What You Can Learn From People’s AI Marketing Successes

29 March 2021Ashley Maxwell11 min read

This article deals primarily with the AI marketing successes we are seeing with large supermarket and superstore chains. Their use of AI marketing is almost a thing of beauty. Perhaps you do not have the data gathering capabilities, or the supercomputer processing power to achieve what the supermarkets are doing in the field of AI. But, you can learn them their methods and apply them to your own marketing strategy. You may not be able to copy their success, but you can certainly improve your own success if you apply their methods in a conservative manner.

Superstores With Club Cards / Points Cards Are Dominating

The notion that places like Lidl and Aldi are dominating is very misleading. Discount supermarkets have started pulling a bigger market share since the 2007 global economic downturn, but in terms of profitability, the likes of Tesco, Sainsburys and Aldi (Wal-Mart) are still dominating.

Since success often breeds success, we have seen some fantastic uses of AI marketing, and a large part of that is due to their expert tracking of user purchases. It started with club cards and points cards that tracked what each user bought, but now that online grocery shopping is popular, the big superstores are gathering more data than ever. As you will learn from this article, the AI supercomputers at work are so efficient that it is genuinely impressive. Yet, it is done with such subtly that only people in AI marketing circles realise they are doing it. It is the same as appreciating ballet. To some, it is a bunch of very thin people prancing around, but if you understand the strength, training, agility and dedication it takes to become a world-class ballet dancer, you suddenly appreciate good ballet when you see it. In short, the AI marketing that supermarkets and superstores are pulling off these days is awe-inspiring, and the efficiency by which they do it is very impressive.

Recording What You Buy Over The Course of 13 Months

Tesco keeps a record of what you have bought over the last 13 months. That number, “13” was not picked arbitrarily. Their market research clearly showed that people’s tastes or circumstances change significantly enough over that amount of time so as to affect buying decisions.

what you have bought over the last 13 months

Plus, the 13-month time period allows the supermarket supercomputers to see how you bought last year. Did you buy Easter eggs last year? No, then isn’t it funny how your advert space in the online shopping portal is filled with crisps, steaks, and gaming offers, and is not filled with Easter Egg offers despite Easter being just days away?

Starting With The Opening Page

When you sign into your online supermarket, a series of apparently generic offers come up. Since the company doesn’t yet know why you are visiting, they offer a broad selection of items. Yet, even that selection of items is not picked at random. They are aware of what you bought last time, when you signed in last time, and roughly how long it takes before you need each item.

The Opening Page on Tesco

If you bought a bumper pack of 250 nappies last week, you will notice that baby items are woefully absent. If it is coming up to the weekend, when you usually buy your weekend meal veggies, note how vegetable offers come up. Again, this is the starting area, so things are pretty broad and generic. However, they are not as bland and obvious as they first appear. What you are seeing is not what other users see when they first sign in to their online grocery account.

The Favourites Page

Many of the larger supermarkets have a favourites page on their grocery shopping website. Take a look at the image below, and you will see the tool asking you to rate your time with your favourites section. It is a rather weak market research tool that is being used by Tesco in this case, but as they say, “Every little helps.”

You think you can dissect me with this blunt little tool

What is more interesting is the seemingly toothless sponsored adverts at the bottom of the page. It probably comes as no surprise to you that the items being sponsored are things that are in your favourites list, but that you haven’t bought in a while. Take a look for yourself next time you are buying.

The Favourites Page

There will be one thing that you bought recently, and then two or three that you bought a while ago, but you haven’t bought recently for a while. The AI marketing tools are offering gentle reminders about the products you enjoy. Take note that getting new purchases is rarely as important as re-selling things people have already bought, especially (but not exclusively) when it comes to consumable items.

Suggestions Based On Raw Data and Intuition

In the image below, you can see that I have bought two types of cheese in the past, and I put them on my favourites list. As a result, the Tesco website spawned an advert for Italian food.

Suggesting Italian Food Purchases

Was this advert born from intuitive human thinking? After all, the types of cheeses I bought are the sorts of things you see in Italian dishes? Or, did the advert come from broad purchasing trends where the AI noticed that people who bought those sorts of cheeses often bought Italian food ingredients too? Was it a sponsorship where the Italian food brand asked that their products be shown near cheese? I hadn’t bought Italian food in a long while, but people in my house had, perhaps that was also a factor? Was it a category thing where the AI decided that whenever people select things from the cheese/dairy section that they are more likely to buy Italian food? Or, was it random luck that this advert just happened to spawn right next to my cheese selection?

Please Buy Again

As you can see from the image below, they are advertising something I have clearly bought in the past. Why would the marketing AI suggest something that I have clearly already bought in the past? And the reason is simple, they want repeat sales. I may be an occasional cheese buyer, and the supermarket wants me to buy it more often so they can enjoy the sales.

Please buy cheese again

There are other factors at play. The AI may have recognised that I have a buying pattern when it comes to cheese. It may have figured that I am overdue for buying my next batch of cheese. What is also more interesting is the price being displayed.

As you probably know, the price of that variety of cheese varies wildly, from packets of different weights, to packets of different flavours and maturities. Yet, the price they show is £2.30. Now, take the three cheeses on my favourites list, add them up and divide them by three, and the average price is £2.50. Coincidence? Notice the 350g weight on the advert. Add up the three sizes of cheese I bought, and the average comes to 320g. Is this all a coincidence? Possibly, but why would it be. Why not use AI in this manner to create the most efficient advert?

Try Our More Expensive Brands

This is a very common marketing trick that supermarkets and superstores play. They target people who have bought the cheaper store brand products, and they push the bigger named brands on them. If you have ever worked in store planning, you will know that sponsors pay a lot of money for supermarkets to push products in this manner. Even things like which products appear at which heights on supermarket shelves is determined by brand negotiations and sponsors.

Try our more expensive brands

In the case of the image above, it is very possible the marketing AI was influenced heavily by the parameters laid out by the sponsors. Nevertheless, it shows the AI doing its job in a very efficient manner.

Three Tries To Catch Another Sale Before The Checkout

This is another Tesco example, but it is a quite common marketing tactic. Before you are able to reach the checkout, the company takes three more shots at convincing you to buy more.

Review Offers Suggestions

The first is the “Review” section where you see what you have bought, and where you may also notice where you “Didn’t” buy certain things. By all accounts, this is actually as useful for the consumer as it is for the supermarket because it is annoying putting an order through and then remembering you forgot something.

The system then puts the consumer through the “Offers” section and the “Suggestions” section. Those will be discussed in the next section, but firstly, let’s give a shout out to Iceland.

Iceland Offers

The Iceland online grocery system does use AI marketing, but it is not harnessed as brilliantly as it is with websites like Tesco, Morrisons, Sainsburys, Asda, M&S and so forth. However, Iceland makes up for it with a few blunter tools, such as their incentives to spend more money. As you can see from the image above, spend more, and you get free delivery, and you are entered into prize draws.

The Offers You Didn’t Know You Wanted

Below is an image of the top offers Tesco gave me. I hadn’t added very much to my basket, which obviously affected their AI marketing algorithm because what they did was very curious.

Tesco Offers

The first two offers were things I have never bought and would never buy, but my girlfriend who lives with me has bought them in the past on her account. They have never been purchased on my account, but using her account, she has had them sent to our house. The marketing AI added something to my top offers that it knows people in my house may like, even if I do not like it myself. Also, with the chicken, I have never bought caged chicken, but perhaps the AI thought it could sell me on some caged chicken because it is cheaper than the free range stuff I buy.

Take a tip from the Tesco’s AI marketing successes and sell to the account holder and the household. I didn’t buy the alcohol in this case, but it was a clever bit of marketing nontheless.

Suggestions Based on Previous Purchases

As with any self respecting AI marketing program, it is going to suggest that you buy things you have bought in the past. Supermarkets and superstores are no different. Superstores are going to wait half a year and then suggest you buy their disposable dehumidifiers again, or their paper plates, or their light bulbs again. The supermarkets are no different, they are going to weigh up what you have bought, what you usually buy, and then suggest products to try and draw more money from their consumers.

Tesco suggested items before checkout

The curious thing about the marketing AI in this case is that it suggests things I have bought in the past, but it almost exclusively suggests very low-cost items I have bought in the past. Notice how there are almost no named brands or overly expensive items on the suggested list.

On this point, I can only theorise as to their AI’s thinking process. Is it because of my previous behaviour where I have only added lower-cost items from the suggestions panel? Is it because it noticed I had spent more than usual, and so assumed I had a budget and wouldn’t want to add tens of pounds onto my order? On the other hand, is it because they are mostly store-brand items, and perhaps this is the area where Tesco pushes its store-brand items? I can only theorise at this stage as to the AI’s thinking, but it is interesting on its own merits.

It’s A Bit Of Clever Marketing – So What?

Yes, the methods and techniques shown in this article are clever marketing. They are highly targeted, they are well-placed, and they are also very efficient. This article is littered with examples of clever marketing, but the point is:

This was all done by AI.

Humans did not make these decisions.

The more data the system has in general, then the better it performs. In addition, the more information it has on you as a buyer, the better it becomes at targeting you. Some of the marketing methods mentioned in this article may have knocked your socks off, and others may have appeared a little obvious. But, the point is that a learning machine came up with these marketing methods. Humans may have set the stage, may have created the framework, and may have constructed the basis for each marketing method. But, it was a machine that made all the decisions. Humans may have built the car, but it is the machine that is driving it to efficiently and effectively convince you to spend more.

How Can We Learn From These Successes?

You probably do not have the resources that massive supermarket chains have, but that doesn’t mean you cannot learn from their AI marketing successes. For a starter, get your customers to sign up for online accounts and start using their purchase history to determine which products you push on them when they are browsing and through email marketing.

Ally their purchase history with page tracking on your website. Tracking what they are looking at isn’t just good for honing your website’s navigation, you can use it to second-guess which products your users may want. This applies both in the moment and in the future. For example, if your user spends a session looking at vacuums, but doesn’t buy, assume that the buyer bought one somewhere else. Then, in two years, when most vacuums start to degrade a little, start pushing vacuums on the consumer just to see what happens.

The methods shown in this article are very sophisticated and very efficient, and yet there is nothing stopping you from trying them. The only difference is that you do not have the computing power and raw data at hand, but all that means is that your AI will not be as efficient. As mentioned earlier, and later in this article, the efficiency by which these AI marketing systems operate is very impressive. Yet, there is nothing stopping you from pulling the same tricks albeit less efficiently.

Isn’t There a Cyclical Problem?

Jenny buys apples and pears. Next time, she sees apples and pears offers, and buys apples and pears. Jenny fancies a change, she sees offers for apples and pears, so she figures she will just get them again.

The cyclical problem is pretty common, and if you want your campaign to become one of many AI Marketing Successes, then you need to overcome it. The larger supermarkets overcome it in four ways.

1 – They sell advertising space to their product producers, which often compels people to buy outside their regularly purchased trends.
2 – They ignore information that is older than 13 months so that certain products appear less prominently in the AI targeting algorithm.
3 – They use big data drawn from local sales, national sales, and from people within the same demographics.
4 – They use the buying trends of other people to alter which items are promoted to consumers, which also helps them capture and exploit current buying trends.

In this manner, the supermarket is able to promote items it knows you want, while also test your buying behaviour in the most efficient manner possible. The efficiency is probably the most impressive element since they could easily promote a wide and random selection of things to see what you buy, and then use that information to sell more to you. But, instead, they take the more efficient route of using national buying trends and demographic buying trends to guess which new products you are most likely to buy. It is both frightening and impressive at the same time.

Final Thoughts – The “Target” Issue Is More Common Than You Know

How supermarkets and superstores use AI is probably one of the biggest AI marketing successes of our generation, but there are many other companies that are experiencing a massive degree of success thanks to their AI marketing.

The “Target” issue is all about how the Target AI marketing algorithm was able to guess that a young girl was pregnant before her father did because of what it learned from her buying behaviour. She hadn’t even a bought a pregnancy test, and the Target marketing algorithm had started sending her baby-related advertising material.

The reason this became national news is that Target acted a little dumb in this case. It would be like your local grocer figuring out you were pregnant before you did and so sent you a big “Congratulations” card through the mail. Target did something almost as creepy, and so it made national news.

Yet, such clever guesses by marketing AI systems are happening all the time, it is just that very few companies are as dumb as Target when it comes to acting on the information. If you are going to learn anything from Target, it is that when your marketing AI pushes new-purchase products on users, it should perhaps be a little more subtle than Target was.