- e-commerce
- Meta Ads
- conversion funnel
Meta Ads for an e-commerce brand: the ratios that tell you if it works
The funnel steps worth watching, the conversion benchmarks, and the point where the problem is the site rather than the campaigns.
Published on · Joh · 9 min read
A brand that sells online has an enormous advantage over a tradesperson: every step between the ad and the sale is measurable. Product view, add to cart, checkout started, purchase. Four checkpoints, four places where the leak shows.
You still have to look at the right ones.
The ratios that count
The benchmarks I use to place an account:
| Step | Pass-through rate | Reading |
|---|---|---|
| Visitor → add to cart | 8 to 12% | Under 8%, the product page or the price is the problem |
| Add to cart → checkout started | around 50% | Under 40%, look at the shipping costs |
| Checkout started → purchase | varies with order value | A drop here is almost always technical |
These ranges are orders of magnitude, not laws. A €2,500 basket does not behave like a €40 one.
What they give you is a sense of where to look. An add-to-cart rate of 3% alongside a good click-through rate on the ad is not a campaign problem: the ad did its job, the page is losing people.
Diagnosis, from the bottom up
The order in which I look at an account that is not performing:
The offer. Does the product sell anywhere else? If word of mouth and organic traffic do not convert either, advertising will simply make the same refusal more expensive.
The funnel. Page speed, clarity of the price, shipping costs announced late, checkout in three clicks or in eight.
The creative. Is the link click-through rate running between 1 and 2%? Do the first three seconds hold?
The audience. Rarely the root cause, often the first place people start tinkering.
The settings. They amplify what works. They repair nothing.
That order is counter-intuitive for most advertisers, who start with targeting. It is the only way not to spend three weeks optimising a campaign that sends traffic to an unreadable product page.
Which event to optimise for
The rule is simple: the deepest event your volume allows.
Purchase if you have enough. Checkout started otherwise. Add to cart if volume is still low. Product view as a last resort, to get going.
The algorithm needs roughly 50 conversions per week per ad set to leave its learning phase. Below that it gropes around, and your results look like noise. Moving one step up the funnel gives you volume and usable signals, and you can move back down later, once sales follow.
One more thing, often set wrong: prefer Meta's standard events over custom ones for optimisation. Custom events are for building audiences and for analysis, not for steering the learning.
Tracking, which decides everything else
Since the tracking restrictions, a significant share of conversions no longer comes back through the browser. Ad blockers, cookie refusals, browsers that cut tracking off.
Two consequences: your campaigns look like they are underperforming, and the algorithm learns on data full of holes.
Installing the server-side Conversions API fixes a good part of it. This is not a comfort setting, it is what determines the quality of everything downstream.
And compare against your back office. If Meta shows 40 sales and your platform counts 55, the channel is more profitable than it looks.
Creatives, and the mistake of rationing them
On an e-commerce audience, an ad wears out fast. The same people see you several times a week, engagement falls, and the cost per thousand impressions climbs.
The usual reflex: keep the ad that works until it collapses. That is the best way to pay more and more for a weaker and weaker result.
What works better: spin it off. A creative that works contains an angle that works. Change the hook, the background, the pace, the person on screen, and you recover performance without starting from nothing. The same juice, pressed several times.
It is also why I do not bill creatives one by one. A client who thinks twice before asking for a new video ends up running the one they have for too long.
Two creative metrics to watch
Hook rate — three-second views against impressions. Above 30%, the video stops the scroll. Below 20%, the first three seconds are not doing their job.
Hold rate — fifteen-second views against three-second views. Around 10% is fine. Below 5%, the script loses people.
Watch the trap: a clickbait hook lifts the first number and destroys everything after it. You pay for views that will never buy anything from you.
Further reading: Why your ads cost more and more and What a Meta video ad costs.
Want a look at your numbers? . We go through your ratios and find the floor that is leaking.