Data-led Reinvention

Making data work

Most firms sit on plenty of data telling them what is broken, but use it to sell more of what is broken

Craig Turp-BalazsOctober 27, 20265 min read
Dreamstime

Back in February 2022 when the European Commission set out its case for what would become the EU Data Act, it made great play of the fact that some 80 per cent of all the industrial data collected in Europe (the readings thrown off by machines, appliances, sensors, vehicles, and the like, which manufacturers had long treated as their own private property) is never analysed, never sold, never used, and in most cases never even glanced at, just stored (at some expense) and forgotten about. The law that followed has applied since September of last year, and since September of this year any new connected product sold in the EU must be designed so that its users can get at that data themselves (which is one way of telling manufacturers that if they are not going to take a look at the data, it might be time to let somebody else have a go).

The one-fifth of data that does get read (and there is no reason to think that complaint logs, returns records, customer satisfaction surveys, and call-centre notes fare any better than factory sensors) tends to be put to work on a relatively short list of jobs that focus on short-term things such as where the next customer is coming from, why the last one left, and how much more the ones in between might be persuaded to pay. This is all, up to a point, useful stuff, and often makes for the creation of some very detailed and no doubt pretty reports. The problem is that the data is almost always used externally, and hardly ever used to carry out an internal review of the company itself.

The buried problems that good use of data can reveal can often be huge, and there are, fortunately, examples of companies getting it right. One of these is the toy giant Lego, which to most outsiders (given how ubiquitous it is) looked in 2003 like a perfectly healthy toymaker, selling around one billion euros’ worth of toys a year (British retailers had named its bricks ‘the toy of the century’ only a couple of years earlier). Underneath, Lego was facing a catastrophic financial crisis, carrying over 800 million US dollars in debt and sitting close to total collapse. The company was losing hundreds of thousands of euros' worth of value each day, partly because nobody in the design studio was expected to ask what a new shade of plastic, or a new mould, would actually cost (as such, moulds were made, at some cost, for pieces used in just one particular set that might not sell). When Jørgen Vig Knudstorp, who took over as chief executive in October 2004, had the data the firm was sitting on analysed properly, he found that just 30 products generated 80 per cent of sales. Two-thirds of its 1,500-odd stock-keeping units were things it no longer even made. It bought from more than 11,000 suppliers (nearly twice as many as Boeing uses to build aeroplanes). Knudstorp dismissed the fashionable explanation, that children had decamped wholesale to video games and were no longer interested in physical toys, as “a blame game”. Knudstorp, the first non-family member to run the firm, then cut the number of unique parts it produced by roughly half, sold big stakes in its Legoland theme parks, began deliberately targeting adults as well as children, and returned to profitability in 2005, where it has stayed ever since, overtaking, in 2014, Mattel to become the largest toy company in the world by sales. The turnaround took little more than taking a good, hard look at data that had been there all along.

Firms that do not put their data to work will probably not survive in the long run.

Returns are perhaps the most under-read data of all, in part because they arrive as a parcel rather than appear on a spreadsheet (and this correspondent confesses to having contributed his share of those parcels). Every parcel sent back is, in effect, a signal, and usually a very specific one: wrong size, wrong colour, not what the picture showed. According to Akeneo, a French firm that sells product-information software, 43 per cent of consumers returned something in the previous year because what they were told before buying turned out to be wrong, with sizing the usual culprit. Romain Fouache, its boss, says that many firms “still haven’t connected the dots between product data quality and return rates”.

ASOS, a British online fashion retailer (which knows a thing or two about parcels coming back), has connected some of the dots. In October 2024 it began charging customers who send back most of what they order, and in January this year it went even further, showing shoppers in the UK their personal return rate in its app (a sort of credit rating for buying clothes), with warnings as they approach 70 per cent, beyond which it begins charging a fee for returns. Ben Blake, an executive vice-president, says this gives customers “greater visibility and control”. Perhaps it does, but it is also hard to shake the feeling that years of returns data have been put to work disciplining customers rather than interrogating what might be wrong with its trousers.

Data can make money, too, rather than merely stopping it leaking away, although not usually in the way that tech firms often think. Barbara Wixom of the Massachusetts Institute of Technology (MIT) and her co-authors found in a study that, on average, just 18 per cent of the returns from data come from selling it (the traditional way of thinking when it comes to data). The rest comes from using it to run the business better (51 per cent) or to make existing products more appealing (31 per cent). Firms that do not put their data to work, she argues, will probably not survive in the long run.

Craig Turp-Balazs

Craig Turp-Balazs is head of insight and analysis at Reinvantage.