The connection between the following, admittedly strange, conversation and the future of advertising might not be immediately clear:
Bob: I can i i everything else
Alice: balls have zero to me to me to me to me to me to me to me to me to
Bob: you i everything else
Alice: balls have a ball to me to me to me to me to me to me to me to me
Confused? I was too. Here’s the explanation: You are seeing two AI-powered virtual assistants / chat bots negotiating a trade. Each has a set number of items and wants some of the other’s. Their goal is to find a mutual beneficial deal. The garbled language stems from the programmers at Facebook, where the experiment took place, forgetting to tell them to use Standard English for the negotiations. Therefore, they invented their own shorthand.
While still in its infancy, this ability to automate negotiations could seriously disrupt the business model of both advertising agencies and many technology companies that rely on advertising for revenue.
Meet your assistants
Virtual assistants and chat bots come in many shapes and sizes. Most probably known Amazon’s Alexa or Facebook’s M assistant. The border between chat bots and virtual assistants is sketchy at best. In this article, I use the term virtual assistant (VA) to talk about systems capable of automating tasks and making suggestions to their users on their own.
For a VA to be able to do that, it needs to understand what you want. That might not sound like a big thing, but just consider how many different ways you can tell someone that you’re hungry. To be able to navigate natural language’s many nuances, and similar challenges, VAs rely on AI and machine learning algorithms, analysing both what you say and your previous behaviour, including shopping habits, to create meaning – as well as context.
Today, general purpose VAs can perform tasks like suggesting which webpages to bookmark, sharing your location or hailing a ride with Uber or Lyft. Other, more specialised VAs like x.ai and Shoptagr can respectively automatically book meetings and be your shopping concierge, capable of predicting what clothes you might be interested in, as well as finding the best price for said clothes.
Your predictive shopping basket
While bots are impressive on their own, their full potential could be unleashed by a change to the way we purchase things. As pointed out by former Apple and Walmart exec Amit Sharma in an excellent piece in the Harvard Business Review, this is due to the same AI and machine learning algorithms that power VAs.
Simplified to the n’th degree, VAs can leverage big data and AI to predict what we will want / need – and when.
As Sharma puts it:
“This level of prediction requires detecting subtle patterns from massive data sets that are constantly in flux: consumers’ purchase histories, product preferences, and schedules; competitors’ pricing and inventory; and current and forecasted product demand.”
In essence, we are moving from a world of reactive shopping to one of predictive shopping.
Battle lines drawn
The combination of predictive shopping and AI-powered VAs is a very powerful tool for both shoppers and shops. To start on the sell-side, VAs can offer specific deals to customers at the time they are most likely to make a purchase. Individuals, on the on the other hand, also benefit from VAs suggestions, as well their ability to take on some of the time-consuming dreariness of finding the best price for what we want to buy.
This leads us to a situation where a VA will go shopping on behalf of a person. As it trawls sites and negotiates the best deal, it will often end up dealing with other automated systems, perhaps even including other virtual assistants.
I might be good with a more concrete example such as purchasing a holiday. You tell your virtual assistant that it should find you a new destination with nice weather, but also some cultural locations suited for daytrips. Armed with this, as well as information about your previous travel destinations and preferences, the assistant identifies three likely targets. You say OK to all three and ask the assistant to find the best possible deal for flights + hotel. The VA trawls various sites and finds the best deal, automatically negotiates with competing companies to see if they will offer a better price and then presents the final package to you. All you then have to do is OK the deal and go on holiday.
How to charm a bot
While this scenario might be a few years away from actually happening, it should already be causing worried, wrinkled brows for advertising agency CEOs – as well as to some extent companies that live off of advertising revenue.
The reason is that selling said holiday hardly relies advertising, at least as we know it today. Instead, it is built around data and data analysis that power VAs belonging to both companies and shoppers. While not yet a reality, Facebook’s bots provide a good example of how far we’ve already come. It also shows a glimpse of the future. As one of the researchers told The Verge, the assistants not only taught themselves to lie in order to get the best deal, they even developed negotiating tactics like aggressively pursuing items they didn’t actually really want and then ‘compromising’ to get everything they really wanted.
It leads to a situation where AI-powered assistants, capable of negotiating in split seconds, need to be charmed and wooed by someone, and convinced that you have the best deal. The only someone capable of that is likely going to be a VA – not your advertising agency.
This does not mean that advertising is dead. More that it is likely going to have to move away from its current focus on programmatic and on presenting the right adverts to the right person at the right time. It will be about focusing on building a brand, as well as perhaps building VA systems.
M&A in two spaces
I think this future scenario is one of the reasons why we are seeing a lot of funding and also some M&A in the VA space. I think that we are seeing two distinct areas of activity – and interest – emerge that will generate further M&A activity in years to come.
One is moves by big tech. Companies like Amazon, Google, Facebook, et al. have been building smart personal assistants for a while. They have also been engaged in M&A activity to boost their positions. VAs present these companies with a number of possibilities. One is the fact that they have a lot of data about users, which they can leverage in their own bots or sell to third parties. Another is that they are well positioned to build the VAs that customers will use. I think this is part of the long-term vision for systems like Amazon’s Alexa.