Most of the conversion rate optimisation advice I’ve read over the years quietly assumes a particular kind of business. Lots of traffic, lots of orders, an average basket of somewhere between £30 and £100. In that world you can run an A/B test on a button colour, get a result within a fortnight, and roll it out with confidence.
Sell things that cost thousands of pounds and most of that falls apart, partly because of the arithmetic and partly because the customer behaves differently. I’ve been thinking about this a lot over the summer, so here’s where I’ve got to.
The numbers problem
Statistical significance needs volume. If a site takes a few thousand orders a week, a test splitting traffic in two will reach a reliable answer fairly quickly. If the orders number in the dozens, the same test might take months to say anything trustworthy, and by then the season has changed, the stock has changed and a marketing campaign has muddied the water.
The temptation is to call tests early because a line on a chart looks promising. With small numbers that’s how you end up rolling out changes that did nothing, or worse, on the strength of what was really random noise. It happens at every size of business, but with low order volumes it’s much easier to get away with fooling yourself.
A few ways round it:
- Test bigger changes. A button colour won’t produce a difference large enough to detect on low volumes. A redesigned product page or a different checkout flow might.
- Measure earlier steps. Rather than orders, measure something that happens more often and leads towards them: adding to basket, starting checkout, making an enquiry, booking an appointment.
- Use qualitative evidence alongside the numbers. Session recordings, user testing with five or six people, reading the questions customers ask the sales team. Small samples are fine for spotting problems, if not for measuring them.
The journey problem
The journey for a high-value purchase can run over weeks, across a phone, a work computer and a tablet at home, often with a phone call or a showroom visit somewhere in the middle. The final online order, if there is one, is the last step of a long process.
That has two consequences. Measuring conversion rate per session makes the site look terrible, because most sessions were never going to end in a purchase that day. And the analytics will credit the sale to whatever happened last, usually a brand search or a direct visit, ignoring everything that did the persuading.
So I find it more useful to ask whether the site is doing its job at each stage. Is it helping people research? Making it easy to save or share something to come back to? Giving them a clear way to talk to a person? Making the final step smooth when they’re ready?
Offline conversions count
On high-value products, a lot of the most important conversions don’t happen on the site at all. A phone call, a showroom appointment, a request for more photographs, a valuation form. If those aren’t tracked as goals, the site is being judged on a fraction of what it actually does.
Call tracking is the obvious gap. If a large share of sales start with a phone call, and in this category plenty do, then not knowing which pages and campaigns drove the calls means flying half blind. The same goes for enquiry forms that end up in someone’s inbox and never get tied back to the visit that produced them.
Where I’d start
With all that said, the things I’d look at first on a high-value site are fairly unglamorous:
- Page speed on mobile. Big, beautiful images are a requirement in this category and they’re heavy. Slow pages lose people before they’ve seen anything.
- Contact options that are obvious. Phone number, live chat if you can staff it properly, appointment booking. Make it easy to talk to someone.
- A checkout that doesn’t feel risky. Clear delivery, insurance and returns information right where the anxiety peaks, not buried on a terms page.
- Payment options suited to the price. Card limits are real at these values. Bank transfer, finance and split payments matter more than they would on a £40 basket.
- Ways to come back later. Wishlists, email-a-friend, saved searches, alerts when a similar item arrives.
On a site where each order is worth a great deal, a small improvement in any of these adds up quickly, though proving it will take longer than anyone would like.