The Art of the Possible: A Conversation with Dave Steck on the Evolution of Shelf Intelligence

Q: What were the earliest goals when you first rolled out Simbe at Schnucks?

A (Dave): It started as a science experiment. We had a robot in one store in 2017, and at that point, we were really just trying to understand customer reactions, what did people think of it? Would it navigate around baskets, shippers, and u-boats left in the aisles? Once we felt good about those basics, we moved into pilot mode in 2018 and 2019.

The goal at the time was more: Can this help us figure out how big our out-of-stock problem is? We knew we had a problem, but we didn’t know the size of it. We started with a 16-store pilot and didn’t involve a lot of teams—just IT at the start, which in hindsight, wasn’t ideal.

But our mission was clear from the beginning: focus on out-of-stocks and it was important to stay focused on that one thing. Everything else came later.

“The goal was never to do everything. It was to do one thing really well—fix out-of-stocks.”

Q: Why was that focus on out-of-stock detection so important and how did you measure success?

A (Dave): When we had Tally in those 16 stores, COVID hit. Shelves were decimated, backrooms were empty, and it became clear very quickly that we could use Tally to update our perpetual inventory (PI) data. If Tally saw something was out, we’d just zero it out in PI and let the system order it automatically. That took the burden off the store teams, who no longer had to manually scan outs.

We had a third-party conduct a financial analysis using conservative assumptions. At the time, the industry average for product switchability was around 55 percent, but Schnucks was significantly higher than that going into COVID. When the assortment shrank due to manufacturer constraints, our switchability index dropped, but we still remained well above the industry average and the value proposition actually got stronger. That made the business case for expanding chain-wide to over 114 stores.

Q: How did other teams start to tap into the data beyond IT?

A (Dave): Store ops was always the biggest consumer of the data. But early on, we also started collecting product location information, where every item was on the shelf. That unlocked a whole new set of use cases. For example, we used that data in our e-commerce app to guide Instacart shoppers through the store more efficiently, and even used it in our customer-facing app to help shoppers find products based on aisle and segment.

Then we realized: why are we printing paper tags based on planogram order when we know where the products actually are? Across the industry, planograms are about 60 to 70% accurate. So we started printing them in “realogram” order based on Tally’s data. That saved time, improved accuracy, and made life easier for store teams.

“Realograms were a game changer. We stopped printing paper tags based on what the plan said and started printing them based on what was actually on the shelf.”

Q: You mentioned the space planning team became heavy users. Can you expand on that?

A (Dave): That one surprised me. I didn’t even realize how much they were using the robot images until they came to tell me. They use it daily. Before doing a shelf reset, they’d check what’s currently on the shelf and identify products that weren’t going to be in the new set. That let them mark items down and sell through the inventory before the reset team came in. Otherwise, it’s a mess—products get removed, carted to the back, and go through a whole clearance cycle that’s just inefficient. This saved a lot of time and hassle.

They also started comparing inventory service counts with Tally data to identify items missed during physical audits, another big unlock.

Q: What surprised you most as more parts of the business started using the data?

A (Dave): Honestly? The data itself. I once said that we were most excited about the data and what we could do with it, but I had no idea how true that would become. Most grocers don’t actually know what’s in their stores. The catalog of what’s supposed to be there is always bigger than what actually is. With Tally, we had a real-time, visual understanding of the shelf. That changed everything.

“The data surprised me the most. Grocers think they know what’s in the store but they don’t.”

Q: You mentioned Tally working in tandem with Electronic Shelf Labels (ESLs). Can you share how that works?

A (Dave): Originally, our ESLs would only flip to “temporarily out of stock” if an item was out at the warehouse, but not if it was just out on the shelf. This was confusing, because if the ESL didn’t say “temporarily out of stock” customers could easily assume that it was available somewhere else in the store and keep looking. But customers don’t care about the warehouse, they care about whether the item is available to purchase or not. So we changed it. Now, if Tally sees something is out on the shelf and it’s not in the backroom either, the tag flips to “temporarily out of stock”. As soon as it’s received, the tag goes back. That consistency made the customer experience much better.

Q: What advice would you give to a retailer just getting started with shelf intelligence?

A (Dave): Keep it simple. Everyone wants to chase the “size of the prize” with 10 use cases at once, but that’s not the way. Start with a single use case like out-of-stocks. That’s where the ROI is. Build the muscle gradually. And don’t overload your store teams with 1,000 tasks the minute you get the data, just give them what they can act on. The goal was never to do everything. It was to do one thing really well: fix out-of-stocks.

“Don’t try to boil the ocean. And don’t let perfect get in the way of really, really good.”

Watch our fireside chat with Dave below.