money leaks. The portal changes nothing without your consent.

Get your account score →

Sources

• Conversion rate: Definition — how Google Ads calculates conversion rate, the 50 ÷ 1,000 = 5% example, rates above 100% with several actions or “Every” counting. Checked 2 October 2026.
• About primary and secondary conversion actions — primary actions appear in “Conversions” and are used for bidding when their goal is, secondary actions are for observation; misconfiguration can stop Smart Bidding from optimising effectively. Checked 2 October 2026.
• About landing page requirements — product data checked against the landing page; disapproval of products or the account; specific product page, matching price, currency and availability, no pop-ups over key information (requirements); pre-selected variant, few redirects, structured data in the initial HTML (best practices); a mismatched page makes people more likely to leave. Checked 2 October 2026.
• Performance Max Optimizations — final URL expansion is on by default in Performance Max and can replace the final URL with a more relevant landing page. Checked 2 October 2026.
• Measure ecommerce — ecommerce events a typical GA4 implementation measures, from view_item to purchase. Checked 2 October 2026.
• Understanding Core Web Vitals and Google search results — LCP within 2.5 seconds, INP below 200 milliseconds, CLS below 0.1. Checked 2 October 2026.
• Ecommerce Conversion Rate: Benchmarks & Tips — Shopify defines ecommerce conversion rate as orders divided by visits to your website. Checked 2 October 2026.
• Product Details Page UX Research Studies — product pages of 343 leading US and European sites benchmarked, 51% rated “mediocre” or worse; guideline topics. Vendor research. Checked 2 October 2026.
• 50 Cart Abandonment Rate Statistics 2026 — 70.22% average of 50 studies; 42% “just browsing”; reasons for abandonment among US online shoppers; 23.48 form elements in the average US checkout, while an ideal flow can be as short as 12–14; 35.26% conversion rate potential for the average large site. Vendor research, sample size not stated on the page. Checked 2 October 2026.
• E-Commerce Cart & Checkout Usability Research — checkout flows of 343 top-grossing US and EU sites benchmarked, 65% “mediocre” or worse. Vendor research. Checked 2 October 2026.
• GetProfit data: 1,360 store-months, June 2025 – June 2026, each measured against the median store in the same month — how ROAS moved with conversion rate, average order value, cost per click and spend; 285 months with ROAS down more than 20%.
• GetProfit data: 114 stores with at least eight months, June 2025 – June 2026 — conversion rate quartiles from product ad clicks; conversion rate by ROAS quarter.
• GetProfit data: 110 stores with at least eight months, June 2025 – June 2026 — typical monthly deviation of ROAS from a store’s own median; best month to worst.
• GetProfit study of 1,404,808 products in 130+ stores over 13 months — conversion rate by category, store median conversion rate, share of spend taken by the top 1% and 10% of products.
• GetProfit portal methodology — the four conversion checks, the 30× typical order value anomaly rule per product-day, the 0.8–1.2 value-per-conversion rule per campaign from 10 conversions, the 17-point data area in the account score.