On September 30, 2026, DS Stream held the webinar Pricing at Scale: Protect Margin & Clear Inventory Across 50,000+ SKUs. The main speaker was Magdalena Okrzeja, AI Transformation Advisor at DS Stream, who has almost 20 years of experience in IT. Bartosz Chojnacki, AI Practice Leader at DS Stream, hosted. The recording is now public on the DS Stream YouTube channel and runs about 43 minutes.
Watch the full recording: Pricing at Scale: Protect Margin & Clear Inventory Across 50,000+ SKUs
What the session covers
The session opens with the problems pricing teams run into once a catalog gets this large, then works through them in a product demonstration.
- Five challenges of pricing a large assortment. The session names them up front: seeing which products earn margin and which bring traffic, pricing new products that have no sales history, trusting what a system recommends, applying business rules on top of the statistics, and sorting through more market signals than a team can follow.
- Segments in place of category averages. The demo groups SKUs by how they sell: margin drivers, traffic drivers, and inventory risk. A category average hides those differences, so one price change across a whole category can cost traffic on some products and margin on others.
- The cold start problem for new products. New SKUs have no sales history, so they are matched to similar products already in the catalog by specification and by product image. A launch price can then draw on how comparable items have sold. The same matching flags products in different families that overlap closely enough to cannibalize each other.
- A pricing campaign built around one primary goal. A campaign starts with the scope of products and a single main objective. From there the team sets limits, adds global rules such as a minimum margin, picks the range of historical data to use, and can run an A/B test to check the results.
- Recommendations that come with reasons. Every recommended price shows the reasoning behind it, and results can be checked SKU by SKU or rolled up to a product family or a whole segment. Teams can accept recommendations in bulk. Where the system is unsure, a person decides.
- Data that is complete, not perfect. The session ends with what this approach needs on the data side. Imperfect records can be worked with, but the core inputs, starting with sales history, have to be there.
Who the recording is for
The recording is aimed at people who set or approve prices across a large catalog: pricing, merchandising, category management, and retail analytics teams. It also suits anyone who supports those teams with data or tooling and wants to see how such a workflow fits together before planning one.
Where to go next
If the session matches the problems you are working on, the AI pricing optimization solution page describes how DS Stream approaches pricing for retailers with large catalogs. You can also book a free consultation to talk through your own pricing setup. There is no obligation.
For background on the method itself, read our guide to retail price optimization at SKU level.
