Nobody actually wants a higher conversion rate. What merchants want is for the traffic they already have to produce more orders, which is the same sentence with the accounting stripped out of it. Conversion rate is just the receipt.
We fund Google Shopping ads for stores and bill a commission on the orders those ads produce, which means we pay for every click before anyone knows whether it converts. Whether we can do that for a given store is not a judgment call. It is a sum with three inputs, and conversion rate is the second of them.
It is also the input merchants get wrong most reliably, and they get it wrong in one direction. The number on your Shopify dashboard is not the number the sum needs. It is roughly 1.7 times too high, for a reason that is nobody's fault and has nothing to do with the quality of your store. That gap is why a store can look completely healthy on its own analytics and still come out marginal when we run our arithmetic on it, and closing that misunderstanding is most of what this post is for.
The Sum That Decides It
We publish this sum in full on the gradient intent model page, and it is short enough to hold in your head:
CPC ceiling = (order value x commission rate x conversion rate) / ROAS target
Read it left to right. An order is worth some amount. We earn a percentage of it. Only a fraction of clicks become orders. Divide by the return we need to make the funding work, and you get the most anyone can pay for a click to your store and still come out ahead.
Conversion rate is the third term, and it is doing something specific in there. The first two terms tell you what an order is worth to us. Conversion rate turns that into what a click is worth to us, which is the only number the auction cares about, because the auction sells clicks and not orders.
That is the whole reason conversion rate belongs in a post about funding rather than in a post about growth. It is the exchange rate between the thing you sell and the thing we buy.
Three Terms, and No Favorites
Because the three terms multiply, and nothing in the sum is added, subtracted or squared, a 20% improvement in any one of them produces exactly the same result. Raise order value by 20% and the ceiling moves 20%. Raise the commission rate by 20% and the ceiling moves 20%. Raise conversion rate by 20% and the ceiling moves 20%. Same number, three times.
We made that argument at length in the sibling to this post, how to raise average order value on Shopify, and nothing here contradicts it. The two posts are the same argument seen from two of its three terms. If you have not read that one, read it alongside this one, because the practical conclusion of both is that you are supposed to stack the terms rather than pick a favorite.
What differs is the character of the work. Order value is set by you, unilaterally, this afternoon. Conversion rate is slower, noisier, and only partly yours: you control the page, the speed, the photography, the shipping message and the checkout, and you do not control whether the shopper is on a train with one bar of signal, whether they already bought this from someone else last week, or whether they clicked to browse rather than to buy.
So this post is not an argument that conversion rate is the strongest lever. Arithmetically it cannot be, because all three are exactly as strong as each other. It is a post about the one term where merchants routinely feed the model a number that is wrong before any work starts.
Your Dashboard Number Is the Wrong Number
The conversion rate on your Shopify analytics is whole-site. Every session counts, whatever brought it. That figure contains people who typed your brand name into Google, people who arrived from your email list, people clicking a link in an order confirmation, direct visitors who bookmarked you, and returning customers who already trust you and already know their size.
That is the warmest traffic a store has. It is also, for most stores, a large share of total sessions, which means the whole-site number is heavily weighted toward shoppers who were never in any doubt.
A Google Shopping click is the opposite of that. It is a stranger who has not heard of you, who saw your product as one tile in a grid of eight, and who is comparison shopping by construction, because comparison is what the format is for. Several of the other tiles are going to get clicked too. That shopper converts at a lower rate than your site average, and no amount of good work on your part fully removes the difference, because the difference is the traffic and not the store.
The 0.6 Haircut, and Why We Apply It
Rather than argue about this store by store, our funding risk engine applies a fixed multiplier. When the conversion rate we have for a store is whole-site rather than ads-attributed, we multiply it by 0.6 before it goes anywhere near a decision.
So a store reading 3.0% site-wide is assessed at roughly 1.8% on the traffic we would send. A store reading 2.0% is assessed at 1.2%. A store reading 1.0% is assessed at 0.6%.
It is a blunt instrument and we would rather have your real ads-attributed rate, which is why the haircut only applies when we do not. It exists because the alternative is worse. Taking an unhaircut whole-site rate at face value systematically over-rates stores, which would mean promising funding stability we cannot deliver, and the merchant finds out about it later and in a more expensive way than a paragraph in a blog post.
Watch what it does to the arithmetic. Take a store with a $120 average order value, a 15% commission rate, and a 3.0% conversion rate on its own dashboard.
Using the dashboard number as-is: $120 times 15% is $18 of commission per order. At 3.0%, the average click is worth 3% of $18, which is $0.54. Divide by a 1.1 target and the ceiling is $0.49 a click.
Using the haircut number: 3.0% becomes 1.8%. The average click is now worth $0.32. Divide by the same 1.1 target and the ceiling is $0.29 a click.
Across our portfolio, the average Shopping click cost $0.23 over the 90 days to September 2026, measured across 92 stores and 40,951 clicks. So the first version of that store has more than twice the headroom it needs and looks like an easy yes. The second version clears the bar by about 28%, which is a real yes but a thin one, and one bad month of click prices takes it away.
Same store. Same day. The only difference is which conversion rate went into the sum. Because the ceiling is linear in that term, the error is exactly one divided by 0.6, which is about 1.67 times, every time. It is not a rounding difference. It is the difference between comfortable and marginal, and it is the single most common reason a merchant is surprised by our answer.
What a Realistic Target Looks Like
The useful way to think about conversion rate is as a ladder by traffic source rather than as one number. In almost every store we look at, the order runs the same way. Direct and branded search sit at the top. Email is next. Non-brand organic is below that. Paid non-brand traffic, which is what Shopping is, sits at the bottom.
We are not going to publish benchmark ranges by vertical, because we have not measured them across enough stores to stand behind them and the ones circulating online are mostly recycled from surveys nobody can trace. What we can publish is the relationship we actually use, which is the 0.6 above, and the arithmetic that turns it into a target.
Rearrange the sum and it will tell you what conversion rate you need rather than what you have. At our measured $0.23 click cost, a 1.1 target, and a 15% commission rate, the required conversion rate on paid Shopping traffic falls out of your order value alone. The third column is the whole-site figure that implies, which is the one you can check against your own dashboard this minute.
| Average order value | Paid Shopping rate needed | Whole-site rate that implies |
|---|---|---|
| $40 | 4.2% | 7.0% |
| $60 | 2.8% | 4.7% |
| $80 | 2.1% | 3.5% |
| $120 | 1.4% | 2.3% |
| $200 | 0.8% | 1.4% |
| $300 | 0.6% | 0.9% |
Two things are worth taking from that table. The first is that a $40 store needs a whole-site conversion rate of 7% to make funded Shopping work at a 15% commission rate, and very few stores in the world read 7% site-wide. That is not a conversion rate problem. That is an order value problem wearing a conversion rate costume, which is why the sibling post exists.
The second is that the column is a curve, not a cliff. Nothing changes character between $80 and $120. The product of order value and conversion rate is what matters, and there are many ways to reach the same product, which is the interchangeability from earlier showing up as a shape rather than as an assertion.
Six Levers, in Order of How Fast They Move
These are ordered by time to effect, not by size of effect. Some of the slowest ones are the largest. If you are trying to change the shape of your numbers before a decision gets made about your store, start at the top.
1. Page speed and Core Web Vitals, which move within days
Speed matters more for cold paid traffic than for anything else you have, and the reason is behavioral rather than technical. A shopper who typed your brand name will wait for your site. A shopper who clicked one tile out of a grid of eight will not, because seven other tiles are one back button away and they are all still open in their head.
The three metrics to look at are largest contentful paint, which should be under 2.5 seconds, interaction to next paint, which should be under 200 milliseconds, and cumulative layout shift, which should be under 0.1. Measure them on mobile, and measure them on your best selling product page rather than your homepage, because a Shopping click lands on a product page and never sees your homepage at all.
In practice the damage is almost always the same short list. Apps injecting scripts on every page. Hero images shipped at three times the pixels anyone will see. Carousels. A chat widget loading before the buy button. Review widgets that block rendering. And, most commonly of all, script tags left behind in the theme by apps that were uninstalled months ago, which cost you speed while providing nothing at all.
There is a second effect worth knowing about. Landing page experience is an input to how Google ranks your ad, so a slow product page charges you twice: once when the shopper leaves, and once in the auction. We are not going to put a number on the second one, because it is not separable from the rest of ad rank, but it is not zero and it points the same direction.
2. Delivery cost and returns clarity, shown early, which move within days
Unexpected costs at checkout are the most consistently cited reason for cart abandonment in every piece of published research on the subject, and they have held that position for years. The shopper is not refusing to pay for shipping. They are refusing to be surprised.
The fix is not necessarily free shipping. The fix is showing the number early. Shipping cost and a delivery window on the product page, not discovered at step three of checkout. A returns policy stated in one sentence next to the buy button, not linked in the footer. If returns are free, that is the single most valuable sentence on the page for a shopper who has never bought from you, and hiding it in a policy page is throwing it away.
This lever is fast because it is a template change and a settings change rather than a project, and it is the one place where this post and its sibling pull in exactly the same direction. A free shipping threshold with a visible progress indicator raises order value and removes the shipping surprise at the same time, which is two of the three terms moving off one afternoon of work.
3. Checkout friction, which moves within a week
Shopify's checkout is already among the best in ecommerce, and that is genuinely one of the reasons to be on Shopify. Which means most checkout damage is self-inflicted, and most of it comes from four things: forced account creation, an interstitial step between the cart and payment, too many payment options presented at once, and a prominent discount code field that sends shoppers off to open a new tab and search for a code they did not know existed.
Turn on accelerated checkout and the wallets your customers actually use. For a returning shopper, Shop Pay is one of the few durable conversion advantages the platform gives you for free. Allow guest checkout, always, and particularly for a cold Shopping click, where asking a stranger to create an account before their first order is asking for a favor you have not earned yet.
The audit takes fifteen minutes. Buy something from your own store, on a phone, on cellular data, as a customer who has never visited before, and time it. Most people find at least one step they did not know was there.
4. Product page quality and imagery, which move within weeks
The Shopping click lands on a product page, and that page is not being judged in isolation. It is being judged against the next three tiles the shopper is going to click. What loses that comparison is rarely ugliness. It is missing information: no sense of scale, no material, no dimensions, a single image on a white background, and no answer to the one question a buyer in your category always has.
Concretely, that means at least five images, including one showing the product in use and one that establishes scale against something familiar. It means the specification your category argues about sitting in the first screen rather than in a tab. It means variants that show what they look like rather than naming them in a dropdown. And it means short silent video, which is now cheap, because it answers movement and texture questions that stills cannot.
This is slower than the first three because it is a shoot and a rewrite rather than a setting. Do not attempt the catalog. Pull the ten products that take the most Shopping clicks and fix those, because in most stores those ten are the majority of the paid traffic and the rest can wait a quarter.
5. Reviews and trust signals, which move within a quarter
For a cold shopper, coverage matters more than score. A 4.6 average across 80 reviews outperforms a 5.0 across 2, because two reviews read as friends and eighty read as evidence. Products with no reviews at all are the ones that lose the comparison quietly.
The way to get them is an automated post purchase request with a photo prompt, sent on a delay that matches when the product is actually in use. Photo reviews are worth chasing specifically, because they answer the scale and color questions your product photography ducked, and they do it in a voice the shopper trusts more than yours.
Reviews also feed back into the ads. Product ratings can appear as stars on your Shopping listings, sourced through Google Customer Reviews or an approved third party aggregator, and stars raise click-through. Higher click-through at the same bid means you buy more clicks for the same money, which shows up as a lower effective cost per click and moves the other side of the same sum.
This lever is slow for an unavoidable reason. It is gated on order volume, and you cannot honestly accelerate it. The dishonest version, which is buying reviews or writing them, is not a conversion tactic. It is a Merchant Center misrepresentation problem waiting to happen, and it is one of the harder suspensions to argue your way out of.
6. Search and filtering, for larger catalogs only
Below a few hundred SKUs, skip this. It is noise, and the work is better spent on the five levers above.
Above that, it becomes the recovery path. A Shopping click lands on one specific product, and often that product is nearly right and not quite: wrong size, wrong finish, wrong capacity. What happens next decides whether that click was wasted. If internal search handles synonyms, typos and category language the way your customers actually type it, the shopper finds the right variant. If it returns nothing, they leave, and you paid for the click either way.
Measure before you buy anything. Look at what share of sessions use site search and what share of those searches return no results. If almost nobody searches, this lever is not your problem. If a meaningful share search and a meaningful share of those get nothing back, you have found a leak with a price tag on it.
The Apps
Thirteen apps and two things that are not apps, grouped by the lever they serve. Every listing below was opened and read on September 1, 2026, and the ratings, review counts and prices are what the public Shopify App Store showed on that day. Several of these were renamed or moved in the last year, so if you have an older bookmark it may now be a dead link. Read the disclosure in the next section before you read this one.
One thing before any of it. Every lever below has a free diagnostic that comes first, and in most stores the diagnostic finds work that costs nothing to do. Installing an app to fix a problem you have not measured is how stores end up with the script bloat that caused the speed problem in the first place.
For page speed: measure first, and it is free
Run your top product pages through PageSpeed Insights on mobile and read the field data section, which is real Chrome users on your real site, rather than the lab score underneath it. Then open your theme and look for script tags belonging to apps you no longer have installed. That audit costs nothing, takes an hour, and in a lot of stores it is the entire speed problem.
Tiny SEO Speed Image Optimizer, by TinyIMG
Rated 5.0 stars from 2,286 reviews, free to install with 50 optimizations a month included and paid plans from $14 a month. Tiny SEO Speed Image Optimizer compresses and resizes images and adds lazy loading, which addresses the most common single cause of a slow product page, which is a hero image shipped at three times the pixels anyone will see. This app was called TinyIMG until recently and its listing moved, so an older link to it will 404.
Avada AI SEO Image Optimizer, by Avada
Rated 4.9 stars from 4,372 reviews, with a free plan covering 100 image optimizations and 20 audits, and paid plans from $34.95 a month with a 7 day trial. Avada bundles image work with a broader technical audit, which makes it the better pick if you want one app to tell you what is wrong rather than one app to compress images. It was renamed and repositioned toward AI search this year, so parts of the listing are now about something other than speed.
Hyperspeed EXTREME Page Speed, by Rvere
Rated 4.8 stars from 145 reviews, one plan at $59 a month with a 7 day trial and no free tier. Hyperspeed does script deferral, preloading and render optimization, which is the layer above image compression. We are naming it with a caveat attached: 145 reviews is a thin base next to the thousands behind the two above, it is the most expensive app on this page for what it does, and there is no free tier to test with. Try it after the free audit and the image work, not before, and only if you still have a number you cannot fix.
Essential Estimated Delivery Date, by Essential Apps
Rated 5.0 stars from 922 reviews, free, and carrying the Built for Shopify badge. Essential Estimated Delivery Date puts a delivery window on the product page. This is the highest ratio of effect to effort anywhere in this post: it is free, it takes twenty minutes, and it answers the question a cold shopper asks immediately after deciding they like the product.
CBB Shipping Rates Calculator, by Code Black Belt
Rated 4.6 stars from 208 reviews, one plan at $4.99 a month with a 14 day trial. CBB Shipping Rates Calculator shows a real shipping cost in the cart based on the shopper's location, before checkout, which removes the surprise that abandonment research keeps putting at the top of the list. It was called Shipping Rates Calculator Plus until the developer moved to house naming.
Return Prime, by Appsdart
Rated 4.8 stars from 723 reviews, free to install with 5 return requests included and then $0.49 per request, or paid plans from $19.99 a month with a 15 day trial. Return Prime runs returns and exchanges through a self serve portal. The conversion argument for it is not the portal, it is that having one lets you write a confident one line returns promise next to the buy button and mean it.
Checkout: mostly not an app, and worth saying plainly
Shop Pay and checkout extensibility are not apps and there is nothing to install. Shop Pay and the other accelerated checkouts are switched on in your admin under Settings and then Payments, and the checkout itself is edited in Shopify's checkout editor. Shopify documents both in its own checkout help pages. If someone sells you an app to do this, they are selling you a wrapper around a setting you already have.
Shopify Checkout Blocks, by Shopify
Rated 4.3 stars from 213 reviews, free, first party. Checkout Blocks adds content and custom fields to checkout and post purchase pages without code. One gate you need to know before installing: editing the checkout page itself is Shopify Plus only, and on other plans this app reaches the thank you and order status pages but not checkout. That restriction is the loudest theme in its reviews and it is the reason for the 4.3.
PageFly AI Page Builder, by PageFly
Rated 4.9 stars from 5,680 reviews, with a free plan that covers one published page and paid plans from $24 a month. PageFly lets you rebuild a product page layout without touching Liquid, which matters when the fix is structural: specification above the fold, variants shown rather than listed, trust content placed where a cold shopper looks. The free plan's one published page is enough to rebuild your single highest traffic product and measure it before you commit. We are not quoting a trial length because the listing did not state one clearly on the day we checked.
Videowise Shoppable Video UGC, by Videowise
Rated 4.8 stars from 219 reviews, with a free plan and paid plans from $9 a month with a 14 day trial. Videowise puts short shoppable video on product pages without the page weight of an embedded player, which matters because video and page speed usually fight each other. Worth knowing what is not here: we looked at the 360 degree spin viewer category and no app in it has enough review volume to name responsibly, so we are naming none of them rather than picking the least thin one.
Judge.me Product Reviews
Rated 5.0 stars from 44,528 reviews, with a genuinely free forever plan and a paid tier at $15 a month with a 15 day trial, carrying the Built for Shopify badge. Judge.me is the default answer for most stores, and the review count behind that rating is not a typo. Automated post purchase requests, photo reviews, and syndication of product ratings to Google are all on the free plan, which means there is no reason for any store to be running with no review coverage at all.
Loox Product Reviews
Rated 4.9 stars from 9,176 reviews, with a free plan for low order volume and paid plans from $49.99 a month with a 14 day trial. Loox is built around photo and video reviews specifically, which is the format that does the most work on cold traffic because it answers scale and color questions your own photography ducked. It is meaningfully more expensive than Judge.me once you leave the free tier, so the question to answer first is whether your customers will actually send photos.
Okendo Reviews and Loyalty
Rated 4.9 stars from 1,333 reviews, free up to 50 orders a month and paid plans from $19 a month with a 14 day trial. Okendo is the option for stores that want attribute level reviews, where buyers rate fit or durability on a scale and shoppers can filter by them. It has expanded from a reviews app into a suite covering loyalty, surveys, quizzes and referrals, so calling it a reviews app is now about half the story. Its listing also moved this year.
Shopify Search and Discovery, by Shopify
Free, first party, and rated 2.7 stars from 464 reviews, which is the worst rating on this page by some distance. Search and Discovery adds filters, synonyms, boosting rules and product recommendations at no cost. We would normally tell you to start with the free first party option and only pay when you hit its limits. Here we are going to be more careful, because the recurring complaints are specifically about irrelevant results on large catalogs and a ceiling of 25 filters, and a large catalog is the only situation in which this lever matters at all. Install it, because it is free and it may well be enough. Just expect to find its edges faster than usual.
Boost AI Search and Filter, by Boost Commerce
Rated 4.8 stars from 1,500 reviews, with plans from $29 a month scaling with your revenue up to considerably more, and a 21 day trial. There is no free plan. Boost is the heavyweight in this category: semantic search, merchandising rules, and filtering that holds up on tens of thousands of SKUs. It is the most expensive thing named on this page at the top of its pricing, and it earns that only at catalog sizes where a percentage point of search recovery is real money. Its listing moved this year too.
Searchanise Search and Filter
Rated 4.8 stars from 1,071 reviews, free for stores under 25 products, then plans from $19 a month with a 14 day trial. Searchanise sits between the free first party app and Boost on both capability and price, which for a mid sized catalog is usually the right place to sit. Note that the free tier is scoped to stores small enough that this lever does not apply to them, so treat it as a trial rather than a plan.
What We Are Not Telling You
Every list of apps on the internet is somebody's affiliate income. Here is ours, in full.
We have no affiliate relationship with any app or tool named above. No referral links, no revenue share, no sponsorship, no commercial arrangement of any kind with any of the developers. The links go straight to the public App Store listing, or to Shopify's own documentation where the thing in question is a setting rather than an app, with nothing appended to them. Nobody paid to be on this page and nobody paid to be left off it.
The ratings, review counts and prices were read off the public listings on September 1, 2026. Those numbers move, sometimes quickly, and a well rated app can be acquired and rebuilt into something else inside a quarter. Treat every figure above as a snapshot with a date on it rather than a fact, and check the listing before you install anything. Four of these listings had moved to a new address since we last checked and five of the apps had been renamed, in the space of about a year, which should tell you how much weight to put on any number in a post like this one.
Three things we deliberately did not print. PageFly's trial length, because the listing did not state one clearly on the day we looked. Loox's free tier order cap, because the listing showed two different figures in two places. And any 360 degree product spin app, because no app in that category has enough review volume behind it for us to stand behind a recommendation, so we named video instead and said why.
We also broke our own habit once. We normally lead a category with the free first party option, and in the search and filtering section we hedged that advice instead, because Search and Discovery rates 2.7 and its complaint pattern lands precisely on the use case the section is about. Nothing on this page pays us, which is exactly why we can say that about a Shopify app on a page aimed at Shopify merchants.
And the part that matters most. A higher conversion rate makes your store easier for us to fund. Look again at the sum at the top of this post: it is our sum, not a general purpose ecommerce truth we found lying around. We pay for the clicks and we bill a commission on the orders, so your conversion rate sits directly inside our own economics, and the 0.6 haircut is not a neutral observation either. It is the number that decides which stores we say yes to. We are not disinterested about any of this. That is exactly why the post exists, and we would rather say so in the middle of the page than bury it in a footer.
None of that makes the arithmetic wrong. It does mean you should check it, which is why every number in it is shown rather than asserted.
Three Ways This Goes Wrong
Moving the site-wide number while the paid number sits still
This is the most common failure and it is a measurement failure rather than a work failure. A merchant does three months of genuine conversion work, watches the Shopify dashboard climb from 2.4% to 2.8%, and reports a 17% improvement. Then the paid campaigns perform exactly as before.
Both things can be true at once, because the site-wide number is a blend. If your email program had a good quarter, or a press mention drove branded search, or you simply sold more to returning customers, the blend rises without the cold traffic moving at all. The term in the sum is the paid Shopping rate, and it is perfectly capable of standing still while the headline number goes up.
Segment it and stop guessing. In GA4, break session conversion rate out by session default channel grouping and read the paid shopping row on its own, or filter to the session source and medium your Shopping campaigns use. In Google Ads, divide conversions by clicks for the Shopping or Performance Max campaign directly. Then track that number, not the dashboard one, and while you are there compute your own store's ratio between the two. If your real ratio is not near 0.6, use yours. We would rather you brought us a measured number than accepted our default.
Buying conversion rate with discounts
A sitewide discount code will raise your conversion rate. It is the most reliable lever in this entire post, and it is close to useless, because of where the money comes from.
Run it through the sum. Take the $120 store at 1.8% on paid traffic. Add a 10% off code across the site. Order value falls to $108. Conversion rate rises to 2.0%, which is a strong 11% lift. Now multiply: $120 times 1.8% is 2.16, and $108 times 2.0% is 2.16. Identical. The CPC ceiling has not moved by a cent, because the sum multiplies the two terms and the discount moved them in opposite directions by the same proportion.
The feasibility is flat. The margin is not. On a 45% gross margin, a $120 order leaves you $54 of gross profit and a $108 order with the same cost of goods leaves you $42. Per thousand clicks, you go from 18 orders at $54, which is $972, to 20 orders at $42, which is $840. That is 14% less gross profit for more orders, more picking, more packing, and more support tickets.
Our side of it makes the point sharper. Commission per thousand clicks was 18 orders at $18, which is $324. Afterward it is 20 orders at $16.20, which is $324. Exactly the same. The metric improved, our revenue did not change at all, and the only party who lost money was you.
Check the margin arithmetic before the code goes live rather than after. Our profit calculator will tell you what a given discount actually costs you per order. A conversion rate earned by a faster page or a clearer shipping message raises the product of the two terms. One earned by a discount usually just moves value from one column to another and charges you for the privilege.
Optimizing a micro-conversion instead of an order
Popups, spin-to-win wheels, countdown timers and low stock counters all have dashboards, and those dashboards report impressive numbers. The trouble is that the number they report is not the number in the sum.
An email capture popup measures popup conversion rate. It can run at 8% and be celebrated while it is costing you orders, because it fires over the product page on a cold visitor's first three seconds, it shifts the layout, and on mobile the close button is a small target. The email list grows. The order rate falls. Both are real, they are measured in different tools, and only one of them is in the sum.
Urgency tactics carry a second cost. A countdown that resets when you reload, a low stock warning on something with unlimited stock, or a strikethrough price that was never charged are misrepresentation, and Merchant Center enforces on exactly this. Losing Shopping traffic entirely is a poor price for a small lift in add-to-cart rate.
If any of that already happened to you, we wrote up the recovery process in how to fix a Merchant Center misrepresentation suspension. The rule that avoids it is simple enough to hold onto: if the mechanism would embarrass you when explained to the shopper, it is a policy risk and not a conversion tactic. Judge every one of these tools on orders per session, measured on the paid segment, and let the tool's own dashboard say whatever it likes.
What to Do in the First Two Weeks
Days one and two: get your paid Shopping conversion rate out of GA4 or Google Ads and write it down next to your site-wide number. Divide one by the other. That ratio is your store's real haircut, it is the single most useful number in this post, and almost nobody has it.
Days three and four: run your three highest-click product pages through PageSpeed Insights on mobile and read the field data rather than the lab score. Then go through your installed apps, remove the ones you are not using, and check the theme for script tags left behind by apps you removed a year ago.
Days five to seven: put shipping cost, a delivery window and a one sentence returns statement on the product page above the fold. Then buy something from your own store on a phone, on cellular, as a guest, and fix whatever you trip over.
Week two: a photo and copy pass on the ten products that take the most Shopping clicks, and switch on automated post purchase review requests with a photo prompt. That second one starts slow and compounds, which is the point of starting it in week two rather than in month six.
End of week two: stop, and then be patient in a way that is genuinely difficult. Conversion rate is much noisier than order value. At 1.8% and a thousand clicks a week you are looking at about 18 orders, and a move from 1.8% to 2.16% is 18 orders becoming 22. That is indistinguishable from a good week.
If you want to prove a 20% conversion rate lift as a proper split test at those rates, you need something on the order of 20,000 clicks per variant before the result stops being luck. At our measured click cost that is around $4,600 of traffic per variant. Which is the honest reason most stores should not A/B test their conversion rate at all. Make the page unambiguously better, ship it, and judge it over a quarter rather than a fortnight.
The Honest Ceiling
Everything above has a limit, and pretending otherwise would undo the point of the disclosure.
Conversion rate is the least elastic of the three terms. For a store that has done none of this work, doing all six levers properly over a quarter is realistically worth something in the range of 10% to 25% relative. That takes 1.8% on paid traffic to somewhere around 2.0% or 2.2%, and it takes the CPC ceiling in our worked example from $0.29 to about $0.33 or $0.36. Real money against a $0.23 click, and not a rescue on its own.
The gap between your site-wide rate and your paid rate narrows with this work but it does not close, and you should not aim for it to. Part of that gap is a defect you can fix and part of it is simply what the traffic is. A shopper comparing eight tiles is going to convert below a shopper who typed your name, in every store, forever. Anyone promising you otherwise is selling something.
Which brings the argument back to where it started. If the table above says you are a long way short, the answer is almost never all of it from one term. It is 15% here and 20% there across order value, conversion rate and commission rate, stacked, which is the case the average order value post works through in full. And if your conversion rate is fine but large parts of your catalog never appear in an auction at all, the problem is a different one, which is what escaping low revenue mode is about.
It is also worth knowing that performance across the stores we fund follows a power law rather than a normal distribution, which we covered in why PPC behaves like a power law. Most stores sit in the long tail and run perfectly good businesses there. And if you are setting this up from scratch, the feed and campaign side is covered end to end in our complete guide to Google Shopping ads on Shopify.
The useful thing about a sum with three terms is that it turns a vague worry into a line you can stand on one side of. Put your own numbers into the feasibility calculator, using your segmented paid conversion rate if you now have it, and find out how far off you are. Sometimes the distance is a quarter of unglamorous work on page speed and shipping messages. Sometimes it is a channel that was never going to be yours, which is a fine thing to learn early and for free.
One caveat on everything above. The click cost, the 0.6 conversion haircut and the target return used in this post are our own measured figures as of September 2026, across our own portfolio, and they are averages over a wide spread of stores. They are published so the arithmetic can be checked, not so it can be treated as a forecast for any individual store. This is educational rather than financial advice. Use your own numbers, verify everything against your own reporting, and treat any figure here that you cannot reproduce as a question rather than a conclusion.
