For those of us working in ecommerce and AI, we can see areas where the two are beginning to merge.
The first AI work we did with the IRP was DeepAgent™ Version 1. Machine learning has an amazing ability to determine patterns and this capability can be used to predict and intervene with buyers to increase conversion rates. Set up correctly, we have seen conversion rates double - radically increasing the bottom line for IRP Customers.
This requirement to predict and manipulate outcomes in favour of more sales is the core of ecommerce. This needs to be both a macro manipulation of the strategies of the companies themselves and a micro manipulation of individual user sessions.
So let’s look at a couple of examples, starting with the macro level.
Macro level example
DeepAgent is left to watch the high-level data and performance of an IRP website. DeepAgent knows who the key people are that it needs to communicate with and can easily identify the topics that it needs to communicate about.
For example, for a client in a Sports market, DeepAgent sees an overall CPA of 12% for a client. It looks through the Traffic Sources that are being managed by the company and sees that Google PPC CPA is 17% run by Agency A. It then looks at this channel’s CPA% and sees that the market average is 10% for the Sports market. It then checks Agency A’s performance and sees that it is 8th out of 10 in that channel. DeepAgent then begins to analyse all the macro information on the IRP Customer and concludes that changing PPC from Agency A to Agency B would increase sales by 30% and decrease overall CPA% from 12% to 7%. It concludes that moving to Agency B is statistically 95% likely to achieve 30% sales growth and save £2500 each month in PPC costs.
DeepAgent then begins a sequence determining who it should reach out to in order to achieve the result of changing Agency A to Agency B for Google PPC. When this change happens, DeepAgent watches again, and watches the market to see what other changes might increase sales. It continually learns from its interactions.
This simple AI example shows the direction in which the IRP platform is moving to engage on macro-level ecommerce strategy.
Micro level example
On the micro level – on an IRP website itself – the site eventually becomes a piece of AI run by DeepAgent. There is no static site; there is just something that is continually learning through logistic regression and serving back information based off what it sees that works.
When the customer is on the website, on the other side of the browser is just an AI machine playing the customer to maximise the value that they have. Changing the home page to achieve the best one-to-one outcome, changing the order in which elements are displayed based on interactions, changing messaging, offers, suggestions, searches … and constantly watching the customer and seeing what the outcome is. It will combine this with watching a Business Intelligence cloud, adapting to changes in consumer trends and even the weather day-to-day.
The end game
For us in RnD in IRP Commerce, this is where ecommerce and the IRP are going right now. Perhaps a bit like playing a computer at chess, there may be something slightly clinical about it. But without doubt it is where it is all going - ecommerce is about making more sales and profits and AI is the way that companies will outperform their competition and become successful.
