Search this question and you will find confident figures: two to three per cent for e-commerce, higher for some sectors, lower for others. They are not invented. They are also not much use, and it is worth understanding why before you make a decision based on one.
Why the averages mislead
An average conversion rate is calculated across businesses that share almost nothing. A site selling five-dollar phone cases and a site selling forty-thousand-dollar industrial equipment are both in the average. So is a brand everybody already trusts and a brand nobody has heard of. So is traffic from somebody's mailing list and traffic bought from a display network.
So when your rate is below the published average, you have learned nothing about your page — and when it is above, you have learned nothing either, except perhaps that your traffic is warmer than most.
The comparison that actually works
Your own site, last month, with the traffic sources held roughly constant.
That comparison controls for everything the averages cannot: your price, your market, your reputation, how people found you. It is the only baseline that makes a change interpretable, and it is available to you for free.
Segment it before you read it
- By source. Search, social, email and paid traffic convert differently enough that a single blended number can hide a collapse in one channel behind growth in another.
- By device. Phone and desktop are effectively two different sites with two different rates, and the phone one is usually both larger and worse.
- By whether they had been before. Returning visitors convert far better, so a good month for new traffic can look like a bad month overall.
How to tell a real change from noise
This is where most conversion conversations go wrong. Small numbers move around on their own, and if you look at a chart every day you will see meaning in that movement, because that is what people do with charts.
A rough guide: with forty conversions a month, an apparent change of a few percentage points is well within what randomness produces. You need either a much bigger effect or a much longer period before the change means anything at all.
The number that is more useful than conversion rate
Revenue per visitor. Conversion rate can be raised in ways that lose money: discount everything and more people will buy, at a lower margin. It can also fall while the business improves, if you start attracting more people who are early in their decision.
Revenue per visitor closes both gaps, and it is the number a business owner actually cares about even when they asked about conversion rate.
So what is a good conversion rate?
| Situation | The honest answer |
|---|---|
| You have no baseline yet | Measure for a month. That is your number, and it is neither good nor bad yet |
| Below last month, same traffic mix | Something changed. Worth finding |
| Below an industry average | Almost no information. Ignore it |
| Very high and revenue is flat | Check whether you are discounting your way there |
| Under 50 conversions a month | Too small to read as a rate. Look at mechanics instead |
The genuinely useful version of this question is not “is my rate good” but “is there anything on this page costing me people for no reason”. That one has answers, and they do not depend on anybody's benchmark.
Measure the steps, not only the sale
A single conversion rate is one number describing a journey with several stages, which is why it moves without telling you anything. Splitting it up costs nothing and turns a vague figure into a diagnosis.
- Reached the page — the denominator everything else divides into.
- Scrolled past the first screen — did the opening earn any attention at all?
- Reached the form or the basket — did the offer interest them?
- Started filling it in — did they intend to go through with it?
- Finished — did anything stop them at the last moment?
The gap between “started the form” and “finished the form” is a completely different problem from the gap between “arrived” and “scrolled”. One is a form problem, the other is an offer or headline problem, and a blended conversion rate hides which of the two you have.
Questions people also ask
What is the average e-commerce conversion rate?
Published figures usually land somewhere around two to three per cent, but the spread behind that average is enormous and it includes businesses nothing like yours. It is a fact about a dataset, not a target for you.
My rate dropped last week. Should I worry?
Probably not yet. Check first whether your traffic mix changed — a burst of cheap or cold traffic lowers the blended rate without anything on your site getting worse. Then check whether the drop is bigger than the week-to-week movement you normally see.
Is a higher conversion rate always better?
No. Discounting raises it and can lose money. Attracting more people earlier in their decision lowers it and can make you more. Revenue per visitor answers the question conversion rate is standing in for.
How long should an A/B test run?
Long enough to reach the sample you decided on before you started, and no shorter. Stopping early because it looks good is the single commonest way to get a confident wrong answer.