A guide to AI retail pricing strategies

Transcript: A guide to AI retail pricing strategies

Notice my shift from data science to customer facing, because I realized I’m a lot better at chatting than I am coding! And also, I just love getting to know the intricacies of businesses and how they could benefit from artificial intelligence (AI). And I think that comes from the struggles and the stresses that I had in my merchandising career.

I was constantly putting out fires, constantly doing a lot of manual pulling that I had no time to even think about how to strategically grow my area. And I think if I actually had AI back in the day, for one, I would have worked a lot shorter hours. But two, I would have actually had a lot more love for it, and I hope that comes across today. So let’s get into it.

So, pricing: the amount of money for which something is sold. Seems pretty simple on this screen, but we all know it’s definitely not. Different businesses, different industries have different pricing decisions that they need to make, different levers that they want to pull with pricing, that even unpicking this for a business can seem quite complex.

So, ultimately what makes it into the boardroom, pricing allows you to be more profitable, drive more value and also attract and retain your customers. But that’s quite a lot.

Again, the world is still quite chaotic. But I’m not here to doom and gloom this session because we all know it. But the underlying message here is it was really chaotic and it’s getting less chaotic. So how do you really navigate the past three years? What about when the new normal is over the next year, when all of this is constantly changing, and that all of these things really play a role in pricing profitability.

You’ve got inflation and how that affects you and your consumer base. You’ve got cost pressures. Your cost pressures are lessening in certain industries in certain areas. So is there an opportunity to actually pass that into your customer base and give them lower pricing, or is there an opportunity to have higher margins again?

Also, shape-shifting consumer habits. The cost of living crisis, online transparency… as consumers we want to perceive value for money and also, in a lot of cases, we want more and more for less. So getting to know your customer who’s in market and what price they’re willing to pay is getting even more complex.

Also, complex networks; businesses have grown arms and legs. You might have acquired different businesses, gone into new sales channels, with new pricing models, complex networks of suppliers, factories, distribution centers and stores — all with their own costing. We speak to a lot of businesses that have grown so quickly that they actually didn’t have the processes underpinning it to be able to control prices across all of it.

And also, excess inventory everywhere. Over the last couple of years, it was hard to really gauge those demand and supply signals for a lot of businesses. And that means that they’ve ended up with a lot of aged stock and stock in the wrong places.

I wanted to drill into the challenge that your teams are facing when they’re trying to price optimally. As a business, you can have tens of thousands of SKUs all with different attributes and landed cost prices. You can have five million plus daily transactions, all with different basket sizes and discounts applied. You can have four different sales channels, all with different pricing models and strategies. You can have eight global markets and those global markets have different cost sensitivities and different cost bases. You can have hundreds of branches, all with different stock profiles and their nuances. You’re setting thousands of marketing campaigns trying to get your customers pointed towards different products. And you’re making tens of thousands of yearly price change decisions.

If you really tally all of that up, you’re really considering billions of permutations for a business to do this optimally. And that’s absolutely crazy to put that on a screen.

So businesses are having to do all these billions of permutations and navigate the chaoticness — still in Excel and legacy systems. I used to pride myself on my Excel skills, and love the complex Excel sheet. But we all know that it has its own limitations when it comes to data pulling and transformation. You’re limited in the time that it takes to pull it and transform it so you can make decisions.

Also, it’s limited to a million rows. And for big businesses today, that doesn’t just quite cut it. And still, as soon as you pull the data, it’s out of date.

So people are making generalized product decisions across maybe multiple categories because they can’t really get into the nitty gritty because they just don’t have time for it.

Also, to go into systems as well. This is a screenshot of a system that I had to use in Topshop back in the day. And you can see it’s a mashup of IBM and internal IT resources. And you can see it’s quite a jarring mashup of Pacman and a digital clock.

I was told I was going to love it. I never did. And we all know that ERPs are quite rigid and you can’t really quite get what you need from it. And you spend your days flicking between screens and copying and pasting and copying and pasting. And that is what everybody’s after. But just to caveat, I’ve met some rock star people that are able to do some crazy things in Excel and these systems, but we want to enable them a little bit more.

So how are businesses dealing with all this chaos? So we speak to a lot of businesses. We have a vast partner network. So how have people dealt with the chaos over the last couple of years? We’re finding a lot of people doing short term to stabilize their cash flow on their P&Ls. And our VP of strategy, Chris Ashley, has nicely penned these, the sticking plasters of profitability, which I’m going to take you through today.

So starting with reaction pricing, promotion and markdown. We’re seeing, with a lot of businesses, that they’re reducing their prices to counteract failing customer demand to try and meet their sales revenue targets or to clear through excess inventory.

Also, on the other hand, people are having to raise their prices due to their ever expanding cost bases. But we speak to a lot of businesses that are having to make this at a top level, based on assumptions and no scenario planning to know how that might affect the rest of their chain.

So you can see here from Mace and Pret that they’ve had to do this themselves. You can see the Pret headline is a little bit more sinister, but I’m that mug that’s still got a Pret subscription even though they’ve raised the prices — so they must be doing something right!

Also, you can see here, cheese prices — one we can rejoice about is that cheese prices have been slashed in latest supermarket cuts. But around nine months ago, we spoke to a supermarket that was spending millions of pounds on reducing their prices, but openly admitted that a lot of it was based on top-line assumptions and Excel transformations. And actually, they didn’t even have anything in place to be able to monitor if that was going to be effective and how to change it. Everybody is in this constant cycle of reaction pricing.

Also, there’s a known trade off curve between price and volume. It’s a really hard one to navigate for your business. And we’re finding that we’re seeing a lot of businesses having to put up their prices but not really understanding how that affects their volume. So people are going up against dwindling volumes, which is really hard for long-term profitability.

So now they’re going in and doing really big marketing campaigns to try and get that customer base back. Or they might be relying on reaction pricing that I just was talking about before. So now you’re in this constant cycle of putting up prices, putting down prices and not really knowing how to stabilize that.

Also, people are doing a lot of cost-cutting initiatives. We’ve all felt it, especially me when I was at the Titanic that was Topshop, that things go. So DCs go, stores go. Also, you cancel product orders because actually you can’t afford to bring it in, or it could be that you’re stopping investing in tech. But in order to run leaner businesses more effectively, you need to start investing in something that makes your life easier.

So if you take anything away from this speech, I’d like to take you to take away this. So McKinsey did a bit of research on 1,200 publicly owned companies and evaluated the effects of different strategies on either returning, maintaining or growing in profitability.

And they concluded that those that invested in pricing strategy, upped their margins whilst maintaining volumes, and had a longer term sustained ROI than any of those sticking plasters that I detailed before. And this is pretty massive. But they also said there was such a small number getting started on their pricing journey, and we find that a lot too. So, why aren’t businesses getting started on their pricing journey?

How can AI to improve retail pricing strategies?

So we find that a lot of people have a lot of complexities in the way that they work — chuck in all those things that I detailed on the chaotic slide — and it can be quite overwhelming to even start on this journey. And that’s why a lot of people were relying on the same way that they priced for a couple of years, or reverting to those reactionary pricing moments.

Or it could be organizational change. So we’ve all probably had to endure the implementation of a new ERP, a new warehouse management system or a new point solution, and the sweat that’s coming down your face trying to figure out whether you can run your day-to-day businesses on it.

Or, three, is an uncertain investment. Ultimately, when times get tough, you’ve got less money to spend on systems. You need to know it’s going to return on its investment. I’m going to try and use the next part of this to debunk all of these and show you why artificial intelligence (AI) is the answer.

AI can handle the complexity so you don’t have to. I said before that a lot of people are working in Excel sheets, and that has its limitations when it comes to data. AI in a productionized environment can have automated data feeds, a robust data pipeline joining it all together and advanced cleansing techniques.

That means that no data point is out of reach and that means that you can reduce silos across different systems that you would want to feed into your pricing, but just can’t get to. And it also gives people an instant data source so they can start making decisions. And that’s insanely powerful in itself.

Also, we find a lot of businesses are having to make generalized pricing decisions across categories, but AI can handle granularity so it can look at every product location or customer group in isolation. This is AI prediction at scale. Also, AI gives the ability to understand the true effects of pricing on demand.

It can simulate hundreds of price changes on each of those products and also output the metrics that it would affect, so things like margin and volume. This is insanely powerful if you want to look at different pricing scenarios across your product range. I’m going to go into this in a little bit more detail in a couple of slides because this is the really cool bit.

Also, AI gives you the ability to optimize against your target metrics. Meeting your metrics has never been more important because you need to maintain that.

This just puts a lot more power in your team’s hands, and it also takes a lot of the risk out of actioning price changes as well. 

AI can also find relationships between products. What we hear from a lot of businesses is, “oh, I actually haven’t actioned a lot of price changes on certain products”, or “I have a lot of newness, how can I really truly understand how price will affect those?”

AI can find similar products through hierarchies attributes and then apply the learnings from that to them, which is some serious power. Also, it can see if you reduce a price in one area, how it might another — and that means that you’ll really start to balance pricing decisions across your range.

Also, dynamic agility. AI is an always-on application. It’s not a one time thing in an Excel sheet. AI learns in light of new data; it’s constantly adapting, it’s constantly learning. So if your consumer behavior changes, i.e. your customer base starts reacting to different products differently or different price points or different promotions.

Or it could be that those external factors are affecting your industry a little bit more. Or it could be your cost prices have decreased or increased in certain areas, or you could have a suboptimal product mix and you need to clear through it. AI can see all that and adapt accordingly, which gives you agility in uncertain times.

So I’m going to dig a little bit more into why understanding the effects of price is really quite difficult, and hopefully break that down here. So we’ve got quite a simple graph here that’s mapping demand volume against price.

In a simple world, you would think that a low price equals a high demand and a high price equals a low demand. But in reality, for most industries and product portfolios, it’s really hard to understand this relationship.

So here, the same axis, but we’ve plotted one product’s daily sales volume and the average selling price against it. And as you can see, there’s quite a loose correlation here. And this is what your team’s looking at in Excel sheets and trying to really figure out how you can pull the price in. And how you can pick a point on that line that is most profitable. This is because there are so many other factors that can contribute to demand, other than price.

So this is starting modeling those factors against price. So it could be stock levels that last year, that product, you had a severe supply chain delay on it. So it wasn’t optimally stocked across all your stores, and that equaled the lower demand. It could be that it’s a really seasonal product, it could be that your industry in certain areas is really struggling because of the economy. It could be that you were insanely promotionally-driven to clear through a lot of stock last year, but you’re trying to wean yourself off of that. It could be that there’s lots of product cannibalization across your range.

So when you really start to model this, this is when you can truly understand the pricing effect. But only really AI can do this for you — and this is AI-modeled price elasticity.

And it could be a case after you’ve done all this, and after you’ve factored in all of these factors, actually price isn’t seen to be a huge lever on demand — it could be that by discounting it, you’re needlessly eroding margin that you could save. Or it could be after you’ve stripped out all of this, pricing is a huge influencer to demand. So that gives you a lot of opportunity when it comes to pulling the strings of pricing.

I know this is quite abstract on this screen, so I wanted to delve into it a little bit more.

So this is one product, one location. This is a visualization about all the factors that AI have found to be contributable to demand. And it’s also applied a weighting of that factor as well. So you can see there’s other factors like marketing, seasonality and website placement. This is really hard for somebody to be able to unpick this themselves, especially when every single day that this product has a different combination of all these factors. It’s really hard to strip out the effects of pricing.

Back to granularity; this is one product, one location. But it could be a product in a customer group. Every single product and customer group will be affected by these factors in different ways and it will be constantly changing. So this is constantly adapting, so you can pick the best price point that allows you to meet your metrics.

Also, the next one I want to debunk is that people believe there’s too much organizational change.

So, ultimately, your teams are still going to have to make pricing decisions based on data — but what if they could get an instant data set that had all the data points and the granularity that they always desired. They had access to AI demand forecasts, so they could start being more proactive than reactive. All of that price elasticity modeling, as well as price simulations across your whole product range and how that rolls up to your target metrics.

This just gives people a lot more fuel to feel more enabled to make those pricing decisions. And also, like I said before, every business is different. You’ll have different logic, different rules, different guardrails — and you are able to constrain AI so that when your team gets a pricing suggestion, they trust it every time. And also a big old wrapper of explainability. Every AI needs explainability so that your team starts to trust it. But then when they do, this is when the serious value arrives.

So the final one — it’s an uncertain investment. These are the metrics that we are consistently seeing when the AI pricing space. This is across a multitude of industries and different business models. So we’re seeing a one to five percent gross margin improvement and, for some businesses, that has meant multi, multi millions of pounds.

Also — for those businesses that needed to use pricing to pull their stock lever and reduce aged stock in that longtail — we’ve seen consistently an improvement of over five percent increase in volume.

You ultimately need your teams to love it and trust it; 100% pricing suggestions accepted. And I put a little asterisk here because, ultimately, it takes time to get to that point. It might be that we need to wrap it in more guardrails, more logic, also enabling teams to use it. And then you can get to this point where you’re really trusting the AI.

So the takeaway from this session, I hope it’s been useful, but you’re still going to have to navigate some chaos, and that’s undeniable. But you could stop relying on the sticking plasters and start utilizing AI pricing power.