The retail sector is today characterized by profound uncertainty and unprecedented change, to say the least. Retail Leader and Progressive Grocer editorial director Mike Troy recently hosted a lively online panel discussion with DemandTec President Cheryl Sullivan and Head of Science Geoff Pofahl to explore how retailers can successfully navigate crises with carefully crafted full-lifcycle pricing. I always enjoy all three of these folks, and as you wold expect, their insights on this topic were valuable.
If you missed catching it live, please enjoy the replay at your convenience. Meanwhile here are some teasers to whet your appetite!
Mike opened by referencing Retail Leader’s recently published report, sponsored by DemandTec, called “Crisis Pricing: The Role of AI in Times of Uncertainty.” He recapped some highlights from this recently published report, then kicked off the discussion. One of the challenges Mike wanted to explore is how retailers can effectively grapple with the fact that little or no relevant historical data exists when shopper changes and preferences have changed so dramatically from pre-COVID days.
Geoff explained how good science can help retailers respond faster to changes in the demand signals than they could otherwise. Science can identify what new signals are emerging and separate them from the noise. It can also determine what historical data still remains relevant. Having science on your side really helps you quickly disentangle meaningful signals while enabling you to respond to with more agility and accuracy in a dynamic environment.
Noting the dramatic uptick in shoppers going online, Mike asked Cheryl her predictions about this shift after the COVID pandemic abates. Cheryl noted that many shoppers formerly reluctant to embrace online have now grown much more comfortable, most notably in grocery. While some may return to in-store shopping eventually, online growth will continue.
She cautioned that its important to note that online pricess sensitivities are different from in-store price sensitivities, so use science to understand sensitivities by channel.
Geoff also addressed the various channels and emerging delivery and pickup models. “In science-based pricing, we factor trends, seasonality, stockouts, etc., all by channel – the model classifies data, and signals are associated with which channel and which fulfillment method. When we feed the science models, they are instructed based on the retailers objective – driving profits, revenue, or margins, for example.”
Mike then steered the discussion to post-COVID retail challenges as retailers move to shed excess or outdated inventory. Cheryl notes that inventory is piling up in many sectors like apparel, and they’ll be using markdown optimization more aggressively. Even grocers, who are aggressively moving many items, are seeing market basket contents shift and will need markdown to clear less popular items to make way for items newly in demand.
Geoff talked about the importance of providing markdown science that can address either short-life merchandise like seasonal apparel or long-life items like grocery center-store staples. “Robust markdown solutions can be a retailer’s best friend as it can help retailers clean the merchandise out as profitably as possible.”
If retailers do adopt science to drive more frequent, algorithm-driven price updates, will they meet with shopper resistance? Cheryl cited recent research showing that in fact shoppers are very accepting of science-based price updates, even frequent ones, as long as shoppers perceive the prices as fair and non-arbitrary. She sees wide acceptance of dynamic pricing as shoppers’ have gotten more experience with Amazon and other vendors using dynamic pricing.
Geoff expounded a bit on the importance of giving retailers a good understanding of science in order to gain trust and speed adoption. Cheryl reminded us that “early pricing solutions had a black box around the science, and retailers were often suspicious of the recommendations that didn’t align with their gut feel. Today, good solutions have full transparency, enabling retailers to understand and more quickly embrace science-driven pricing. As a former professor, Geoff is particularly attuned to the importance of good communications and education in helping retailers truly understand what science can do for them.
The panelists discussed how KVIs, which are traditionally quite stable, have been shifting very quickly in the COVID-19 era. Cheryl is a proponent of using science to quickly identify which items have highest price sensitivity today. Geoff discussed pioneering work within the DemandTec science team to productize KVIs, which stands out in the industry. It enables retailers to prioritize price changes based on known (and the most current!) KVIs. Cheryl noted that science also factors in cross-item price effects like cannibalization and affinity, and Geoff explained how these are directly addressed in the science.
To sum it all up, Mike asked our panelists what key features a retailer should look at when evaluating AI-based pricing. Cheryl’s tips:
- It’s important to look for productized science around critical items like KVIs, market basket analysis and determining price strategies for categories and items. Relying on one-off analytical services engagements to update each of these crucial areas doesn’t cut it at a time when retailers need to respond dynamically to fast-changing environments.
- Remember the importance of full lifecycle pricing breadth, from initial price to promotion through markdown, without having to go to separate vendors for each phase and creating silos.
Look for a vendor whose science goes across that whole portfolio, plus manages vendor trade funds.
- Remember that science that addresses both long- and short-line markdown helps you clear inventory profitably.
Geoff weighed in also:
- Be wary of one-size-fits-all science approaches.
- Look for science that evolves and adapts as markets and circumstances change -that’s the true power of ML and AI. It should be ever-evolving.
- DemandTec is constantly poking and prodding and looking at how to innovate our science, even applying it to our inputs to better identify new signals. Recent examples include moving to automated, productized science around KVIs and market basket analysis, while also enabling retailers to continually re-align items and categories for their optimal price strategy.
There were lots of questions posted in the Chat area from listeners, and we’ll be posting another blog soon to get to all of those, including ones that we were unable to address during the panel due to time constraints. Check back soon! Meanwhile, enjoy the replay.