First, they can ask whether it is artificial intelligence or just automation. Any system where a model is applied and an action is taken based on that model is not AI, it is automating. In theory, you could build the model yourself and have the computer spit out a list of fixes to make today. Don’t get me wrong – this is good. The Retail Alliance built this process back in 1992 and we’re still amazed at how few people use it. But it wasn’t smart, it wasn’t intelligent, it just did math really quickly and created an operating discipline for the retailer.

Second, AI will never make your store more engaging, so get over that. You, the retailer, need to do that. Automate the tasks you can automate, then focus that extra time on doing what you can to make the customer want to come to your store.

AI absolutely works well for repetitive tasks, but portraying the usefulness of AI as little more than an automation tool is silly. Like anything new, first movers often make mistakes and sometimes get suckered by salespeople pushing the latest be all end all. Nevertheless, AI is making excellent strides where logical use cases and due diligence drive deployments. As an example, AI is part of the bedrock of the banking industry behind the scenes and increasingly consumer facing. To think that retail is some exception that won’t follow suit is ridiculous.

There’s always at least one bright shiny object in the neighborhood. AI is one, IoT is another. It has value, but not in the ways that have been explained. And what’s worse, you cannot really trust the usage statistics retailers supply on these technologies (IoT and AI). Because many don’t even really know what it is, they say they’re working on it, but probing slightly deeper, we discover that they truly do not know.

So who’s being taken for the worst ride? I think retailers have to use the “sniff test” to figure out what technologies can do for them and, for the most part, I think they do. It’s the surveys that have bad data.

I mean, realistically, do you need AI to tell you that a shelf needs re-stocking? That’s crazy. There are any number of ways to figure that out, with or without an accurate perpetual inventory.

“Retail is detail” is a classic phrase which is still valid today. Retailers should try technology out in a micro-environment and then analyze carefully the result before any full scale roll-out. Secondly, it is not always wise to belong to the early adopters in areas where the technology, benefit or acceptance of the consumer is not proven. Sometimes it might be wiser to first analyze best practices across retail sectors and then judge if a certain AI solution would be an improvement for one’s own company.

Let’s not underestimate the ability of a retailer to evaluate technology that can support and even help drive their strategy.

Retailers of all sizes are well accustomed to the method of “try and test before you buy or commit.” It’s critical to know what outcomes you desire and the levers that impact those outcomes – and how technology plays into that process. This isn’t the magical kingdom, but one of collaboration, hard work and many questions.

Software companies and consultants that are not upfront and honest about what their technology recommendations can and cannot do truly poison the well for those of us that do want to help retailers, manufacturers and shoppers maintain and keep our industry healthy. The old saying “buyer beware” should hold true for AI. Those that are taking advantage of this fact (yes, the larger companies especially) should be taken to task before it screws up another technology that can and will help the industry.

Retailers have only themselves to blame. Retailers have had at least 20 years to learn and understand the digital world. AI and its retail applications are reflective of the fact that retail technologies continue to evolve whether the C-suite has an interest or not. Here is the deal, a person or entity can only be “taken for a ride” if they choose to be ignorant. Digital is not a side note, it is the business of retail today. With digital comes the opportunity to leverage new retailer-specific customized AI driven solutions in many areas from improving ROI in the supply chain to banishing the bane that is the $208 billion return market by offering shoppers recommendations based on a shopper’s sensory preferences. The litmus test for a retailer? Know what you are solving for, which is a human problem. The use of technology is the opportunity for a digital solution to solve a physical world problem. Take the time to talk with vendors, dig deep into detailed use cases you are trying to solve, ask… Read more »

AI’s biggest promise isn’t automation, it’s smart pattern recognition. Notice I said promise, not deliverable — at least not yet. Speeding up routine functions is necessary (assuming you aren’t sacrificing accuracy or service in the process) but not sufficient. Ditto for pure pattern recognition. Without the ability to correctly translate those patterns into insights that inform and benefit the customer’s shopping experience AI won’t be more insightful than a child staring at the clouds and “seeing” horses, airplanes, and space ships. There is a huge difference between the potential of “machine learning” and “Artificial Intelligence” which will remain no matter how many times the terms are used interchangeably. In machine learning, systems optimize based on the recognition of patterns. In Artificial Intelligence — as its most rapid proponents see it at least — systems transcend pure functionality and translate those patterns into new insights or activities. If machine learning is all about running fast, AI ought to be all about choosing new directions to run in. Machine learning gives us data. AI ought to give… Read more »

Great article and I agree with the guidance and concerns. We also need remember that today’s AI is primarily yesterday’s Big Data with a shiny new label. So all of the challenges of Big Data apply to AI, and AI is no smarter than Big Data.

Computers have always helped companies thrive by automating repetitive tasks. That’s what they do better than people. We need to keep that in mind.

Artificial Intelligence has great promise to help retailers be more nimble and respond to customer/consumer needs. However, a retailer must step back and have the foundational data pieces in place before trying to implement any AI solutions. I mean foundational pieces such as clean POS data, clean loyalty data and adequate IT systems to allow AI to deliver against its promises. A retailer needs deep pockets to set up their own AI shop with the expertise to manage AI on their payroll. Most retailers will look to the outside for help and a retailer needs to find an outside solution provider that is strategically interested in helping the retailer get their foundational data right FIRST before implementing any AI solution.

There are far more fundamental operational and customer experience challenges for retailers to take on before they should consider any AI focussed initiatives. The need for personalization is driving retailers to become increasingly proactive, predictive, and agile to create outstanding customer experiences. However, AI is the next wave of the evolutionary path for retailers, and this calls for a crawl, walk, run strategy, as there are far more change management and organizational considerations to account for.

Purpose-led AI initiatives that solve real business cases are what is needed in the retail space. AI could potentially enable retailers to be one step ahead of the consumer, and drive personalized engagements that keep them coming back for more. Until the execution strategies are in place to operationalize any AI insights, it will, unfortunately, remain a shiny object and another takeaway from the NRF 2020 around the future fo retail.

AI is not for the faint of heart and requires deep understanding of data, both customer and operational. The best analogy I can think of is a precise surgical tool to solve very specific issues. It can be applied more generally, but you can use a kitchen knife instead, as well. In retail there are specific examples where AI has shown success: demand forecasting, automated recommendations, and similar. Applying the tech to other scenarios takes time, data scientists, and clear understanding of the problems that need to be solved. The best AI solutions won’t be sold as AI but as an embedded part of a data or analytics solution. It will be invisible but impactful to the bottom line. Retailers are typically not being taken for a ride. Smart execs already understand how to think through the best avenues to apply, test, and validate the success levels of AI or alternatively purchase the tech as part of a broader set of solutions to mitigate risk. It can be a powerful value for retail, and the… Read more »

As with most technologies, AI is not a solution for all that ails retail. To be most successful, the use of AI must be aligned to fit problems versus a solution in terms of a problem. Key areas to explore include efficiency of operations and better shopper experience.

Just measure it. It’s that easy. Run a side by side comparison vs your current algorithms and make the vendor prove the difference. Anything else is irrelevant. Boil the AI’s performance down to a few basic variables, like conversion rate and unsubscribe rate. Then benchmark as many vendors against it as you like.

AI has been helping innovative retailers for literally years. Without naming names, a HUGE retailer has increased their demand forecasting accuracy by more than 17% with AI. Another retailer is driving SKU movement well below their top-tier movers, and realizing margin gains due to increased movement in the slow moving, high-margin items. AI is happening. 1) Define the business problem you need to solve, 2) Determine your desired outcome, 3) Investigate all potential technologies to address the challenge and see if AI bubbles up to the top.