In today’s competitive landscape, AI presents an opportunity as much as a pitfall for any physical retailer. In this part of Adaptable Category Management series, host Simon Hernaus talks with Helen Kom, Inventory Optimization Product Director at Leafio AI, to uncover the thin line between AI success and failure.
In the interview with Helen, you’ll learn how to:
- Refine underlying processes before implementing AI solutions to prevent inefficiencies and avoid ‘automating chaos.’
- Conduct strategic experimentation such as optimizing store layouts and pricing strategies, to achieve significant revenue increases, including a 35% rise in one store.
- Balance human intuition and AI-driven processes in category management, with an eye towards a more balanced 50-50 division of tasks in the future.
—
— Intro
»A good category manager’s very analytical mind must be backed by love for retail,« says Helen Kom in an interview discussing pitfalls and opportunities on the road to more streamlined retail category management. Then laughingly adds: “I think I could be a good category manager.”
I can only agree, as someone who’s been in the field for more than 25 years. Fitted with in-depth understanding of computerised optimisation methods, she could even set the standard for a successful category manager in the AI-driven era. Yet, for now, this remains an untested hypothesis because Helen serves another role.
As a Product Director of Leafio AI’s software platform for retailing, she navigates through topics like demand forecasting, inventory management, and all things retail automation. She doesn’t sugarcoat the dangers that lurk. Luckily, I guess – both for my pursuit of adaptable category management 2.0, and for Leafio’s current and potential clients. Matter-of-factly she points out that software alone – no matter how capable – is not a magic wand for solving what’s not there, stating:
»Without processes in place, sophisticated software just automates the chaos.«
This assures me again: Helen is the right guest for Omnibus Insights, a blog covering my journey into Streamlined Category Management – a model aiding physical retRailers to avoid the pitfalls of promotional hype on the road to real value.
— Category Management Insights From A Retail Optimization Expert
In our interview, she shares insights on key topics that will assist physical stores in navigating the turbulent retail sea. While it alludes to some future-proof developments, the overall tone is focused on the present, providing learnings for practical implementation into your processes.
SHO: How can retailers ensure that software solutions based on AI and machine learning really work for them?
Helen Kom: From our experience, implementing software solutions without refining the underlying processes can lead to inefficiencies. Simply automating processes that are not well-designed is akin to automating chaos – a scenario we aim to avoid.
The scenario we use at Leafio – recommended for anyone – is the following. Our core objective is to thoroughly comprehend the current state of customer experience and business processes. Only after gaining a comprehensive understanding of the existing situation can we propose recommendations and implement changes in collaboration with the customer. Subsequently, we are prepared to offer the right software solution.
SHO: In digital times, a “responsive” retail must quickly open the way for the flow of new products, as this is the key to success. Can software assist them with this?
Helen Kom: In the modern world, we are used to deal with a substantial 40 percent influx of new items each year. In consumer electronics or toy retail, this figure can rise to 70 or even 80 percent throughout the year. Thus, making predictions and forecasts becomes imperative given the dynamic nature of these factors.
Our approach to this challenge has involved numerous experiments within our platform.
It’s a nuanced situation that depends on the category, the role of the SKU, and the level of novelty involved. We adopt different strategies for introducing new SKUs compared to when we’re replacing existing ones, which is a relatively simpler task.
Various approaches were tested, and we found that cluster analysis was effective. Unlike relying on categories on the customer side, which can vary significantly, our clustering model categorizes SKUs within a category into clusters with similar demand behavior. We then analyze name, weight, volume, price, and other characteristics based on the specific retail vertical. The model makes decisions regarding which specific cluster the SKU belongs to.
The introduction of this SKU is based on predictions, utilizing average figures within the cluster for similar SKUs.
SHO: What role will human input play within an AI and ML-driven category management?
Helen Kom: From our observations, it seems that in category management, there will continue to be a substantial amount of work—let’s call it intellectual rather than manual—that involves intuition, understanding trends, and drawing from field experience. It’s challenging to precisely quantify the current division, as it heavily depends on operational excellence and the company’s development level. However, as of now, let’s assume it’s around 70% intuition and 30% automation.
Looking ahead, perhaps in the next five years, I anticipate a shift towards a more balanced distribution, something like 50-50 or 40-60, favoring a more analytical approach. This transformation is driven by the fact that tasks related to customer experience heavily rely on the assortment mix. When customers enter a store, they are not just seeking the familiar; they are also eager to discover new and interesting products. While certain items or brands serve as dependable go-tos, part of the shopping experience involves encountering something novel.
SHO: The success of stores is increasingly dependent on intangible products and services – are planograms thus losing its importance?
Helen Kom: Ideally, we aim for comprehensive equipment coverage, but practical constraints exist. Still, we strive to include as many categories as possible in the planogramming environment, even incorporating services like gift cards or lottery without physical items visually represented. Balancing efforts, we focus on categories with tangible outcomes, as having all sellable SKUs on the planogram is crucial for financial analysis. In a recent case study with a US-based customer, our planogram solution reduced the time to create a new store planogram from 15 days to just four and a half days for all categories, a significant achievement for their ambitious retail chain expansion.
SHO: Planograms increase standardization, on the other hand, the customer is increasingly demanding – he also wants surprises and offering that exceeds his expectations. How should retailers respond in this environment?
Helen Kom: We’re avid experimenters, and recently I had a fascinating meeting with a retail customer overseeing 120 supermarkets. Six months into their strategic experimentation journey within one store, the noteworthy outcome was a remarkable 35% increase in revenue. This improvement wasn’t limited to planograms; it extended to addressing issues like out-of-stock shelves in the back room, emphasizing the role of operational efficiency.
These outcomes underline a key idea – strict adherence to planograms and category placement might limit opportunities for experimentation and surprising customers.
The example demonstrates that embracing experiments and enhancing the overall customer experience can substantially impact revenue. Therefore, as retailers, it is essential to be proactive in making changes within the stores to better cater to the needs of consumers and enhance their experience. This perspective, with a focus on customer satisfaction, is a valuable approach for retailers to adopt periodically.
SHO: How should retailers resolve the eternal gap between buying and selling, which can also be called the archetypal internal conflict of every retailer? Is purchasing closer to proactive sales or to the operational orientations of logistics?
Helen Kom: In each and every case, we encounter we face the gap between purchasing and category managers. This gap is only absent in scenarios where demand planners and supply planners operate directly under category managers. From our experience, having a unified categorization algorithm is not the norm.
When considering which department purchasing should be closer to—logistics or category management—I would lean towards logistics. This is especially true for smaller businesses.
Regarding the specific scenario where category managers identify SKUs for each store’s assortment mix, the primary goal for demand planners and supply planners is to fulfill this assortment mix accurately, avoiding lost sales and overstocks simultaneously. From this perspective, it’s more beneficial for them to be closely aligned with logistics rather than category managers.
SHO: AI is everywhere, promoted as a magic stick – and we know it’s not. How can someone who works daily with AI describe main pitfalls of using AI for retailers?
Helen Kom: I agree that there is a lot of hype surrounding the use of AI and machine learning in retail. However, there are several areas where AI can genuinely enhance the efficiency of retailers. Customer insights, especially derived from receipts, demand forecasting, promotional management and promotional demand forecasting, sales forecasting, also. These are several areas where AI can genuinely enhance the efficiency of retailers.
As for the pitfalls, I’d mention three points.
Firstly, these technologies are very data-consuming. You need to have a lot of data stored with a long historical perspective, and it is very costly to save this data. And not all the retailers can afford to store the data for long periods of time.
The second thing is to clear the data because we need to understand that there are a lot of different trash things in the data, and you need to definitely understand how. So you need to have an approach how to clean the data from these spikes and trash.
And thirdly, you need to design the process and afterwards to apply the AI engine to it, not vice versa, OK? Because you will have just the data scientist on your side, and he will try to find some insights. He will not succeed even if he will like make it for five years because we need to align the technologies with the processes and to understand what is.
SHO: The role of category managers is also changing in the new retail world. What are the qualities those entering this field should have? Or reversely: What traits should retailers seek from candidates for category management posts?
Helen Kom: First of all, I wouldn’t recommend becoming a category manager for a person who doesn’t love retail, OK? So you need to love it. You need to feel it. You need to understand that you really are fond of everything that is going on in this retail environment.
Then you need to prove your decisions based on data. So you need to analyze the trends. You need to analyze the performance. You need to see how —- going on now to make some assumptions that can be proved by data, so to have a very data-driven mind.
Thirdly, you need to be a great negotiator. First of all, from the perspective of negotiatitor with the vendors and suppliers. And on another hand, you need to negotiate with the commercial director and with the CEO of the company.
And fourthly, I think that you need to have an analytical mind. It is connected with these statistical and math things. So you need to understand the correlations between different factors.
And of course to work hard.
SHO: We’ve discussed changed shopper habits, technologically advanced solutions. But there is another mega challenge in front of all of us. Can you point out your organization’s contribution to sustainability?
Helen Kom: Leafio was inspired by nature because, you know, it is very important for us to do something that has a positive impact on the future for our children.
First of all, with all our solutions, we are trying to reduce waste. We have statistics that especially in the United States, the percentage of write-offs is approximately 50%, and it is a horrible situation for our environment, for human beings generally, and for particular retail companies who are losing money. That’s why we have tried, and we are making everything possible from our side to develop algorithms to cover not only goods with long shelf lives but also for goods with very short shelf life. And it is not a secret, that’s for supermarkets, they generate not less than 50% of the revenue.
Secondly, if talking about shelf efficiency, it is the paperless technology. We have a mobile application for shelf efficiency like how it is done usually. Before the implementation of such solutions, the planograms are printed out on paper, and each and every equipment has this letter, like a piece of paper with the printed planogram. So let’s assume that we have the supermarket and the supermarket has, let’s say, 2000 types of equipment. So we will have 2000 pieces of paper every day. When the planogram changes, the piece of paper needs to be reprinted, that’s why it is a very paper consumer thing. When we are switching to shelf efficiency, we don’t need paper at all because we have the mobile application and everything is reflected in the mobile application.
— Outro
We had to conclude the discussion with Helen, and sustainability seems to be an OK topic for that. Category management’s role in this area is rife with conflicts, as efficiency and sustainability don’t always work hand in hand. Mostly they don’t. And future-proof “streamlined category management” definitely requires a dedicated place in this aspect.
As a final thought, let’s return to Helen’s advice at the beginning: well-founded processes are a prerequisite for successful implementation of software solutions. I have no doubt that these words show how important is the right amount of commonsense, especially in delicate processes like retail category management. We are clearly navigating into (through) AI-driven retail reality, and simplicity and clear thinking will be even more important now.
—— THE END
Important Links:
Leafio AI: about company, solutions, and expertise
Helen Kom’s Articles on Leafio AI
Omnibus Workshops
Linkedin
