Points for personalisation
Loyalty points for Sainsbury’s and its partner businesses are earned through the Nectar programme, and these points are then redeemed for discounts, special offers, and similar incentives. By utilising the data, Jardine and his team were able to identify areas of growth and influence shopping decisions for millions of people across the nation that would positively impact their health and habits through personalisation.
In the UK, nine in ten customers now have at least one loyalty card, with many more owning multiple loyalty cards. For Sainsbury’s and its customers, the Nectar programme earns over £200 million in value via points annually, with a Nectar card being used 32 times every second – this puts Sainsbury’s in a unique position to identify real-time shopping habits throughout the year with a high degree of accuracy.
Through investment of £70 million in personalised pricing algorithms, Sainsbury’s utilised machine learning to tailor product recommendation personalisation and pricing suggestions for their customers.
Award-winning personalisation
Sainsbury’s has gained a lot of experience in building algorithmic machine learning customisation engines, and the data team wanted to take that experience and knowledge that is traditionally commercially focused and apply it to another significant challenge: national health.
In the UK, two-thirds of adults are classed as overweight, which is a serious concern for long-term health and the impact on the NHS which spends millions every year tackling weight-related health issues. Jardine explained that it was a challenge that Sainsbury’s felt obligated to tackle, with the company making the commitment that 85% of its food sales on volume would be on items that are healthy or better for consumers.
Jardine and the data team explored the choices available around health options when it came to customers selecting their groceries, noting that the existing system was relatively impersonal. People make different food choices for different reasons, and this meant Sainsbury’s needed to take a very granular approach to personalising shopping experiences for customers through automated machine learning algorithms.
The first challenge was identifying what Sainsbury’s customer guidance journey looked like, followed by getting customers connected to multiple touch points to help guide the journey further. Despite sounding straightforward, Jardine explained the process was far more nuanced than expected. One major hurdle was understanding what constitutes a healthy product – the UK government has a classification system, but it was considered a binary system by Sainsbury’s wanted a detailed approach; for example, many fruits have high sugar content, so what metrics should be used to determine healthiness?
The Sainsbury’s data team worked closely with the nutritional team and existing healthy eating guides to understand the existing data and identify areas of improvement. This was then developed into a scale of healthiness and the healthy eating automated machine learning algorithm for personalisation. There are more than 25,000 products across 17 different categories that are in a constant state of evolution, with products being added and removed, that require classification.
Due to the scale of the project, this could not reasonably be handled manually, so automated pipelines needed to be built to make the process seamless. This was designed in such a way that as soon as a product is added, it is scored and placed to sit in the healthy eating programme.
Once this was achieved, the data team were then able to apply the project across the customer base to better understand what they are buying, their basket compositions, key food groups, and how Sainsbury’s could provide the opportunity for better food choices. This led to a stronger understanding of what was preventing customers from certain choices and how Sainsbury’s could bring better balance to customer options.
Three main areas were uncovered:
- Cost
- Inspiration
- Time and effort
The assumption was that budget was the main hurdle for most people, but the answer was more nuanced. With time and effort and inspiration being concerns, the suitability of ready meals became an important factor. This would then be coupled with brand pricing hierarchies to help develop the most suitable suggestions and personalisation.
The data team noted that this was not an exercise in encouraging customers to buy large quantities of fruits and vegetables, but rather an opportunity to help people achieve a balanced diet that works for them, their schedules, budgets, and circumstances.
Integration into the customer journey
The data team then had to action the data-driven recommendations into a seamless customer journey that reaped the desired results.
For those shopping online, the data indicated most customers were habitual in their grocery shopping journey. The data team used this as a starting point to evolve the product recommendations to begin influencing the repetitive purchases they would make. By raising the visibility of certain products with healthier scores over those with unhealthier scores, the data team was able to give people more prominent healthier options. The result of this action was that the average product health score increased two time over due to the recommendations.
Gamification was also utilised just getting people to download an app was going to be inadequate. To make a lasting change, the data team needed to make people want to be engaged with their journey. This was achieved through providing snippets of information in a yearly roundup, such as informing people that they were the number one purchaser of a specific item in their town. The added benefit of this gamification was that customers would share their results on social media, further enhancing the reach of the project.
The other aspect of the challenge was to tackle the issue of inspiration for exploring new recipes, ingredients, and flavours. This was achieved by recommending healthy alternatives that aligned with previous purchases, but were varied enough to change up the result, such as recommending butternut squash or swede.
The data team took the gamification factor even further by introducing a fruit and vegetable challenge where customers were encouraged to add more portions of fruit and veg to their baskets and extend their repertoire by trying new items. This would be personalised for each customer.
Furthermore, the data team can take a combination of the customer loyalty score and how sensitive they are to the price of other products known to be liked and then send up to ten different products that are personalised. This has resulted in more than 260 million personalisation offers every single week.
Making a difference
Sainsbury’s innovative approach has resulted in a 200% improvement in product health scores and a substantial 130 million portion increase in fruit and vegetable sales. The way this programme has worked now means that Sainsbury’s can truly invest in the right products for the right customers and the right amounts.
Jardine emphasised that the journey was not about implementing a huge change, it was about presenting options to customers to improve their awareness and provide educated alternatives. The way in which the project was delivered improved health for customers as well as improved sales metrics for the business, making it a win-win situation thanks to data.
Looking forward, as a 155-year-old business, Sainsbury’s is now prepared for a data-driven future at scale. The team worked diligently to build a separate set of machine learning pipelines, real time adjudication system, real time wallet capability, notifications, and more, which sat alongside a series of legacy technologies, but was tightly integrated.
The final point Jardine made to the audience was that it is essential for the team to get out of the office and engage with customers where appropriate. He explained that it is easy to get stuck behind a desk and swamped in algorithms when you have access to such a large pool of data, but getting out and having conversations with active customers can bring insights and ideas that have been missed. It is here where some of the most poignant personalisation wins can be made.
Get involved with industry discussions here.
