Ratha pitches Dfirst, an AI tool that optimizes discounts for retailers. The platform predicts how price changes affect sales, then recommends discounts that maximize profit. Clients saw gross-profit gains from 7% to 48%, with an average of 26%.
Hi, I'm Ratha, CEO of Dfirst. At Dfirst, we help you stop losing gross profit during discount campaigns. So among all the brands here, can you please raise your hand if you're doing discounts right now, either at your physical stores or on your website? Okay, I would say 30%, but I hope that seeing that Black Friday is coming soon, so you will be doing more discounts. And so today what we observe at retailers we talk with to choose the discount today, normally we look at 4 things. We look at the sell-through rate, we look at the stock level, we look at the age of the product, and we look at the competitive pricing at best. And we put rule-based around this, but it give quite good discounts. But actually with accurate AI, it can help you much, gaining more, much more gross profit. So with AI, actually you can take into account much more factors. So we do take into account the factors you are using right now, but with more factors such as, for example, uh, trends that are hidden in your transaction data, characteristic of the products data, seasonal data, and all the external data we can help you fetch. You can find the right discounts for each product. So, the sweet spot to put for each. So, Deferse, we are super specialized on this. So, actually, we built Deferse also with our frustration. We on the consulting side, though. So, we were helping multiple retailers across the globe like Sephora, L'Oréal, Pepsi, for example, on discount and promotion optimization use case. But it was, let's say, quite a high-cost consulting project and one-shot. So we said that after building this from scratch for more than 5 years, let's do it as software that is super, super fast for retailers to use. So you see later that we use only 2 or 3 weeks of implementation and which cost much less with implement— sorry, with maintenance included. So we build this with our experience making it a software and able to achieve super high accuracy to allow you to sell at a higher price when you can and at a lower price only when you need to. Good. Some sneak peek of how it works. So in the algorithm, there are two big steps. The first one is the elasticity of discount prediction. So it's essentially how much volume are you gaining on selling on each of each product when you add more and more discount. We have a super high accuracy, and then we give it to the second step, which is the optimizations under your constraints. So we can take it like, for example, your business constraints. So from your experiences of doing merchandise planning for decades or multiple years, such as, you know, that for this group of product, this type of product, during this season, you shouldn't put more than, let's say, 20% discount, but you sell anyway. You can put constraints on margin, for example, depending on your suppliers when you negotiate with them. You can have some of the products that you want to keep more margin than the other. And competitor rules, like you want to say that for this type of product, comparing to this competitor, you can be a bit more expensive, but not more than 10% more expensive, for example. All these constraints, thanks to your use, to your insightful experiences, combining with the goals. So the goals, it depends on what is the time of the year and also what is the type of the products. Let's say we are on Black Friday. You can have permanent collections that you, uh, that sell quite well, but you want to make it attractive for consumers. You put it on sale, but you maximize gross profit. And you can have a seasonal product, end of season, during the same moment, so Black Friday, that you want to maximize more the clearance or reducing the stock cost. With all this, the algorithm will give you the right discount for each product that you can apply either at your physical stores or online. And we are able to, of course, provide you the right discount tailored for each channel or each region or even each stores depending on the, on the volume of your, of your data. So in the, in the interface, we built it so that you can use this powerful AI without any technical knowledge. We have the Decide part, the Understand part, and the Measure part. So in the Decide part, you can set the scope of the products, the campaigns, the campaign dates, the constraint, the goal. The understand part, we help you recommend what is the right discount for each product, but it's not a black box. We also explain you, try to, we also explain you, sorry, why. Why? Because we have this level of stock, because we have this incoming stock, because we have this trend of the sales, because this is the financial uplift that you will gain by doing this discount that we recommend. Lastly, we have the measure part that allows you to do A/B testing as much as you want to measure that differs really help you gain much more gross profit than your existing strategy. So, these are some examples of our customer success story. We actually have two typologies of clients. We have clients who are resellers of other brands like BSTN, who is kind of JD Sport. With headquarters in Germany, serving across EU. We help them gain +25% gross profit uplift. And another type of other brands that sell only their brands. So we have Sikjun, for example, French streetwear brands, where we help them see that there are so many products, they put too much discounts, and thanks to us, they have gained +36% gross profit uplift. Hi. So our customers' gross profit uplift range between 7% and 48%, so with the average of 26%. So let's say if you work together, you have 98% of chance to get more than 10% gross profit uplift. So as I said earlier, we have helped a lot of retailers around the globe on this use case, super specialized on that. And today it differs. For example, we have Verbode, which is one of the biggest kids' wear brands in France. We have also some luxury brands that I cannot name it here, but very well known in Italy and in France, for example. So thank you very much. Thank you, Ryan.
