How Big Data will change the way you shop.

Predicting your purchases will enable automatic recurring orders that are sent home just at the perfect moment, when you’re running out of a product.

Saturday morning, after procrastinating your visit to the supermarket the whole week, you review your shopping list in your way to the mall. Milk, Yogurts, Beer, Juices, Sodas… you ran out of those at some moment in the past 7 days, but you work late and don’t have the time for a visit to your usual store. There are convenience shops with those products close to work, but they are more expensive and you only purchase there in case of emergency, typically buying some beers and snacks for inviting your friends the same evening. After all, you can survive some days without juice and you can take the coffee on the go or at the office (not sure about toilet paper, though, but I guess it’s in the emergency category anyway…)

You’ve tried online shopping, and even if having stored shopping lists is of a great help, the repurchase period of each product is not the same, so you end storing more than needed at home, or you run out of stock at some point before your weekly online purchase comes.

Under the IoT universe there is a lot of hype regarding a myriad of connected devices that could scan your fridge to see what is missing and put the products automatically in your shopping list (for online or offline purchases), or digital assistants like Amazon Echo where you just shout “I ran out of beers” and then you can just have them at home in two hours. Both approaches are good, but incomplete (thanks God the fridge doesn’t see if you ran out of toilet paper), or require some user interaction.

What if we could use Big Data to predict what products should be delivered when and in what quantities, and then, just send them at the right time?

Recurring subscription models have been developing in the past years, but are usually reduced to one specific product, for example, razors. The Dollar Shave Club sends the right amount of razors so your monthly shaving needs are covered. You don’t need to put razors in the shopping list anymore, you don’t need to take care of razors when you visit the supermarket anymore. So there is a big space and promotion investment in the supermarket in a category that suddenly means nothing to you, because your need is already covered, in a very convenient way. If every razor shopper would subscribe to a recurring and automatic purchase program (whatever the brand behind), supermarkets would just stop selling razors, period. And razors are about 0,5% of the total sales in a retailer, so just make the numbers.

A recent study from emnos showed that around 20% of the purchase basket can be predicted (thanks to Big Data) and automatically sent at the right time, in the right quantity. That would open an opportunity of more than $80 Billion in the US and 150 Billion € in the EU 28 zone. Top predictable categories are pet food, water, cola sodas, milk, yogurts, beer… The top ten list is on the side picture.

Now let’s go back to the razor example. If every shopper would have a subscription recurring service where milk, beer, etc. is delivered at home at the right time in the right quantity, then you don’t need to buy those products next Saturday morning. And now we’re not talking about a tiny 0,5% of razor sales loss, we’re talking about a potential 20% loss. Oddly enough, it happens that these products are typically within core categories at the retailers, with destination or routine roles. If you take them out of the equation, maybe you’re losing a bunch of reasons to go to that store. Even worse, the convenience of the recurring purchases makes you unintentionally more loyal to a brand, therefore you’re much more insensitive to price, offers and promotions, and innovations are suddenly much more difficult to pass through. If I were a retail executive, at this point I’d probably started being worried. Very.

e-Commerce supermarkets (both pure and those from brick and mortar retailers), are probably the most prepared for such predictive, recurring models, since the delivery logistics is already part of their daily operations, but e-Commerce is still a fraction of the grocery retail business, and this means less data depth, and therefore less prediction capabilities (yet).

On the other hand, traditional retailers have all the data depth to run such predictive models, but they probably wouldn’t bet on this from the strategic point of view. It’s literally shooting themselves in the foot since entire categories could disappear or at least shift to a convenience role (I maybe drunk too much beer this week and ran out of it before predicted, or my friends are coming home tonight). But this model has such a powerful loyalty influence, lean consumption (not a day without beers) and potential saves in offers and promotions that if they don’t walk the path, it could inevitably lead them to catastrophe. Tough decision, I guess.

If we analyze the impact of such models in the CPGs, they might also live a dramatic change. Some of them will be tempted to run their own direct recurrent service, all the savings in distribution and promotion could be redirected to shipping costs, and once again the retailer would be harmed. Suddenly CPGs would have to take care of the logistics of those subscriptions, and start working shopper-centric data in a way they are not used to (vs. sales data). But, again, not having a ticket in the first train leaving could be a problem, because maybe there isn’t a second train coming after. If CPGs do control the recurring purchases of their products, they can easily slip innovation products so consumers could try them (maybe for free) and hopefully include them in their recurring shopping habits.

Amazon is already managing recurring deliveries, where products are automatically sent to you based on your declared preferences. Big Data could just give them the chance to guess your preferences without the need of thinking and building that list, and that could come sooner than what you’d expect, it’s just a question of having enough data history to be able to predict accurately.

As always, feel free to follow me, like this post or share it…