AI-driven pricing is transforming the way businesses set and adjust prices in real time. But how do you know if your company is ready to leverage AI for pricing optimization? In this on-demand webinar, we’ll break down the different types of AI-powered pricing strategies, where they work best, and how they can drive revenue and margin improvements.
Key takeaways include:
- Identifying the right AI approach for your industry and business goals
- Assessing if your data is ‘AI-ready’ for effective pricing optimization
- Practical steps to implement AI pricing for maximum impact
Whether you’re new to AI-based pricing or looking to refine your strategy, this session will equip you with the knowledge to make informed decisions and drive success.
Speakers:
Dominic O’Regan – Senior Strategic Consultant, PROS
Pol Vanaerde – Founder, EPP Pricing Platform
Full Transcript
Good afternoon, everyone. Welcome to another EPP webinar. Are you ready for AI based pricing and lock the power of AI to optimize pricing and maximize profits? My name is Rita. I am a member of the events marketing team at EPP. And today, it’s a great pleasure to have with us Pol Vanaerde, founder of EPP pricing platform, and Dominic O’Regan, senior strategic consultant at PROS. Before I hand over, I would like to explain to you how this webinar works. We’ll be answering questions questions live during the presentation.
…
So feel free to submit yours at any time by clicking the q and a button on the control panel. This webinar will be recorded and will be available afterwards on EPP Prime for EPP Prime members. Thank you for your attention, and now I’d like to hand over to Paul and Dominique. Thank you.
Thank you, Rita.
It’s all about AI and the readiness about AI and pricing that we’re going to discuss, myself with, Dominic. Dominic, short introduction from your side, maybe. What’s your role in, at PROS?
Sure.
So strategic consultants, at PROS, we’re like, we’re like a presales function. So we primarily support our sales teams and our sort of partner marketing teams and things like that. And and our role is to really understand what pros can do, the sort of the detail of our capabilities, and work out how that might be applied for particular prospects and businesses that we talk to or partners that we’re kind of working with, for them.
And I’ve been working for PROS now for about six years, I think, roughly speaking, rounded up or down, wherever it is. And, and prior to that, I’ve worked in other pricing, and CPQ solution, providers. So my understanding is a little bit broader than just, pros and the way that we go about things, which is always kind of handy, if you like.
Absolutely. Yeah. I see.
Have you Probably more like fifteen years.
I don’t know. Something like that.
Okay. Good. Good. That’s good to have you here. We we would like to discuss four things, if I remember well.
It’s about, yeah, the enabling your tools, to be AI ready, to set, value based prices or embed your pricing strategies with with AI. Mhmm. We’re going to discuss, the practical steps to implement.
We’ll discuss the, the famous question, should we start small or big? And finally, of course, we’re going to discuss what are the most important enablers.
Dominic, let’s start with, the first, yeah, and and a very important question, I think. How do you know that you’re ready?
Yeah. Yep. Yep. I think it’s a I think it’s a good question. And and you almost have to answer the question by looking at a slightly different question that will help you to work out if you’re ready and and where you might be, ready.
So in in a way, you kinda have to start by thinking about how can AI be used in the context of pricing and price setting or price optimization or what whatever you want to you want to call it. And and that kind of requires people to think a little bit about the different sorts of AI that are out there and how they might be applicable in their particular business context. So I have only two slides to share in our in our chat today, and the first one is to help kind of illustrate exactly that.
So I just took a a selection of some of the most common AI models that we see in pros, and there are other AI models available both in pros and and and in general. Right? But but I thought I’d focus on just some of the ones that we see on a regular basis or that we use on a regular basis just to illustrate some of the differences. Because I do think that the first step is understanding, is there a model that would be a good fit for the sort of challenges that I’m that I’m facing? Right?
So as a as a as a sort of an way of introducing them, I would actually kind of start with the revenue management context. And I’d start with that even though it’s the one I personally know the least well. But I’d start with it because it’s perhaps the oldest in terms of, pricing AI. The concept of revenue management in things like airlines has been around probably longer than any other application of pricing AI. So it’s almost the daddy of pricing AI, if you like, at least from an age perspective.
And and at a really, really high level, what revenue management is doing is balancing, algorithms that try to work out what a customer might be prepared to pay. Right? That’s always gonna be important. But they’re balancing that against some sort of a forecast that is driving the expected kind of, opportunity and therefore working out, if I sell something today, would I have been able to sell it for more if I’d held out and waited?
Things like that. So it’s kinda balancing those two, two things. And it’s very, airline specific, and the label revenue management is very airline airline specific. But it’s just one one example.
And like I said, one I probably know a little bit less than the than the others.
But but closely related to it is something that people call capacity aware optimization.
And you can think of that as being a little bit similar to some of the elements of what happened in revenue management in airlines, but not, designed necessarily for an airline context. So where you have a single opportunity to sell something in the same way that a seat on an aircraft, you can only sell up until the point it takes off. Right? You can think of the same sort of idea in other business contexts. It might be things like theater seats. It might be other things like, other travel related industries like ferries and all sorts of other kind of, contexts.
But anytime that you really have a defined perishable item, if you like, then you can think about a capacity aware optimization. And the big difference between this one and the revenue management one is that the algorithms are developed so that you don’t need to create a kind of a separate forecasting, module. And that’s really important because that’s very difficult but possible in things like airline industries, but very difficult and often impossible in other industries.
So it’s kind of a a blended, approach. Mhmm. Mhmm. Now what I see much more on a day to day basis in business to business sales, which is where I spend most of my time, so where I have most familiarity, are the algorithms that that we’re looking at on the on the bottom of the page here.
And so I I we can think of those as algorithms that are designed PROS calls it ecommerce, but you can think of it as algorithms that are designed to set a sort of take it or leave it style price. So if I’m publishing prices on a website, an ecommerce site or something like that, or if I’m or if my the way I sell is simply to say, here’s my price. You take it or you leave it, then you can think of it as being the right sort of algorithm. And you and the core concept with a take it or leave it type algorithm is that the algorithm is trying to learn when I set prices at a particular level, what reaction does the market give me in terms of volume?
Now you might be targeting not volume. You might be targeting margin. You might be targeting revenue or volume, but it’s learning from the understanding of the reaction as a volume reaction. My price goes up. My volume goes down, and that leads to whatever margin impact. And is that good or bad and and so on. Right?
Scaling of dynamic pricing. Yeah. Right. Yeah.
Yeah.
Where we see probably the very biggest focus in b two b is in that last, corner on this on the slide where we think about negotiated, prices.
And and and, actually, sometimes I I’ve engaged with companies where they have a list price and then they negotiate. And nobody actually pays the list price. And sometimes they come to us saying, hey. We think we’d like to help with our list price. And we kind of end up saying, well, we can do that if you like.
But you’re gonna get no real value compared to working on the thing that people actually pay if that’s the negotiated price. So we do tend to see more focus in in negotiated than, than list or ecommerce.
No. That that’s an interesting thing. So your AI tool is helping on on, yeah, on real selling price point, let’s say, instead of optimizing the the list price, you optimize the the selling price. Is this is this identical for contract pricing versus project pricing, spot pricing, or do you have different is this a different model?
Or The the concept is the same, and the way that models work for spot pricing, contract project, for anything that is negotiated that is, kind of in a b two b context, the model is the same.
But Oh, and it’s interesting, Dominic. We got Yeah.
It it it reminds me at at the request of of a company recently that said, you know, but, one of our challenges is we have twenty percent spot pricing and eighty percent contract pricing or the other way around. Yeah. Have you used the same model, or are we stuck here? So your answer is it’s the same model.
So it depends what you mean by model in reality.
You you would use the same model concept, but you would actually have a version of it that is learning from and designing to do the spot pricing, and you would have a version of it that is learning and from and design and designed to support with the contract pricing. So you so it’s the same sort of model concept, but it would be a distinct learning model designed to learn to do that that individual job.
That that But both both are productized?
Did this Yes.
Exactly.
These models exist, maybe.
Yes. Yeah. Yeah. Yeah. Yeah. In fact, it is a single, for pros, it is a single, model type.
But if you feed it the right data, it learns to to to to price that negotiation context.
And I know we come data in our conversation anyway.
Right?
Yeah. No. We come to the point that you say, if you if you upload the right data, and and that’s the first question that we’re focusing here on. When do you know that you have the right data that that you’re ready with your data?
Because that’s always the the first question that’s right. Yeah. But if you don’t have the right data, etcetera. And I remember a quote from you a couple of days ago preparing this, webinar that you said, well, well, Paul, there’s no such thing as bad data for AI.
Can you explain me it a little bit?
Yeah.
Because it is an extreme statement, and and and it was designed to catch your attention, shall we say?
Okay. You taught me. Yeah. Yeah.
But so the thing with data is when people say that they’ve got bad data, sometimes they mean that there is missing data.
Mhmm.
And that’s one sort of, challenge for an AI to kind of deal with.
And sometimes what they mean is that they’ve got data that doesn’t reflect what they think is good pricing. And that’s a a almost a separate sort of bad data, challenge if if you like.
And and in reality, I genuinely think there’s no such thing as bad data. There might be data that you don’t want to impact how you price tomorrow, but it’s not bad data in the sense that it does represent something that you genuinely sold probably, if it’s in your data, and a price you genuinely sold at. It’s just maybe extremely rare or impossible that that reflects something you want to do going forward. So it really comes down to how good is the AI at understanding when data is something it should react to and when data is something that it actually, frankly, should ignore because it’s gonna be more accurate, more stable, more reliable if it ignores it.
And so k. The more sophisticated the engine under the covers, the more capable it is at dealing with things like sparsity, which is where there’s missing data, or with, misleading data, things that don’t represent what you actually want to do, going forward. And that’s where the more the newer learning algorithms are much, much more stable in those sorts of, difficult data, worlds, shall we say.
Okay. Interesting. So your point of view, alrighty, I say, don’t be afraid if you’re having less accurate or less, consistent data. You can still use the engine. It will learn and and etcetera.
Yeah. I mean, there are limits. Right? There are limits with anything like this. But in the main, when somebody says we’re not sure about our data, ninety nine percent of the time, it’ll work.
So there’s gotta be a one percent out there. Okay.
And and I did come across a business not that long ago that that that had no product category, no cost, no data about their products, very little transactions, and they genuinely weren’t a company ready for using AI and pricing yet. But that’s so rare. I don’t think most people need to worry about that.
I have a question from Shahrukh Gupta who said, what model would you recommend for a commodity based product in b two b environment, revenue mentioned model, or the negotiated model?
Probably.
In most, most commodity markets, negotiated is the is the right one because the price that you’re setting is a negotiation, whether that’s a long term contract or a short term contract.
In commodities where the where the price moves in a quite volatile fashion, some industries will be very spot purchase based negotiation. Mhmm. Some will be long term contracts, but with some sort of a tracking against an index, but you’re still negotiating where you are compared to that index or compared to a list price or some other, reference frame. Right? So the fundamental answer is always, if it’s actually a negotiation, then the negotiating model is is almost certainly the right one. And what we didn’t point out is that the we talked about the sort of ecommerce and list price one being focused on the volume reaction.
Right? And that sounds like something you’d want for everything. Right? But but, actually, there’s something better for the negotiated, environment, and that’s to focus on how do my prices cause or or lead a particular, win rate or win probability.
Mhmm. Because in a negotiated environment, you if I lower my price for my customer, the overall market demand doesn’t go up or down. And so instead, I’m more or less likely to win that business. And so instead of volume reaction, we focus on probability of winning and using that as how the algorithms are gonna learn and get better and better and better. And, again, separate to what’s the price? Are you looking to get more margin or revenue or volume or whatever else? Right?
Right. So from the first question, how do you know that your data is AI ready? The answer is ninety percent of the cases you’re ready even with, less consistent, or less reliable, data you can handle. Yep. Now the second question is then, okay, when you believe you’re ready, to implement, what are the practical steps now to build your game plan? What is the game plan? How how do you start?
So so just like What what’s your best practice?
Just like the first question, I’m gonna dive into a slightly different answer as the starting point. Okay. And and in the same and we’ve been talking about data, which is really good because, actually, the very first thing to think about is what sort of data will I need if I’m gonna be using one of these algorithms.
And and and do I have that data available, and is it easy for me to bring it, bring it together? So the I mentioned at the beginning, I think I’ve only got two slides for you. Right? So the other one is kinda data data focused. So if we’re gonna cover that off and then and then see what see what’s left. So whatever algorithm you’re looking at, the very first thing to think about is, what is it that I’m trying to price? And that may that may sound like a monumentally stupidly simple, thing.
But as an example, if we were thinking about that negotiated space, am I negotiating a long term contract? Am I negotiating a project, or am I negotiating just a volume that you might buy today or tomorrow? Right? What is it that I am negotiating?
And this is probably the most important thing to to think about because once you know what you’re on negotiating, that’s the core thing that you need to feed your AI to learn from. So AI will always say, I want something called transactions.
So if it’s a spot purchase, the transaction is the record of who bought something. Right? So the order or the invoice or something like that. This person bought this at this price after a negotiation. Right? That’s the sort of core.
But if it’s a project or a an agreement, then the transaction is the point that they agreed the agreement, right, rather than the purchases that happened.
Purchases were interesting, but the core item is we made an agreement on this date. What did we agree and at what price? So the starting point is what is the transaction? What is the main thing that I want my AI to learn to do?
Because Mhmm. I need to feed it the history of exactly that same thing. That’s my sort of core data that I that I want. Right?
Clear. And and within that within that transaction, you might think about what have I got. Right? I’ve got a I’ve got a product.
I’ve got a customer. I’ve got a price. I’ve got a cost. I’ve got all sorts of things.
I know they wanted it really quickly. I I know that it was raining. I know whatever else. Right?
So what data do you have about that Mhmm. That negotiation moment, if you like?
There is go on. Yeah. Well, there is a a question from Louis Vega who says, yeah, you talk about what price am I setting, but if you have discounts connected with the list price, of course, so you have, on invoice discounts. But what if you also, use off, invoice discounts, rebates, end of year things? What what what about that?
Should you Those those, those rebates really, really confuse things, don’t they?
So so the the it’s it’s easier to think about it in a, yeah. Yeah. Forget them. They don’t exist. We’ll just focus on the price and the discount or the net price or whatever else. Right?
But the way that you kind of deal with rebates in the world of pricing AI is in two or three different ways. Right?
K.
So so you can think about what is the total giveaway. What do I need to give my customer in the negotiation across rebates and discounts?
So I might feed the engine the transactional information that includes my costs, my agreed price or discount, and what I ended up paying in terms of, rebates. Right?
I might that might be what I feed it, and I might ask the system to end up telling me what my total giveaway should be. So once I’ve given them some rebate, I know what’s left that I put onto, discount.
Or some businesses, the rebates are very standardized.
Everybody has you know, I’ve got one rebate scheme. It applies to all of my customers.
And then what you can do is you can well, you can sort of automate, and you can almost ignore it. Not quite, but almost ignore it in your price negotiation context because it’s so consistently applied across your customers. So it really depends on how.
It the big question is who negotiates the rebate? Is it a centrally negotiated function, or is it part of the sales negotiation alongside the discounts? That influences what’s the best way to to approach it.
But, fundamentally, it is just a more complex landscape of what am I trying to price. Absolutely right. Because the same logic can be applied to what’s the rebate giveaway on its own just as it can be applied to what’s the discount giveaway on its own. So I don’t want to make things any more complex than they have to be.
There are often ways to simplify things, so that you can focus first on one thing or another and not have to boil the ocean all in one go. Right?
Yeah. That was a question.
Should we exclude then the rebates, or should we find a different way to If they’re if they’re very systematic, then I would start by ignoring them.
And then I would think about, do I want to make my model a little bit more complex or sophisticated over time?
Right. Yeah. There’s another question coming in, asking, if you have different manufacturers, some for instance, a channel that is more considering volume and another channel that is, more considering the product life cycle, for instance, pricing. Mhmm. Can you use the same model or not?
Normally.
And normally is the best is the best answer. Not absolutely exclusively always the case. The reason I say normally is what if you think about it, you’re you might want the answer from the model to be more driving volume or revenue in one channel and more driving margin, in another channel, for example. Mhmm. But that’s more like, what do I want to achieve from the model? The model is still learning and can tell me, hey. This is a good price if you wanna drive maximum revenue, or this is a good price if you wanna drive maximum margin.
So that’s more, what do I wanna achieve from it?
And considering the life cycle, or or do you So that could be of revenue or or margin, that that’s kind of easy.
Right? Yeah. When you think about, okay, there’s a there’s a product life cycle component, then the life cycle and the life cycle stage of the product is is just another attribute. Right? So it’s just another piece of data that I need to give to my model that I might have for one section of my product portfolio or channel, and I don’t have, for another.
And and good AI, when it’s fed that data firstly, if it’s missing for one channel, it’s easy. The AI is gonna go, well, I’ll do nothing with it. Job done.
But but if it’s available across everything but only really matters in one channel Okay.
The AI should be able to do is identify where the life cycle is actually resonating and driving a differential performance opportunity.
And you might find that there are little corners of the world outside of where you expect it to where actually there is an opportunity made available by having that information.
Right?
Right. And if you if you if you come back to the question, of your game plan, what what what what is the biggest opportunities for you with AI in in in implementing price strategy?
I think we’ve we’re we’re okay in terms of slides now, aren’t we?
Yeah. Yeah. That’s fine.
Yeah.
So what’s the what’s the biggest, opportunities?
I I kind of think of it from from the perspective of where is the biggest reward, the fastest reward, and the cheapest reward to get to. So and that’s not always the same answer for every business, obviously.
But But but that that that’s a bit the the low hanging fruit that I heard in the past and Yeah.
Consultancy as well, but it’s not always the best way to start.
There there’s detail. Right? That that’s an easy, glib thing to say, but, actually, it is quite easy to understand where is the fastest place to find a return in terms of pricing AI. So as a as an example, we’ve been talking about negotiated prices.
Right? Well, if I have spot purchase negotiations and I have long term agreement negotiations, one of those, if I start pricing better today, I’ll see the impact immediately the next day. Right? And that’s the spot purchase environment.
So whereas I might have a bigger overall prize in my negotiated space, but it will take me maybe more than a year to collect all of that because everything doesn’t come up for renegotiation until when it’s due and so on. Right? So Mhmm. So some things are slower to capture the value.
So how quickly can you capture the value is is a function of that price environment. A a list price can be a fast way to capture value if people are paying it or if people are paying a fixed discount from list. Both would give you a volume of business that you immediately impact.
Mhmm.
But but sometimes companies operate in an environment where there’s an expectation that list price is not changed regularly. Right? Yeah. So so that might slow down your opportunity to collect the value as well. Right?
Yeah.
So this kind of the so the type of pricing will have a big impact on how quickly you can get to value. And for me, if there is spot purchase business and there’s enough of it, that’s always where I would choose to start. And I would choose to start there for two reasons. One, as we’ve just been discussing, it’s really, really fast in terms of return if there’s enough volume there. Right?
Right. Right.
Other reason I would start there is that the data tends to be easier to collect.
So if you think about it, for a spot purchase, all I need is the the order or the invoice, and everybody records that. Right? Nobody is not recording that information. Right?
So your ERP has that as a data item already, and all we need to do is maybe add some customer and product information to it.
So Correct. Yep.
It tends to be the easier technical space and one of the fastest to deliver the value. So if there’s big enough volume there I also think that pricing AI is still something that’s reasonably new to a lot of people.
They’re still debating where should we invest in AI compared to other AI type projects. Obviously, that’s silly. The answer is pricing, but I understand that they have to go through that conversation internally.
But you therefore kinda want to pick somewhere where you can prove that value easily fast and at a low relatively low investment level as well.
And so then still and still, David, I I often hear the the the reflection made in companies, and that’s why the third question from me to you is what you what is your experience in when you start with AI, you implement, you see the opportunities.
Should you start small and try to scale up fast, or should you think more holistic, more even global Mhmm. And and then customize in countries or Yeah. Categories or whatever.
What’s your what’s your experience in that, the w? So small and scale up or think big and customize?
Design big and start small. So so let’s imagine for a moment that you said, right. I’m gonna go with, my spot purchase or my negotiation. Doesn’t matter. Right? That I’ve picked my space. This is where I want to focus.
I would then say that the AI is better when it has more data.
So even if I said, right, I’ve got lots of spend in spot purchase. That’s where I’m gonna focus. I want to start and prove the value in this country or something like that. Right?
Mhmm.
I would still say design the model and give it the data, but more even if my proof of concept, my initial rollout is only one country because it will be cleverer in its ability in one environment when it has a broader dataset presented to it. And it doesn’t fundamentally it doesn’t normally make the project significantly, heavier to do that. Right?
So design as broad as you can even if They start small.
Starting small to get a proof, get a win, get everybody on board, get a good story, but then support kind of a rollout that can happen quite quickly because your model’s already there and ready. Right?
Yeah. Well, as you’re precocious, Dominique, I have an interesting question from you from Mhmm. A participant. He said, I understand that, process, the AI solution embedded in the pricing platform and your platform.
What about integrating existing external AI pricing oriented solutions into commercial models?
What’s what what make we choose for Prose?
So external AI pricing.
So there’s so there’s a there’s a couple of different ways of of thinking about that. One is to say, that would be ridiculous. We’re brilliant at pricing. Why would you want to use somebody else’s?
It’s a big lift, but, but there’s a few different sort of angles. So it really depends on where the question is coming from. On on the one hand, there are situations where what a company requires is something that isn’t perfectly aligned with an existing AI model. They’ve put some of their own investment or development into some intellectual property, and, actually, that might influence and adjust the type of model that they need to operate, but they want it embedded in a pricing platform.
They want it industrialized. They want it stable, productized, and all the rest of it. Right? So there are situations where we will host customers’ own algorithms as part of our pricing platform for them so that they can have that productized approach.
Mhmm. And there are cases where we’re almost creating a little bit of a bespoke blend, if you like, or using our AI and adding elements to it.
There’s there’s another angle that the question might relate to.
Probably not, but it’s worth mentioning, which is that when you think about the data you’re feeding an AI to learn from, that doesn’t have to be just the data that you own in your own environment, your transactions, your customers, your products. Right? It could be external market data that you’re collecting that you want the AI to learn from.
So that could be competitive prices or price indices or material indices or Seasonal influences temperature.
Exactly. Exactly. So so sometimes this sort of question actually relates to that might be good, but I need it to also learn from what’s happening in the competitive pricing space or something else. Right? And that then just becomes additional data that you would give to your model for it to learn from.
So it depends a little bit on exactly where the question’s coming from.
Yep. Yeah. Good answer.
My last question is, you start implementing your game plan. You’ve selected your opportunities.
You you trusted data that you have. You start with your game plan. What are all the most important enablers? Some people say it’s data first.
Some people say it’s system. Some people say it’s people, and some people, I hear rooming. No. No.
It’s money.
It’s not yet.
What is the most important enabler for you, Dominique?
Let’s take two of them off the table straight away. Yeah. So it’s not it’s not money. And Okay.
Okay.
That’s that’s easy for me to say.
I haven’t gotta find the money and buy it. Right?
But these things are a lot lot lower price point than they were five, ten years ago. So people typically have an expectation that this is something very, very expensive to do. Mhmm. And and it’s certainly not, of this of the order of magnitude of implementing a whole pricing platform, for example.
Just focusing in on on AI can be very efficient and light, and especially when you look at a sort of an initial reasonable scope, right, to get that that proof point. So so money, we can’t completely dismiss it, but it’s no longer the big hurdle that it was in the past for pricing AI. Right.
I also think that data isn’t the hurdle. Everybody thinks it is. You should think it is. You should force a vendor to work out that it is okay and why and all the rest of it. But like we said earlier, overall, it’s unlikely to be a real challenge in the way of going forward.
So you’re kind of left with, the other two, right, which is the system, the vendor, somebody like PROS or others. Apparently, they exist others. I don’t know. Anyway, so the vendor, the system, and, and, of course, the the people.
So I would actually argue that those two are interconnected.
That’s weird because everyone is panicking about the data, data first, data first, and Yeah.
Your point of view is garbage out. Yeah.
Just feed garbage in. Good AI should be able to identify what is garbage and what isn’t. Otherwise, exactly what intelligence is there in it. Right?
That’s fine. Yeah. Yeah. Yeah.
I mean, if we’re learning, then we ought to be able to learn from the from the mix of data what’s good and what isn’t good. Right?
Anyway But I see I hear so many organization talk talk about global data leaks, tools, or whatever.
No?
Yeah.
Improving data is always valid, always good, always healthy. But it’s really rare that the data isn’t good enough already to make a substantial step forward.
That’s an important message.
Or you kinda go, what are you waiting for? Right? You know?
An important message. Yep. Yeah.
So so I think it comes down to system and people.
Now of the two, people is the most important. Totally.
But but, weirdly, system is deeply interconnected with that. And so what what I mean is, no matter how good my AI is let’s imagine my AI is absolutely perfect, genius level, never gets anything wrong, perfect answer to every single question a hundred percent of the time.
If my employees don’t believe it, am I gonna get the value compared to the number that they got yesterday that they didn’t believe either?
No. Right? So If you don’t trust trust the system, you’re out of course.
Exactly. Exactly. So you’ve got to find a way to get them over that hurdle. And some businesses, that’s easy because they’re trusting, and they love you, and they believe you.
And in others, they’re like, yeah. Arms crossed. Let’s see. You didn’t know about pricing last time you gave me a number.
Let’s see if you do now. So so the the system then becomes really, really important in can it help you build that trust.
And and the way you help it build that trust is by making sure that the AI has a really deep understanding of what the salesperson was going to think themselves.
Mhmm.
So for list prices, it’s easy.
You don’t need salesperson to agree with you if you need it.
But for negotiated, that’s where this sort of trusting is really, really critical. Right? So if your AI can be really, really accurate, not just in working out what the price should be, but in working out what the salesperson was going to actually go for themselves, that can be a little bit kinda like, really?
But good AI will first work out exactly what price it thinks that salesperson was gonna go for.
And can AI help? That’s a question that, that I just see in the the chat is Imagine that you’re in in a declining market all of a sudden. Market declines for a couple of reasons.
Can AI also recommend you on on, you know, your future price points if you are estimating a decline in your demand?
Yeah. Definitely.
So good good AI will do will do two things. So the good AI versus bad AI, how quickly the AI responds to a new market reality is one good measure of how good that AI is. Right? So you should be feeding it information regularly.
Regardless of how often you change things, it should be learning and feeding regularly. Always up to date if if at all possible. And then how quickly it will respond to a new market reality is really interesting. Right? Can it detect that there is it’s much harder to sell in this environment? Maybe a competitor has done something crazy on price or or the other way around. Maybe there’s an opportunity because somebody’s someone else’s warehouse has had a fire or I don’t know what it is.
But And does does AI give me the explanation?
Well, AI pricing AI doesn’t yet do market research and tell you, hey. Somebody’s got a fire in their warehouse.
Let me it’s it’s go for a time. But what it can do is say, suddenly, there’s an opportunity I think the if you’re prepared to go.
Inconsistencies are yeah. Yeah.
Yeah. Yeah. And and capture that. And then and then the other thing is it should be tracking how things are changing over time. So in a general market trend situation, the AI should be able to work out where I am for today, where I am for tomorrow. And if I’m and spot purchase, it’s not really that important because it’s very in the moment. Right?
But if I’m agreement if I’m setting an agreement for the next year Yeah.
Then I really do wanna take into consideration where is it heading, not just where is it right now. Right?
Exactly.
And and, actually, sometimes there is a difference in the impact of not where a market is in terms of absolute value.
So maybe your costs are volatile and or they’re on an upward or downward trend. Right? And you can predict that out.
Sometimes the rate of change is something that customers respond to positively or negatively disproportionately compared to the absolute value. So models will often learn from not just the overall trend in its direction, but what do you do when it’s particularly sharply moving up and down?
Okay.
Let me just for that over time as well.
Yep. Yep.
Yeah. Great.
I’m I’m checking the questions. I don’t think there are additional questions. Well, there is one that I’d like to share with you, Dominic, but you can maybe, respond a bit later. It’s about, as pros, examples or samples or deep quick demos on contract pricing, learning models. I think you can take it off, afterwards.
I will share you. No. It’s anonymous.
If if the person wants a quick demo, please, share your, name, and then we can As a as a software company, we’re quite keen on talking to people and showing them stuff.
Can you imagine? Yeah.
Fantastic, Dominic. Thank you so much.
Thanks for the great conversation.
I learned a lot myself as well again Pleasure. As always from you. It’s always a pleasure. So, thanks. Rita, they’re all yours.
Of course. Thank you. Thank you very much, Dominic and Pol, and thank you to everyone who submit their questions.
The recording will be available in just a few days in the webinars and presentation section on our website. Before we close, I’d like to take a moment to highlight an upcoming event you might find relevant, the EPP Ignite AI, the future of pricing, taking place on September seventeen and eighteen in London. It is a unique opportunity to connect with peers and to dive into how AI is being applied in real business settings. If today’s session resonated with you, this is definitely an event worth exploring.
You can find out more and register at ignite dot pricing platform dot com. Thanks again, Dominic and Pol, for sharing your insights with us, and thank you to everyone who joined. It’s been a pleasure having you with us. I wish you all a great afternoon.
Thank you so much.