For shops and makers
Nearly half my orders never touched my website
My conversion rate looked bad enough to justify rebuilding the shop. Then I checked what was actually in the number, and the answer changed what I did next. Here is the check, and it takes about a minute.
Ed Howe · 4 September 2026 · 7 minute read
I run a small product business alongside this one. Last year I looked at its conversion rate, compared it to the figures people quote online, and concluded the shop was broken. I got as far as sketching out a rebuild.
Before spending the money I went looking for which pages were losing people. That is when I found it. Forty three per cent of orders had no online-store session at all. Nearly half the people buying from me had never been on the website in the way analytics counts it.
The shop was not broken. The number was.
Why a conversion rate lies to makers in particular
A conversion rate is orders divided by sessions. Everyone stares at the top of that fraction and nobody checks the bottom. For a maker or an independent shop, the bottom is usually wrong, because a lot of your sales never look like a website visit:
- Someone met you at a market, took a card, and ordered from their phone a week later with a link you sent them
- A repeat customer messaged you on Instagram and you sent a payment link
- A wholesale or trade order went through by email and got keyed in by hand
- Someone bought on a device that blocked the tracking, or arrived from an app that strips the referrer
Every one of those is a real order from a real person. None of them behaves like the website visit the benchmark assumes.
The thing that actually costs money
A wrong denominator does not just make the number look bad. It points you at the wrong job. I was one afternoon away from spending four figures redesigning a shop that was converting fine, while the actual opportunity, the people who had already bought once, sat untouched.
The check, and it takes about a minute
You are comparing two numbers that should be close and often are not.
- Find your total orders for a month. Any shop platform shows this on its front page.
- Find your sessions for the same month, and find the orders your analytics attributes to those sessions.
- Subtract. Anything left over is orders your website never saw.
If the gap is small, your conversion rate is roughly honest and you can act on it. If the gap is large, stop using that percentage to make decisions. It is measuring a different business from the one you are running.
Do it for a full month, not a week. Markets, fairs and trade orders arrive in lumps, and a week can easily land entirely inside or entirely outside one.
What to do once you know
The answer is not to fix the tracking, or not first. It is to notice what the gap is telling you: a meaningful part of your business is relationship, not traffic. People who already know you, buying again, through whatever channel is nearest to hand.
That changes where the next hour goes. Chasing more visitors is expensive and slow. Making it easier for people who have already bought to buy again is cheap and fast, and almost nobody does it properly. In my own shop, repeat customers run between a quarter and a third of orders, revenue is up 24.6% year on year, and 32% of that growth happened over five months when I was working thirty hours a week for somebody else. That was not a traffic strategy. There was no time for one.
So: a short email to past customers that goes out without you writing it. A saved message for market contacts. An order confirmation that says something more useful than "thank you for your order". Unglamorous, and it compounds.
The general version
Before you fix a number, check what is in it. Almost every alarming figure in a small business turns out to be a measurement problem, a seasonal lump, or a definition that does not match how the business actually works. The dashboard is confident and the dashboard does not know about the market stall.
That is not an argument against measuring. It is an argument for spending ten minutes on the denominator before spending four figures on the numerator.
The Health Check is ninety minutes on your actual numbers, then a written plan with prices. Most of the value is usually in finding out which of your problems is not a problem. Every account is set up in your name, so if you stop working with me you keep all of it.