Sign up

O’Key – Case Study

O’Key already used the top-rated Oracle RPAS software to manage replenishment, yet they continued to struggle with inventory inefficiencies, especially stockouts and overstocks of promotional inventory. Could another system increase efficiency?

O’Key tested the Evo Transfer software directly against Oracle RPAS to see if its Prescriptive AI could deliver better results than traditional predictive software.

The result: the Evo system delivered +23% inventory efficiency over Oracle RPAS with higher product availability and less leftover inventory post-promotion.

increased inventory efficiency +23%
by replacing Oracle RPAS with Evo AI

How O'Key







Context: about O’Key

O’Key is a leading Russian grocery retailer, with over 100 super- and hyper-markets in 2 brands. O’Key has expanded rapidly in recent years, thanks in part to its online ordering service run through its hyper-markets.

While O’Key is a popular brand, grocery is an incredibly competitive and highly fragmented industry. O’Key remains competitive in its many locations primarily through local, targeted promotions.

While regular promotion cycles are effective, they make replenishment incredibly challenging. Demand is volatile and locally diverse, leading to unexpected stockouts. Extra shipments to prepare for increased demand, on the other hand, regularly lead to leftover overstock post-promo.

The challenge: unpredictable inventory needs related to weekly promotion uplifts, causing regular stockouts and overstocks

Despite using the top-rated Oracle RPAS software, inventory issues remained a challenge at O’Key. Armin Burger, CEO of O’Key, needed a more efficient approach to replenishment.

He was frustrated by the regular stock issues, yet he was doubtful that any solution could do better than a well-established system like Oracle RPAS.

It’s difficult to predict the impact that our weekly promotions will have on demand. Some discounts exponentially increase sales while others have only a marginal impact. To ensure sufficient product availability, we tend to overstock promotional items.

These overstocks create significant waste and cyclical markdown pressure. We needed a more sustainable solution, yet I wasn’t sure one existed. We already used Oracle RPAS, which is widely considered a top predictive supply chain solution.

-Armin Burger, O'Key CEO

Additional, critical problems stood in his way:

  1. Highly complex logistics: Both brands stock their own private label and 3rd party products. Replenishment of hundreds of thousands of SKUs happens on a daily basis across a large, geographically diverse area.
  2. IT integration difficulties: Oracle RPAS was built into all IT systems in stores and at the corporate level. A new system would have to replace Oracle functionalities and integrate seamlessly into the existing infrastructure.
  3. Store manager and supply chain manager hesitancy: Staff was already comfortable with the Oracle system and sceptical about learning a new technology, especially if it would deliver only minor improvements.

The solution: prescriptive, not predictive, supply chain software that differentiated at a per-store and per-product level every day

O’Key partnered with Evo to test Evo Transfer against Oracle RPAS. Burger wanted to see how the Evo recommendations would stack up against those given by Oracle. He was curious whether Prescriptive Artificial Intelligence could actually improve inventory efficiency over a more widely-used predictive approach.

We felt frustrated by the error rate and inventory inefficiencies that we were experiencing with Oracle RPAS, but I doubted that another forecasting software could make a significant enough difference to justify the costs of switching. The ROI of a new system would have to be overwhelming.

Evo’s AI, however, used a more responsive, prescriptive approach, not a predictive one. I was intrigued to see if this agility would translate into higher efficiency.

Evo implemented a responsive replenishment strategy: daily replenishment recommendations per SKU for each store.

This approach relied on:

1. Tracking historical sales and market data

Evo combined historical sales data with extensive market and competitor data to better understand local demand. Impact is magnified by monitoring sales of over 300 critical products and the consumer behaviour of 14% of the Russian population.

2. Estimating product sales in each store

Evo Transfer estimates product sales potential based on SKU attributes, product category, and overall market demand. The AI calculates the impact of promotions using a real-time, responsive forecast.

3. Providing daily recommendations on full-price and promo replenishment needs

The Evo system provides daily input on both full-price and promotional inventory replenishment needs. The AI then translates this using O’Key’s existing logistics rules and delivery schedule into simple, actionable replenishment recommendations.

To measure impact, Evo Transfer was initially deployed in 5 stores in different regions alongside the Oracle system during a rigorous A/B pilot test.

Evo Transfer’s recommendations were not implemented but instead compared to those made by Oracle to see the direct impact when sales are precisely equal. Burger wanted to see which system would best address its inventory issues.

Regular overstocks and the resulting waste was our biggest problem, so we wanted to focus on the impact in just that area. I needed to know if Evo’s prescriptive approach could minimize the need for markdowns without reducing product availability.

Pilot impact: +23% increase in inventory efficiency

During a one-to-one test against Oracle recommendations, Evo Transfer increased inventory efficiency by an extra +23%.

After just one replenishment cycle, the Evo software had decisively outperformed Oracle’s. The returns were astonishing.

The Evo software more accurately estimated both regular and promo demand. This resulted in -40% less replenishment needed to maintain product availability and a -23% reduction in leftover stock after the promotion. Fewer markdowns and overstocks resulted directly from the Evo approach.

Evo delivered incremental gains beyond what I believed possible from existing supply chain software.

Additional results: -11.4% reduction in stockouts and -16.2% reduction in overstocks

After Evo Transfer proved itself to significantly reduce inventory efficiencies over Oracle, Burger actively deployed Evo Transfer in all O’Key Group stores across all Russian regions. Soon, the Evo system had entirely replaced Oracle RPAS.

Thanks to a collaboration between O’Key supply chain managers and the Evo team, the transition to the new technology went smoothly.

The Evo system fully integrated into the existing O’Key infrastructure, allowing store operations to continue without disruption. In fact, the simple interface and reduction in required replenishments reduced store effort. The average daily shipment volume decreased -21.6%.

The Evo software lifted the efficient frontier for the inventory/stockout trade-off for both regular and promo product sales. The new system further reduced stockouts by -11.4% and store overstocks by -16.2%.

The traditional limits of supply chain forecasting no longer apply.

Armin Burger

Evo provided a superior solution. By addressing our core supply chain pain point, they improved over Oracle RPAS. Evo proved that with a prescriptive approach

Want to learn more?

Hey! Was this page helpful?

We’re always looking to make our docs better, please let us know if you have any
suggestions or advice about what’s working and what’s not!

Send Feedback