This short case study tells the story of how Tesco and its marketing data analytics consultancy partner Dunnhumby used Big Data to re-engineer the UK Supermarket Sector in the 1990s and 2000s. Some years before the term ‘Big Data’ first entered popular usage.
Timing and background
The year is 1993. Four years after Tim Berners-Lee proposed the concept of ‘the Internet’ and five years before Larry Page and Sergey Brin founded Google. Whilst the enabling technology for the Internet had been in place for many years, it was Berners-Lee who first started to demonstrate the capability, to meet his own project requirements to share information. His proposition was simple: Create a technology platform which in concept could allow any data to be shared between anyone or anything. By 1993, bandwidth was still limited and error-prone, and processing and server capacity was limited and – relative to today – expensive. But early adopters like Jeff Bezos with his Amazon ‘bookstore’ had already seen the opportunity[1]. Investors, analysts and technologists were similarly excited: What would a world of effectively limitless and frictionless connectivity, data access and developing processing power look like?
The UK supermarket sector
In 1993 the UK supermarket sector was led by Sainsbury’s with the highest market share, followed by Tesco, Asda, Morrisons and Waitrose respectively. There were also many small independents, and Aldi and Lidl had yet to enter the market. Whilst the players had different brand positionings[2], in reality the product offerings were quite similar, customers had almost zero switching costs and – crucially – customers had plenty comparative data[3]: They could simply switch to a competitor down the street if they were unhappy. As a result, margins were thin and competition was fierce. Advertising was via traditional TV and press channels and adverts, billboards and Public Relations.
The opportunity
Terry Leahy, then marketing director at Tesco, knew instinctively that the Internet would bring step-changes for retailers like Tesco. So, he asked his marketing team to investigate potential loyalty card implementations and to review the existing, largely non-automated, market. Again, loyalty schemes were not new, but Leahy and others knew that some combination of Internet connectivity (data access) to link customer purchase actions and habits electronically and really quickly – in ‘real-time’ – and to respond to them with flexible and responsive product promotion and targeting could yield returns.
Joining the dots
Clive Humby, a mathematician, and his wife Edwina Dunn had set up their marketing data analytics company in 1989 and were already working with companies like electronics and communications giant Cable & Wireless and car company BMW when they were ‘spotted’ presenting at a marketing conference by Grant Harrison, the lead project manager at the time for Clubcard. Dunnhumby was invited to partner on development. Tesco was ambitious to see its new Clubcard venture work – or at least learn why it would not – so the company, with Dunnhumby moved forward quickly with a low-risk pilot in five stores, soon to be 14. Leahy and the Board were also keen to seize first-mover advantage; whilst the early internet had many weaknesses, they knew that by dint of simple human endeavour and natural evolution that it would in time mature into a standardised, reliable and cheap data communications platform.
Secret sauce – tuning the numbers
The idea behind Clubcard was simple: The more that customers spent, the more they would be ‘rewarded’ and ‘thanked’; the ‘thank you’ was a key part of the marketing and promotional strategy often forgotten by commentators and analysts. Clubcard catalysed a kind of ‘virtuous circle’ effect similar to the so-called Amazon Flywheel concept. Being both analytical and a mathematician (he had previously worked at the data consultancy CACI) Humby knew that he had to test and hone the results, particularly how much money to give off on promotions. He worked out that 1% would work and stimulated repeat buying but that 2%, at the time, actually caused the project not to pay for itself. He was also well familiar with the ‘multiplier effect’: Just one more purchase per visit per customer would result in millions of increased turnover. And he also knew that turnover was not the important point on its own – what mattered was growth and the opportunity for increased revenue – increased profitability and growth numbers were what the City was looking for in its stock valuations.
Integrated marketing
Another aspect of the Clubcard model was that traditional ‘snail mail’ closed the promotions loop – customers were sent vouchers…they went shopping…they were sent more vouchers…they went shopping again. Tesco also knew that they had to invent new shop models, particularly smaller format Tesco Express stores, that catered to local customer catchments and demographics. The key point here is that Tesco understood that its operations and marketing needed to work as an integrated end-to-end process. Another case of joining the dots.
Implementation
Developing and rolling-out Clubcard was a classic ‘Advanced Project Management’ exercise: A lot of unproven Point-of-Sale (PoS) technology; unknowable – but partially predictable – responses from customers; the need for dynamic pricing in order to manage revenue-against-growth-against margins; the data ethics concerns and regulatory/legal risks of mass-collection of customer data; printing and distributing millions of plastic cards; maintenance and update of customer databases. With all of this complexity some may wonder how the project succeeded: The answer was: (i) Professional, proven, experienced project management resourcing; (ii) Chunking-down the work to manageable and self-contained sub-projects and, (iii) Clear communications both within sub-projects and appropriate cross-project information and data exchange.
Being data-driven
Tesco and other organisations will be clinical in keeping base-level data clean: Customer address; purchase history; purchase-location history. Based on this simple initial set of data it is relatively easy to infer additional meta-data (data about data) such as individual customer demographics (age band, income band, gender, education level etc) and so-called event information (births, new homes etc). But even with this information, it is no mean task to determine exactly what could and should – and should not – be done with this data. Understanding the data soon becomes beyond the capabilities of traditional computer processing, even with sampling and the availability of the most sophisticated toolkit of standard statistical tests. Hence the need for so-called ‘AI’ algorithms and machine learning – proprietary information that Tesco and others keep closely guarded for good and valid reasons.
The outcome
Clubcard helped Tesco in three measurable aspects: (i) It helped the company leapfrog Sainsbury’s to become number one in the UK supermarket sector; (ii) It led to a dramatic increase in financial turnover, and (iii) It helped towards a three-fold increase in stock price[4]. The fundamental step-change that Clubcard yielded can though maybe be best summed up by a quote from Ian Maclaurin, Tesco’s Chairman in 1994, who, after a board presentation from Dunnhumby remarked: “You know more about my customers in three months than I know in 30 years.”
Clubcard now – and the future
Clubcard no longer provides such an advantage as it did when it was launched in 1995, but Tesco still benefits from the after-glow of first-mover advantage and the concentration of expertise and experience that engenders. The challenge now for retailers is how to rebalance the often dry online transactional experience with the physical bricks-and-mortar experience to provide a richer overall shopping experience for customers. This is a hard task for companies, society and governments even. But one that is now being investigated via ventures such as till-less shops (Amazon Fresh and now Tesco Express pilots) and other more far-reaching initiatives such as Facebook’s attempt to catalyse and lead market growth of the so called Metaverse.
References
“Scoring Points – How Tesco continues to win customer loyalty”. Clive Humby, Terry Hunt, Tim Philips (Kogan Page, 2007).
“Omnichannel Retail – how to build winning stores in a digital world”. Tim Mason, Miya Knights (Kogan Page, 2019).
“Big Data Demystified – how to use Big Data, data science and AI to make better business decisions and gain competitive advantage”. David Stephenson (FT Publishing, 2018).
“Business at the Speed of Thought – succeeding in the digital economy”. Bill Gates (Penguin Business, 1999).
Notes
[1] Bezos formally founded Amazon in July 1994, advising his early stage and family investors that there was a high likelihood of failure (!)
[2] It is a gross simplification, but Sainsbury’s broadly communicated ‘quality’ whilst Tesco sold itself on the idea of reducing the cost of the weekly shop through the message of ‘value-for-money’: Quality at a similar level to Sainsbury’s at a price comparative to Asda. This was later developed via iconic advertising popularised in TV advertising by famous English comedian and actor Dudley Moore.
[3] Customers’ could compare the costs of their weekly shops item-by-item. And as ever there was regular sharing of information about price hikes or discounts and offers between friends and between neighbours ‘over the garden fence’.
[4] The increase in stock price was also partially attributable to other factors such as general economic growth in the UK at the time and the so-called ‘Internet dividend’ discounted into many stock prices.