Predictive analytics, predictive modeling and the technology it powers has become one of the most exciting techniques in the marketing toolbox. One recent study found that 32% of US marketers hope to start using their data to predict customer trends in the next five years.
Predictive analytics is using data to improve the performance and effectiveness of campaigns. More specifically, mining and modeling customer data to anticipate outcomes and behaviors. Through the use of predictive analytics, companies can make more effective business decisions, enabling marketers to engage with their customers with relevant content. Some examples include:
Marketers know ‘one size fits all’ is the old way of thinking. They have been grouping customers together, based on attributes thought to make them more suited to a particular service or product, for a long time. For example, people of a certain age and income might respond to adverts for sports cars or racing bikes. Consumers who show an interest in Yoga might also have an interest in healthy eating.
Yet, while these approaches might be drawn from customer data, there is still a degree of guesswork and assumption involved. Predictive segmentation, on the other hand, combines both insight-derived customer needs with real customer responses.
Behavioral clustering uses behavioral data and demographic/firmographic data to identify commonalities and trends to create new target segments. This means a predictive algorithm automatically segments an audience, rather than a marketer.
For example, a predictive algorithm can take into account what channels a customer prefers, how much they spend and how frequently, whether they tend to buy sale items, and so on. When customers exhibit one (or some) of these variables, an appropriate marketing response can be initiated.
Product-based clustering is slightly different, in that it tracks buying trends. For example, customers who mostly buy clothes for children from a retailer, or gluten-free food from a grocery store. A predictive model can be used to create appropriate offers or emails for those in these segments. A good example would be the coupon that appears at the bottom of your grocery shopping receipt. While generic coupons might see a coeliac offered money off pizzas or pasta, a predictive model would present coupons for wheat-free equivalents. One would be far more likely to be used than the other.
For B2B companies, product-based clustering can be useful to identify cross sell and up sell opportunities. By placing existing customers into clusters, marketers can compare trends with the services or solutions that they buy. This can help identify customers who have not yet met the buying potential of their peers, and prioritize the sales team accordingly.
Predictive analytics is a great way to uncover historical trends in your customer’s transactions and behaviors to help better predict what they will do or buy next.
This webinar, delivered by Mark Jameson, Chief Operating Officer at BlueVenn, who will show examples of customer segmentation using predictive models and how to use these to improve the success of your marketing campaigns.
Propensity modeling is making a prediction; a calculation that a customer is inclined to behave in a certain way. Along with tracking and anticipating buying habits, common propensity models include predicting the likelihood of engaging with an activity. This could be opening an email, clicking a link, signing up to a loyalty scheme and so on. Other propensity models include:
Propensity to churn – A model that tells you which of your customers are most at risk of leaving you. This is usually by identifying whether they exhibit similar warning signs as past churned customers (high use of customer support, low activity and so on). By singling out the high scoring risks, marketers can focus retention efforts on them.
Propensity to buy – At the other end of the scale, this model looks at behaviors which show an inclination to purchase. For example, what emails a customer has opened, what they have looked at on your website, how they have responded to your campaigns, etc. These customers are likely to need only a little encouragement to buy (and are less likely to seek a discount).
Predicted share of wallet – This refers to the percentage of a customer’s total spending that they have allocated to your products or solution over a given amount of time. For example, if a company has spent $900 of their yearly $1000 budget on a solution, they are less likely to want additional services that a company that has spent $90 of a $1000 budget.
Data quality considerations
Predictive models can achieve great results for businesses, particularly where decisions have previously been made based on gut feeling or influenced by bias. However, predictive models and segmentation are only as clever as the data that powers them. The trouble with customer data is that it isn’t always accurate and naturally goes out-of-date.
Vital to the use of any form of predictive analytics is a continually updated and accurate customer database. For example, with a Customer Data Platform processed by a Single Customer View. This will ensure you keep making precise predictions and accurate segmentations that serve your company well.
To learn how the BlueVenn Customer Data Platform can make use of predictive analytics to build personalized and relevant customer journeys, join us for our regular live software demonstration. Or get in touch with us to book your own, one-to-one BlueVenn Customer Data Platform consultation.
LIVE NEXT at:
Register to watch how to combine customer analytics and real-time, omnichannel customer journey tools to create personalized and contextual customer experiences in this BlueVenn Customer Data Platform software demonstration.
In the session we’ll define how to:
- Use predictive analytics to make real-time decisions that positively affect the customer journey
- Improve targeting of campaigns using customer segmentation and RFV analysis
- Use real-time personalization to better engage customers
- Integrate online and offline channels into the BlueVenn Customer Data Platform to create a true Single Customer View