KPIs: Edited chief scientist Michael Ross on retail’s ‘red flag’ metrics

Without giving too much away, what will your LIVE 2026 session be about?

It’s all about ‘measure what matters’ and how retailers need to think about the measurement of their business performance. My contention is that the world is in a mess, a mess that started at the birth of the internet and has got steadily worse over time.

When online was 2% or 3% of sales and relatively separate from stores, you could run your stores and online businesses independently. But over time, online has become a much bigger percentage in terms of people transacting online, and it’s changed customer behaviour. People are browsing online and buying offline, or browsing offline and buying online. They’re in stores on their mobiles. So you can’t look at stores and online independently anymore.

At some point in the last 30 years, we recognised that what retailers measure and how they measure was woefully no longer fit for purpose. But the challenge is that people haven’t yet worked out what it is. How do you instrument an omnichannel, customer-centric business to understand what’s going on, to drive the right Monday morning trading conversations and align the organisation?

Can you give an example of this problem?

My epiphany was at Figleaves. In 2004, when I was CEO and we were about £20m-£25m turnover, I hired my successor, Robin Tyrrell, who’d been running Amazon UK. After a couple of weeks, he sat me down and said, “Michael, you’re measuring all the wrong things.

“You’re telling me that you’re 92% in stock at the SKU level, which for lingerie is the individual size variant. It doesn’t matter. What matters is, are we in stock of the products that customers are looking at?” This is an Amazon measure of viewed availability.

We looked at our unweighted availability and thought it seemed pretty good. Then we looked at our viewed availability, and it was horrible – about 70%. It was horrible because we had lots of very fragmented products, lots of marketing traffic landing on products that were sold out or very fragmented, and lots of sort orders on the website that were very good at driving traffic to bestsellers but not so good at pushing them down when they started fragmenting.

Other than viewed availability, are there other obvious KPIs that retailers probably aren’t taking as seriously as they could be?

Absolutely. At LIVE, I’m going to talk about Goodhart’s Law, which says that when a metric becomes a target, it ceases to be a good metric.

Philip Green, when he was running Topshop, set up the Topshop website and incentivised his team based on three things: drive sales, drive conversion rate and increase average basket size. The clever people running his business worked out that they could run free next-day delivery above £50 promotions all the time. What does free delivery above £50 do? It drives sales, it drives basket value, it drives conversion rate, and it massacres your profit because you’re FedExing all these orders next-day.

The moment you set metrics as targets, you drive lots of very dysfunctional behaviour.

Is there anything else that retailers aren’t tracking?

Maybe the most important flavour of metrics is failure metrics. When we talk about conversion rates, viewed availability and delivery on promise, they are all, to some extent, averages. The nature of a digital business is that you can both measure and take action at a very granular level.

One of the things that retailers need to track that they typically don’t is failure metrics. For example, how much inventory value do you have on your website that nobody has viewed?

A recent example – a retailer in the US with $50m of inventory, talked about average stock turn and average product conversion rate. It turned out $15m of inventory hadn’t been viewed on their website in the last four weeks.

Inventory not viewed is, for me, a very good example of a de-averaged metric. It’s a high-powered metric because it moves away from “What’s my average stock turn?” to “Can we shine a light on where we think there are opportunities to take a different type of action?”

What are the red flags you see?

They’re measuring average outcomes.

I asked someone from Amazon, 20 years ago, “What does Amazon talk about on a Monday morning?” He said they talk about four things: viewed availability, viewed price competitiveness (are we price competitive on the products that customers are looking at?), delivery on promise (are we delivering to customers when we told them we would?), and range competitiveness (do we have everything our competitors have?).

I said, “Do you talk about conversion rate?” This is the key. He said, “No. The Amazon culture is, you can cry about conversion rate, you can’t do anything about it. We talk about metrics we can action.”

That is the profound insight. Conversion rate is an outcome of having the right products, the right prices, and the right availability. Measure it if you want, but don’t talk about it.

Most retailers on a Monday morning will spend all their time talking about aggregated average outcomes, and that’s just a recipe for sub-optimisation and misery.