Elastic boosting
Increased items sold by 40% and Ozon's total GMV by 1.6%

What our team does
I work on the promotions and product-advertising section for Ozon sellers, helping them promote products in search results, drive more sales, and take part in special promotional campaigns. The goal is to get sellers engaged with promo mechanics and build long-term loyalty.
Problem
We currently run a set of promotions with three fixed discount tiers and three matching boosting tiers. This design gives sellers no incentive to raise their discount beyond the minimum — they set the threshold discount just to qualify, then have no reason to offer more.

Research
Analytics showed that about 60% of sellers add products only to promotions that require the minimum discount needed to earn a search-ranking boost.

Interviews and takeaways
I've prepared a guide and a set of questions for user interviews to understand how they're currently using promotions
Sellers don't understand the differences between promotions within a single campaign, which hurts the share of products enrolled
Sellers set their discount based on product margin, but are willing to increase it if a meaningfully bigger search boost is on offer (a “meaningful” boost is 15–30%)
They track promo sales through the Ozon seller dashboard analytics or third-party plugins
Adding products to different promotions is inconvenient — it means opening three separate Excel files and constantly switching between them
Hypotheses
For each hypothesis, I sketched rough designs to make it easier to discuss ideas with the manager and present them to the commercial team.




Scoping
Together with the product manager, we reviewed the hypotheses and selected the first and third ones.
We decided not to go with the second one because that approach to pricing and discounts might be inconvenient for users and difficult to implement.
Test
After thoroughly refining the mockups, I prepared prototypes and questions for usability testing with users

Final result
In the final version, I added reference prices for each boost to make it easier for users to understand the pricing
A/B test & Impact
After completing development of the feature, we ran an A/B test on 50% of sellers and got some pretty good results