How transactions handled by AI assistants will impact marketing

If you own an Alexa device today, together with an Amazon Prime membership and setup your Echo with 1-Click Buying in your Amazon account the right way, you can already experience today how it feels like to let an AI assistant handle a transaction on your behalf:

"Hey Alexa, order more coffee please"

In the not too distant future we will be able to escape the narrow frame of Amazon with a Prime membership and actually use so called "Large Action Models" to let AI assistants handle all kinds of transactions on our behalf and solve real problems for us:

• make an appointment for a haircut next week
• reserve me a seat for my train ride tonight from Munich to Berlin
• book me a flight to New York for next weekend
• do an analysis if there is a cheaper phone contract available
• calculate the costs of a family trip to Mallorca in June
• get me a car booked to the airport for 6pm

AI assistants will become more and more the primary platform we engage with and device like the hu.ma.ne AI pin or the Rabbit r1 will offer exactly these functions with the help of LAMs moving forward.

The impact on search, the web and other touchpoints will be noticeable

Trying to research numbers on how many people actually make use of the ordering feature of Alexa, I unfortunately only got outdated and mixed results (for example here and here) from a couple of years ago.

Apparently it wasn't a big thing back then (yet) and trust issues came along with it. But that's years ago and the incredible rise of ChatGPT might change the framework for that, as the usefulness will increase, the error rate in understanding natural language improved and we gain more trust and experience in working with AI assistants in general - it's simply a next logical step.

Going through the list of examples from above: Today we would search for a solution, we would use a website or app to get the job done. How is this about to change I ask myself these days.

I just wrote a few weeks ago on the apparent comeback of The Intention Economy, and that this old idea of us simply voicing an intention and in the background brands will compete to offer me the best possible solution according to my preferences, will finally be possible to implement.

Currently the impact on search engine traffic is not truly noticeable, despite the big success of ChatGPT and millions of people using it every day. The experience of search is changing though and this long learned habit is simply evolving for now with new capabilities through LLMs - and it might even very well be, that for taking actions with the help of AI assistants search will remain a key entry point for quite a while.

Marketing will need to wrap its head around this new touchpoint within the Customer Journey

Nevertheless I truly believe that for two stages of the customer journey direct interaction through natural language with an AI assistant will play a key role moving forward:

1. Purchase Stage

You already know what you want and simply want to get it. Here it will be a key task for marketers to be top-of-mind with consumers, so they say:

• book me a flight with Lufthansa to New York
• get me roasted coffee beans from Segafredo
• buy me a ticket for Napoleon in the nearest UCI cinema available
• get me a BMW from Sixt at the airport later today
• reserve a car for 6pm to the Airport with Freenow

We will want to get consumers to voice their brand preferences in these natural language interactions and for this it will be key, more than ever, to have strong loyalty and awareness.

I expect in key industries like for example travel or FMCG a change in priorities in how they allocate their media spendings and which tactics they apply in order to influence these kind of interactions with AI assistants moving forward.

2. Consideration Stage

But also for the consideration stage - which is currently the primary battleground for a lot of performance marketing - I see the potential for AI assistants to gain a lot of momentum with consumers especially as more complex the problems get they are able tackle for us, so that we don't have to research prices, compare best deals and take care of all the individual puzzle pieces ourselves:

• I want to travel for a 4 day shopping trip to New York with my wife right before Christmas. Please propose the best options not succeeding our budget of 4.000 Euros for the entire trip
• I need for our upcoming summer holiday a new set of swimsuits, diving gear and other water toys for my two boys 6 and 9 years old
• my microwave just broke down and I need a cheap replacement by tomorrow delivered during the day to my neighbor, while I am at work
• I am planning the kids birthday for my 7 year old daughter and we want to play some games with up to 8 children in our garden. Please propose three games for a birthday party and put together all the required materials and gadgets for it

Whatever our intention is, we delegate the research task to an AI assistant and expect to either act based on known or learned preferences - or we will be presented with options through natural language (if appropriate given the complexity of the task) or on a screen to choose from different options. That might happen on the phone, tablet or even TV.

The experience you can get from Bing Chat (try it if you haven't yet) - which is powered by ChatGPT and labeled Copilot now - already comes pretty close to this idea, misses though the part to be able to take action and put together full solutions for more complex intentions.

This will fore sure generate different marketing approaches and I am sure that search engines, upcoming AI assistant vendors like humane or Rabbit, as well as established players like Amazon, Google and Apple will come up with new ideas and formats for advertisers to compete to be selected within the process.