Make smarter, more targeted campaigns with AI and predictive analytics

If you’re a marketer, your dream tool will predict customer behaviour using artificial intelligence. It will allow you to target them in the most effective way just as they’re about to buy or churn. And you won’t need any technical expertise to do it.

Profusion has just entered into the next phase of a knowledge transfer partnership with the University of Essex. Combining our expertise we’re building upon our product that’s already in development – AI Marketer – so you can use AI and data science to do what you need.

Profusion head of data science Henrik Nordmark explains more in this Q&A with Louise Scott

What is AI Marketer? What benefits does it bring to organisations?

It’s a predictive analytics tool that helps you inform your marketing, improve customer retention and ultimately increase your income. It enables you to target your customers at the right time. Basically, you can really maximise your impact (and comply with GDPR at the same time).

Who is it for?

It’s for marketers at any organisation. We expect it to be mainly for retailer and travel companies, supermarkets – anywhere there are relatively frequent purchases, repeat purchases. It wouldn’t work well for wedding or funeral companies, for example, because a customer doesn’t come back again and again! The predictions become a lot more difficult when dealing with long customer lifetime cycles.

When will it be available / launch to market?

Right now we’re beta testing it with several customers and are planning to launch it commercially in 2020.

How does it enable smarter targeting?

It tells you when you’re most at risk of losing a customer or when they’re most likely to make a purchase.

In its current form it has three models. There’s the customer churn model, which figures out the stage of the customer life cycle a customer is at and when they’re going to leave you. Then there’s the propensity to purchase model, where we make very short-term predictions about the probability of a particular customer coming into your store in the next five, 10 or 20 days.

The third model is the customer lifetime value model. This takes the long-term view of things, looking far into the future at the behaviour we would expect from a person in one, five or 10 years’ time. Based on their past behaviour, we can predict how much they’re going to be worth to you in the long run.

How will it develop during this next knowledge transfer partnership with the University of Essex?

What I’ve explained are things we’ve already built so far. This KTP will build upon that, adding some new modules. We want to develop a send time optimiser that shows the best time or day of the week or month to send customers an email. It could be that people only buy something when they’ve recently been paid, for example. We want to look at integrating open data sets such as weather and Twitter data – anything that can help the predictions to be more accurate. Also, we’ll use natural language processing to understand the comments customers make about products they buy. Anything that will make the job of a marketer easier, so they can make decisions that aren’t reliant on gut instinct alone but based on solid data.

What’s the history of AI Marketer so far? How did it come into being?

It came about as we started doing these kind of projects – predicting churn and propensity for different clients. Every time we did it from scratch, bespoke to the particular client. But we realised if we put extra effort into this, we could make the algorithms a bit more generic and create a churn model that doesn’t just work for one client but for hundreds of clients. So we looked at how to package them in a way that many companies could use them, make them more cost effective, and sell the tool as a piece of software at a lower price. It makes it more accessible to smaller companies.

Why and how are Profusion and the University of Essex working together to develop the tool? What do they each bring to the table?

There are things we could develop on our own at Profusion. But there are areas of niche expertise and benefits to be had from incorporating research that professors at the University are doing on machine learning, natural language processing and statistics. And at Profusion we try to give a real-world application to these new ideas being developed by academics – this gives us an edge over other companies that are working on similar tools.

Interested in finding out more? Contact us and we’ll arrange a chat.

If you like this article, try these:

Please rate this

Author

Louise Scott

Comments
(0)

View more recent postings

+ Add comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Loading...