At IMRG, we deal with averages a lot. We take in data from lots of different retailers and provide benchmarks across a wide variety of metrics. Providing averages is what we do, and yet this article cautions against taking them too literally; to be clear, averages are really, really useful and can be crucial in informing business decisions, but it’s interesting to take stock and realise what an average actually is telling you, and what it is not.  

First though, a quick market update. We’ve had a really good start to the year for eCommerce growth, with revenue up around +3% year-to-date against a forecast of +1% for 2025 as a whole. There are still six months to go and it can all change obviously, but growth in the market generally feels a lot more consistent and reliable. That said, May was a lot flatter than we had become used to this year; many industries reported a bit of a slump in fact. June has started off with positive growth again though, so it could be that the bank holidays were just a bit bunched up due to Easter being so late and people overspent a bit toward back end of April. Just a blip perhaps.

Now: back to averages. If you are trying to work out the average of something, by its very definition, you simply add all the values together of the participants within the group then divide the total figure by the number of participants. It makes a lot of sense and gives you a figure that is representative, revealing and handy. For example, the average engagement time on eCommerce sites (which means how long visitors spend on them before leaving) averages out at around two and a half minutes. To inform planning, knowing how long you have to convert visitors into customers sharpens the focus around optimisation.

Why Context Is Everything in Interpreting Averages

And yet, we have to remember exactly what averages are – a collection of values taken from a number of participants, which in our context are retail businesses who may have widely varying propositions, customer demographics and price-points. A retailer selling expensive, infrequent purchases such as sofas or beds will see different performance in certain metrics than those selling low-cost, repeat-purchase items such as make-up.

Even when you account for such variation, the interesting consideration arises when you look at the individual values that make up the average and plot them on a graph. Take the below example, which shows the individual values for 90 retailers across the previous three quarters for the percentage of visitors to retail sites who proceed to a product page. The average runs at around 60% for all three quarters which means, incredibly, that 40% of visitors don’t actually view any products on a site that purely exists for that purpose.

When viewed like this, a few things become apparent. You get outliers – a couple hover around the 20% mark and a small selection are over 80%. But if you knock out the outliers, who are so few they aren’t going to skew the average, you realise that an average of 60% actually means there aren’t that many who are exactly on that figure, instead the overwhelming bunching up of values sits between 50-70%.

Averages Don’t Equal Targets

So: 60% is the average and it is accurate and useful to look at averages, but actually far more participants have rates closer to 50% or 70%. Hence, if you are working out a strategy for how to improve a metric, having the average is a great guide but that’s not what you should be aiming to achieve exactly as not many sit on the average. Instead an average of 60% tells you that good performers in that area sit at closer to 70%.

Averages are really useful and should inform decision-making, absolutely – but always remember that hardly any participants will sit squarely at that value so trying to achieve it exactly might be a misleading aspiration.