Substitutions are a pain point, and it’s a difficult problem to solve. While AI often gets bandied around as a solution to solve anything, this is one case where the application of AI could have a meaningful impact on the substitution challenges. Taking the guess work/judgement out of the stock pickers’ hands will be helpful since even well intended stock pickers can make suboptimal choices. At the risk of claiming this to be a panacea – I think it’s an especially good use case.

Out-of-stocks and substitutions are a huge frustration for online grocery shoppers and in-store product pickers. Artificial intelligence should help improve substitution decisions. However the root cause is inventory accuracy. Without real-time inventory accuracy, grocers, shoppers and pickers will continue to experience frustrations with out-of-stocks. Accurate, real-time inventory is the Holy Grail.

Substitutions represent one of the largest compromises retailers ask customers to make. Brand, size, price, etc. all contribute to the customer’s decision-making process. To the extent this can be modeled, the need to do something they don’t want to do or compromise is lessened. Any method or process to handle out-of-stocks needs to be clearly and proactively communicated by retailers.

Yes, AI can certainly help as it can quickly assess more data than humans can and come up with recommendations based on past habits and customer profiles. However it won’t always be perfect and it is unreasonable to expect it to be. The best position for retailers is to maintain good stock positions so that substitutions are minimized.

AI can solve a lot of things, but predicting brand and flavor preferences of individual consumers is probably not one of them. Food is a very personal item for most consumers. It’s very difficult to understand the attributes they use to choose individual items, and get it wrong and you’re charging a customer for something they don’t want. My recommendation is to ask. I was getting texts from Instacart at one point about what to substitute for missing items. You could also ask the customer if they want to allow substitutions. Bottom line, work on supply chain issues to improve availability.

Huge problem, potentially good solution. Two caveats – you need thousands of substitutions for data before the AI solution will be any good and if you keep having out-of-stocks, shoppers will stop coming to your site.

Substitutions are a definite pain point for online grocery, and it is an area that needs to be addressed to help minimize shopper frustration and keep people using the services long-term. AI can help with that, especially for retailers that are filling orders from stores since leaving it to human shopper judgment is too risky. Costs to remedy consumer complaints can be higher too – if someone isn’t happy with the substitution retailers make it right with a refund – but that isn’t a sustainable solution from a profit perspective.

Irrelevant substitutions created a lot of frustration last year for online shoppers, personal shoppers, and grocers alike. Many customers were not happy with the substitute – in some cases, it wasn’t even a substitute!

While using AI to solve that problem is commendable, I wonder how brands are planning to handle it. A McKinsey study found that 46 percent of U.S. consumers had migrated away from their usual brand during the pandemic due to product unavailability. Of those, nearly 50 percent expect to make these switches permanent.

Fascinating. I question whether the “real-time customer approval” of the substitution can actually work in practice because on-shelf availability can change so much – is the picker going to wait by the shelf until the customer has approved the substitute? Unlikely. As a shopper, one of the biggest issues I have seen in this area is being prevented from choosing a given grocery item at selection time based on in-store availability at that time of order rather than projected availability at the time the order will be picked. Providing visibility of what is in transit from a DC or supplier is a key ingredient to fix this before AI can have an impact. In terms of providing recommendations of substitutes, using AI makes sense in terms of driving consistency in approach. While the video talks about 100+ factors involved in making the recommendation, it would be fascinating to understand how deep these go – for example is the wider basket considered, as that could well influence what is and what is not an appropriate substitution.

Substitutions are serious. Retailers must find a way to have one single, correct inventory number as close to real-time as possible. AI isn’t magic. It has to follow the “garbage in, garbage out” rule, too.

AI should be used to augment retailers’ ability to approximate real-time inventory at the item level. By augment, I mean using RFID and other technologies to better track inventory throughout the supply chain in the DCs and MFCs, and in the stores. MFCs have the potential to be the keeper of real-time inventory.

Substitutions? Real-time inventory certainly is the only true solution. Short of that, the only fair choice to the shopper is either “no substitutions” or being able to specify exactly which substitutions would be acceptable. There should also be zero back and forth between the shopper (consumer) and the “shopper” (Instacart or whoever is actually picking the order). What the shopper wants is called primary demand — primary demand is what the retailer must know for accurate buying and assortment planning.

From a consumer perspective, this is a huge pain point especially if you have food allergies. AI would remove the guesswork and simplify the picker’s role where they are hired to just pick product ordered. Infusing technology to better understand the customer as they shop is exactly what we need across all verticals of retail, not just grocery. Smart substitutions or upselling the right products aligned to the customer is how you keep the customer long-term.

AI is a powerful weapon in retailers’ substitution arsenals. Currently, Walmart’s substitution parameters are black and white (allowed or not?). Mr. Venkatesan’s comments explain why parsing substitution criteria is a slippery slope. A good first step is for retailers to make it crystal clear how shoppers can approve/disapprove substitutions and therefore absolve retailers from backlash. The primary opportunity for AI is to avoid out-of-stocks in the first place. The other side of the issue is to ensure that retailers get the sale despite lack of availability for particular items. There is a big gap in perception between retailers having the shopper’s back or pulling a fast one. It’s worth the effort to stay in the first category.

The pain level is directly proportional to the effectiveness of the retailer’s logistics and supply chain, but substitutions are a critical part of the customer experience for e-grocery as well as physical and omni-grocery. Based on Deloitte’s 2020 Fresh Food Consumer Survey, about 72 percent of stock-outs are substituted – so this definitely drives an important part of the business. As most of us know, stock-outs have been a thorn in many a retailer’s side. AI can help on the substitutions if properly configured and managed. Asking the customer can also help, especially for stable assortments. However at the end of the day this is entirely a logistics, inventory management and demand forecasting issue. Reduce the stock-outs in the first place. AI would be better suited to help manage the demand forecast before worrying about how it addresses the substitutions.

Seems like real intelligence could play a bigger role before AI steps in. How about some form of ranked choice voting? The need for substitutions is now a known, no-surprises element of e-grocery shopping. It’s great that some grocers seek text approval for substitutions. And “no substitutions” is an option. But how about pre-approved substitutions? Most customers will know in advance what substitutions they can live with. They’ve probably shopped the store dozens if not hundreds of times. They know what else is on the shelf. They know what they’ve tried and accepted and tried and rejected in the past. So how about they spend 10-15 minutes filling fill out some kind of questionnaire so that the grocer has real intelligence on file instead of having to resort to AI as the first alternative? Customers can’t know all the options but they can help by setting some boundaries up front.

Yes — and no. Properly applied and managed, out-of-stocks can be satisfactorily alleviated with AI. However if the core data and rules are “bad” (which is too often the case when combining disparate data sources), the results can be a disaster.

To address this haunting retail pain point, AI efforts must begin with a thorough understanding of incoming data sources, a single and universal “source of truth” for every item, and a merchant to establish the rules. Short of this, someone may get calculators instead of pasta.

E-grocery out-of-stocks, expired items and product omissions are major pain points. Substitutions are a moderate pain point that can make us reconsider in-store shopping to gain certainty of what we’re buying.

Yes, grocers need to consider investing in tech to stay competitive by aligning with consumers’ expectations. AI allows exact product matching, so grocers can offer substitutes that satisfy shoppers’ specifications.

If AI can help with accuracy and efficiency, it should play a role in the e-grocery shopping experience. Any shopper that has used e-grocery for any period of time knows substitutions will be a part of their cart – so if a grocer can reduce the dissatisfaction with the substitute – it’s a win for both the shopper and the grocer.

In my opinion, substitutions are among the greatest threats to online grocery shopping customer satisfaction. Walmart is smart to invest in technology to help improve this significant pain point. I think AI can help, but I am not convinced it will eliminate the problem entirely. However even modest improvements will have a significant impact on the shopping experience, and could potentially be differentiating.

Out-of-stocks have been a problem in grocery retail for years. Most research shows that one in eight shoppers leaves the store without at least one item they went in for. Ordering online has made the OOS challenge exponentially worse, and while AI may help in the substitution process it only addresses the symptoms, not the problem.

Stock-outs and substitutions are likely the number one issue causing dissatisfaction in online grocery purchases for consumers. AI is a good technology to apply to this challenge, however there is the issue of training the AI to reach an effective success rate. Kudos to Walmart for promoting this as a solution and setting the expectation with customers that their substitutions will get better over time. The question is, how long will it take, per customer, for the AI to make not only better choices but the right choice in that customer’s mind? The best way to solve this problem is to prevent stock-outs. I would hope that retailers leveraging AI for the substitutions are also using AI to create better forecasts for replenishing those store shelves with the right products to avoid the stock-outs in the first place! This is very much like discussions around returns in e-commerce – the more you can prevent them (using AI), the better your customer experience!

AI is definitely a good option for optimizing the substitution process and improving customer experience. The problem is that many retailers still haven’t automated the problem one step behind substitutions – real-time stock availability. It is hard to act on substitution pain points if you don’t even know when you’re actually out of stock.

Substitutions may be a deal breaker for “e-grocery” — nice to see the “e-” is making a comeback, by the way. Consumers aren’t going to reward you for being out-of-stock and coming closest to getting them what they want because, well, they want what they want. AI can get us to a “better worse choice” but the only way to get to a good choice is to give the consumer what she or he is asking for.

Substitutions due to “out-of-stock” are frustrating to both the customer and the retailer. However they also present opportunities. It’s one thing to say, “We are out, so try this.” It’s another to explain why it’s a good option. With the right message, this can actually be an opportunity that turns frustration into delight.