Earlier this month I wrote about the site passing a million sessions in a month, and about being wary of that kind of number. A couple of people asked the obvious follow-up: fine, so what do you look at instead? This is my attempt at a straight answer, at least for the kind of business I work in.
Most ecommerce reporting is built for retailers who sell lots of fairly cheap things. Conversion rate, basket size, sessions, bounce rate. Those dashboards come out of the box in Google Analytics and every agency deck uses them, which is a big part of why everyone reports on them. When the thing you sell costs several thousand pounds and people take weeks or months to decide, a lot of that standard set either stops meaning much or actively misleads you.
What I’ve mostly stopped reporting
Site-wide conversion rate. At Watchfinder the online conversion rate is a small number, and it always will be, because nobody buys a Rolex on their first visit after seeing an Instagram post. Worse, it moves for reasons that have nothing to do with how well we’re selling. A good month for content or YouTube brings in a lot of enthusiasts who are browsing for fun, and the conversion rate drops. That shows up on the report as a dip, when often it means the content is doing its job.
Bounce rate as a quality score. Someone who lands on a model page, reads the specs, checks the price and leaves might be doing exactly what they came to do. I’d rather know whether they came back.
Raw session totals. Covered that one already.
What I look at instead
Enquiries, not only orders. A lot of high-value purchases start with a question. A phone call, an email, a chat, a request to view a watch in a showroom. If you only count the online checkout you’re missing a big chunk of what the website actually does. We’ve been working on call tracking with ResponseTap so phone enquiries can be tied back to the channel that drove them, and even in its early days it has changed which campaigns look good.
Both sides of the business. Watchfinder buys watches as well as sells them. Someone asking for a valuation on the watch in their drawer is just as valuable a lead as someone browsing to buy, and often more, because good stock is what everything else depends on. So the sell-side enquiry numbers sit on the same report as the buy side, not tucked away in a separate tab.
Returning visitors and time to purchase. How many visits and how many days it takes before someone buys tells me more about the customer than any single-session metric. If that gap is getting longer, I want to know why before it shows up in revenue.
Repeat customers and lifetime value. Collectors rarely stop at one watch. A customer who buys, has a good experience, and comes back in eighteen months to trade up is worth far more than the first order suggests. That changes what you can sensibly spend to acquire them in the first place, and it’s the number I’d most like finance and marketing to agree on.
Brand search volume. People searching for your name is about the clearest sign that your other activity is working, including the stuff that’s hard to attribute, like TV, PR and the YouTube channel. It’s slow and noisy month to month, so I look at it over a year.
The awkward ones
There are two numbers I watch that rarely make it into anyone’s headline slide.
The first is returns and cancellations. A campaign that brings in orders which are then returned is worse than one that brings in nothing, because you’ve paid for the traffic, the shipping and the handling. So I want returns netted off in any channel report before it shows me orders.
The second is cost per enquiry by channel, set against what those enquiries eventually become. It’s hard to do properly, because the lag between enquiry and sale can be long and the joins between the website, the phones and the CRM are never as clean as you want them to be. I don’t have this working the way I’d like yet. Even the rough version I have is more use than cost per click.
The caveat
None of this is unique to watches. Furniture, cars, jewellery, anything people think about for a long time before they buy runs into the same problem. And I’m not saying conversion rate is useless; for checking that a page change hasn’t broken something, it’s fine. It just shouldn’t be the number that decides whether marketing is doing its job when most of the decision happens off the page.
If you work in a similar sort of business and track something I’ve missed, I’d genuinely like to hear about it.