Adding Artificial Intelligence (AI) and Machine Learning (ML) to traditional customer behavior models enables you to generate better predictions from your data and create better personalization for your customers.
Marketing is a juggling act. You want to sell product, but you don’t want to create an over-reliance on discounting. You want to gain loyal customers without having to give margin away, and you don’t want to train people to expect discounts. If you’re advertising 50% off every second week, people will simply learn to wait for the next sale rather than buy at full price.
What you really need is a way to promote just the right offers to just the right people; optimizing engagement without cannibalizing revenue. If you know Jane loves iced tea, you can stop giving her discounts on iced tea; she’ll still buy it at full price. But to get this right, you have to know Jane and her preferences. That means you need customer and contextual data, and you have to be able to get value from it.
Modeling customer behavior
Marketing departments are no stranger to data modelling; most will use some level of statistical analysis to interpret customer data to inform marketing strategy. Common models include propensity, affinity and churn.
Propensity models: what are customers most likely to buy?
Knowing what products people like means you can create personalized promotions to bring offer-sensitive customers into store. It also means you can avoid promoting products to people who are likely to buy anyway, and instead give them offers for add-on, up-sell or related products. The types of data used for propensity modeling are:
- Previous transactions and products purchased
- Propensity to buy similar products
- Lookalike audiences and their purchase behavior
- Historic behavior and propensity to buy different categories
Churn models: who’s likely to leave, and how can we stop them?
Personalization increases stickiness, and keeping sticky customers costs businesses five times less than acquiring new ones. We’ve found that customers who see relevant offers one week tend to engage in the following weeks expecting further personalized offers. Re-engaging churned or at-risk customers is a matter of discovering which customers are most likely to leave, and what’s most likely to make them stay. Churn model data includes:
- Customer activity including purchases and app activity
- Specific behaviors: day of the week, daypart, location
- Tenure: how long each person has been a customer
We’re already doing this. How do we do it better?
When you’re only dealing with a few customers, data analysis isn’t a daunting prospect; you can even run your data models on a spreadsheet. But when you have data from 150m users on your platform, there are better ways to extract meaningful insights and make recommendations.
Adding Artificial Intelligence (AI) and Machine Learning (ML) to traditional customer behavior models enables you to generate better predictions from your data and create better personalization for your customers. ML-powered models learn as they work with data, so as your customers interact with your marketing, your messaging and offers become more targeted and relevant. These models can also pull in contextual data and react to situational changes a lot quicker than someone with a spreadsheet can.
Personalized marketing works
Personalization results in higher spend, more visits and fewer churned customers. When people receive offers that are customized to them, they’re going to be more engaged and come away with a positive impression of your brand. AI and ML-powered customer behavior models make personalization possible on a massive scale. And since customers are crying out for personalized experiences, the brands that can deliver will come out on top.
We’ve run ML-powered predictive models on the Plexure platform and seen first-hand the difference these advanced models make to marketing effectiveness, and to the bottom line.
You may also be interested In…
The “Dopamine Effect”: The psychology behind personalized marketing
The fifth and final blog in our series covering the changing face of consumer engagement examines the “dopamine effect” – how it plays into the psychology behind personalized marketing and how big data can help brands leverage this reaction.
How McDonald’s Japan utilized technology to optimize the customer experience in-store
In a recent virtual fireside chat, Plexure hosted McDonald’s Japan VP of Digital Marketing, Raphaël Mazoyer and discussed customer engagement through digitally enabled service, content personalization and mobile ordering.