Boosting Retail Margins: Price Optimization Strategies with Machine Learning

Don’t miss the upcoming AI, Machine Learning and Data Science Conferences www.datascience.salon/ Book your pass today to learn more about generative AI and machine learning in the enterprise.
Presented by Maia Brenner, Data Scientist at Tryolabs
Setting the right price for a good or service is an old problem in economic theory, marketing, and business practices. However, as companies have more data than ever about their business and their customers, the world is moving towards new data-driven pricing strategies.
Machine learning tools allow us to incorporate structured and unstructured data in order to get accurate demand forecasts. But most importantly, with Machine Learning tools we can learn, for example, the price elasticity of demand for each SKU on each location and day, and therefore understand the willingness to pay of each customer. With these demand curves estimation and optimization algorithms we can learn which is the right price to set at each moment in time. In addition, with explainable AI tools and with randomized control trial experiments we can get new business insights, and deliver high impact price recommendations.
In this presentation, we will focus on how Machine learning is reshaping price optimization and will show outstanding results obtained by leading retailers around the globe.
In particular, we will walk you through our Machine Learning approach, which helped a Luxury Retail Company boost its gross margin by 28%. We will share the algorithms, techniques, and data sources applied, which led our client to gain over half a million dollars during a 10-weeks price optimization experiment.

[@jimbocho660]

A very nice presentation. Interesting and informative. Thank you.

[@DataScienceSalon]

Thanks for watching!

[@harshaneekhra3553]

nicely explained!

[@sahilkakkar5628]

Very insightful video, thank you

[@babakheydari9689]

great and informative

[@hsoley]

Very nice and informative!

[@brandoncharles5413]

So would profit be an input feature of the bayesian optimiser? As in, you multiply the predicted sales by the gross margin and use that as feature to optimise the price.

[@juiguram7177]

what is the type of optimizer used , the audio is not very clear

[@gamilfarea4276]

Thank you for sharing. What was the name of the optimizer?

[@naveenvenk]

did you get the optimizer name ?
@18:50

[@matiastorres392]

hi ¡ , nice presentation, but i coudnt catch whats is the variable that you predict with the model. Its just the demand for the nexts week?

[@khanhtruong3254]

I think the model predicts the sales (the demand), i.e. how many units will be sold. And the price is in the features (independent variables) when building the model to predict the sales. By doing so, we can arbitrarily change the price from a to b to see how the sales vary in the price range.