Blog
Traditional retailers are in dire need of price optimisation. The help they need is finally at hand.
![](https://faculty.ai/wp-content/uploads/2023/02/Featured-223521.png)
Tesco achieved 7.2% retail sales growth in the last financial year – sounds pretty good in the current climate, right? It’s certainly a lot better than many other retailers will be seeing right now. Tesco rightly celebrate that they’re the only traditional retailer to achieve share growth over a 3 years.
But there are two unignorable problems. Firstly, the growth is far from organic – volumes are down by around 7%, more or less neutralising the benefit of price increases. Secondly, Tesco have made a £1bn investment in price – holding back some of the inflationary pressures passed on by their suppliers. This results in a 50% decline in bottom line profit. Spare a thought for retailers with similar inflationary pressures but without the top line growth.
Tesco laud their claim to be the most competitive that they’ve ever been on price. But it’s costing them, and every other retailer. Inflation will ease but the competitive dynamic won’t, and no-one can afford another £1bn to make sure they’re as competitive as possible on price.
Say it quietly – retailers need to find a pricing strategy that isn’t just based on being cheaper than the next retailer. Margins are going to have to be restored. There’s not a lot of fat to cut. So what’s the answer?
Price optimisation is a reality – for some.
The situation calls for net revenue management, with price optimisation front and centre. It’s not a new idea – the relationship between price and demand, what affects that relationship, and how to model it are all well researched areas. It offers the prospect of small changes to pricing that add up to big gains in margin.
Digitally native e-commerce retailers are frankly way ahead in this regard. Amazon uses analytics to optimise pricing for market to market conditions in real time. Uber combines willingness-to-pay with supply and demand analysis to optimise their pricing second by second. Any retailer with this capability is well placed to tough out difficult times and manage market competition effectively.
But what of traditional brick and mortar retail? The picture is not so rosy, with pricing decisions often left to judgement within rules on minimum percentage margin and gap to competitor pricing. This results in at best reactive pricing, and at worst cedes control of price to the competitive dynamics of the market – resulting in a constant downward pressure that retailers can no longer afford.
Almost all traditional retailers have attempted some form of price optimisation, only to find it either didn’t work or never got used. It’s perhaps unsurprising that there’s a lack of faith that price optimisation is worth pursuing again.
The trouble is, they can’t afford not to – getting price optimisation right can unlock 300 basis point EBITDA growth.
It would be easy to assume that traditional retailers are somehow behind the curve; unable or unwilling to keep up with technological advances. But the reality is that the incumbent price optimisation solutions on offer today – be they price elasticity studies, virtual store experiments, or black box AI – are not fit for purpose.
Why price optimisation in traditional retail has failed…so far
The first step is to be clear on what the shortcomings of attempts at price optimisation have been.
1) A poorly conceived methodology: Price optimisation has previously been treated as a vanilla regression problem i.e. training a model on historic price and sales data, and using it to predict the future. To understand why this is a problem, consider two facts of traditional retail. Firstly, prices in traditional retail don’t tend to change much – there is very little pricing signal in the data. Secondly, in portfolios of thousands of SKUs, complex interactions mean that a price change on one SKU might affect the sales of others. Naive application of a regression algorithm will fail to account for both problems, and the result is poor, potentially dangerous, recommendations.
2) The outputs aren’t useful: Approaches to brick and mortar price optimisation have ranged from little more than high-level insight to sku by sku recommended pricing. The former clearly doesn’t provide the granularity needed for line by line price optimisation. Even if accurate, the latter approach very often fails to explain its recommendations. This is unacceptable to the business users who – when faced with the potential risk of getting pricing decisions wrong – tend to work around or ignore such systems.
There is also a fear that retailers should face and overcome – that optimising price will compromise their competitive position. This is NOT what good price optimisation is about. Rather, it should be conceived of as a balanced approach of finding margin where price is less sensitive, and investing it where competitive pricing is crucial – resulting in both happy customers and healthy margin.
Most importantly, the decision makers and domain experts in retail businesses should remain in control – and this is where a solution that is more than just a model becomes important.
Decision Intelligence in price optimisation.
Decision Intelligence brings together cutting-edge AI algorithms with seamless user interfaces that are focussed on enabling human experts to make the best decisions extremely rapidly.
How can this approach transform price optimisation? We firstly need to find the right AI for the challenge. This means a sophisticated ensemble of techniques that can carefully extract pricing signals, understand and utilise a network of product similarity, and reliably search for complex optimisation of multiple objectives.
It then requires a deep understanding of the requirements of the pricing decision making process. Pricing is a constant decision process for retailers, it’s useful to conceptualise it as a Decision Loop.
- Observe: ingest and decompose the drivers of your current sales performance
- Understand: users must be able to see and trust how a model estimates future performance
- Decide: find pricing that meets multiple commercial objectives. Flexibly examine scenarios.
- Act: integration of decisions into parallel processes and systems.
Intelligence at every stage of the decision loop, for every iteration of the loop, creates compound improvement over time – offering retailers a new and enduring growth lever.
Just a model – even a really good one – isn’t enough to enable price optimisation. But with Decision Intelligence, retailers will be able to cyclically manage their pricing in order to make pricing the growth lever it should be.
True price optimisation is now within reach for traditional retail. And it couldn’t come at a better time.
Joe Mills is a Pricing and Revenue Management specialist in our Retail & Consumer team. If you would like to learn more about our intelligent price optimisation capabilities – please get in touch.