[Summary] Predictive Marketing by Omer Artun & Dominique Levin — 3 Takeaways, 2 Quotes, 1 Question.
How marketers can predict future customer behaviour.
Predictive Marketing by Omer Artun and Dominique Levin is the perfect primer for marketers wanting to learn about the world of predictive analytics.
Predictive analytics is the process of using advanced math in order to predict individual customer behaviour and to group customers into the most actionable and meaningful ways.
While many of us marketers are familiar with doing post-campaign analysis and the process of mining historical data for insights, the key difference with predictive analytics is that it’s forward-looking — it allows marketers to identify future possible outcomes and therefore proactively develop and adjust our strategies.
One important thing to note is that the book’s target audience are marketing VPs and directors. While it briefly covers how to implement predictive analytics in your organization (i.e. building an in-house team, outsourcing to an agency, or buying an out-of-the-box solution), the book’s focus is mostly on high-level strategic thinking and exploring the various tactics that can be unlocked through predictive analytics. It teaches you more about how to think versus how to do.
I highly recommend this book to any marketing managers or directors (especially in eCommerce) looking for inspiration on developing more compelling and personalized marketing experiences.
3 Takeaways
#1: Predictive analytics can improve the effectiveness of acquisition efforts by predicting customer lifetime value.
We know that not all customers or leads are equal. By gathering the right data, marketers can understand factors that signal future lifetime value, which can include:
- Size of the first order
- Time spent on the first visit
- Number of product pages viewed
- Time from first engagement to purchase
- App features used
- Use of a discount code in the first order
From these factors, marketers can better predict the quality of leads or customers acquired from different channels or audiences. For example, one might discover that customers acquired through couponing affiliate sites tend to have a shorter lifetime value with lower margins, and therefore reduce commissions offered through these sites.
The key is that the analysis is forward-looking because it allows marketers to proactively allocate budget according to expected lifetime value (versus reactively from post-campaign analysis).
#2: Predictive analytics can enhance customer segmentation strategies through the use of clustering methods.
There’s a difference between “segmentation” and “clustering”:
- In general, segmentation is the process of arbitrarily deciding how to group your customers. An example would be deciding to segment customers who have spent less than $500 in the last year vs. customers who have spent more than $500 in the last year — why was $500 the selected threshold, and is that the only relevant dimension by which to segment?
- Clustering, on the other hand, is the process of grouping data based on similarities and statistical connections — in other words, letting the data discover the segments. Examples include product-based clusters (based on what types or categories of products are bought together) or behaviour-based clusters (based on things like frequency of purchase, use of discounts, average order value, only shopping on Cyber Mondays, etc.) — these dimensions could also be combined to create more complex segments.
Key to note is that clustering is meant to be a guide to segmentation — simply grouping data together may not result in meaningful or actionable segments.
While it’s still up to the marketer to decide what inputs or outputs are to be used, the advantage of clustering methods is that the segments become more powerful, rely less on gut feel, and — most importantly — can be predictive. Imagine knowing what segment a customer falls into based on their first purchase and therefore being able to predict what items they’ll likely buy in their second or third order.
By knowing the future needs of a customer, marketers can have strategies in place for each segment such as replenishment emails, targeted promo notifications, or next-sell product recommendations — all of which serve to increase overall lifetime value.
#3: Predictive analytics can allow companies to be more proactive with retaining customers.
Similar to acquisition, certain signals can also predict customer churn — for example, not having logged in within the last 6 months, having a recent order refunded, or even the time of the year such as the holidays.
I feel like we can all relate to the experience of calling our phone provider and “threatening” to leave unless they can offer a better deal on our data plan. A lot of churn management programs tend to be reactive and only cut a deal to those who actually call in. Sure, it’s less expensive, but the downside is that it could be too late because your customer has already signed with your competitor.
By anticipating churn ahead of time, the advantage of being proactive is that you’ll likely save way more customers before they’ve even realized that they want to leave. You could argue that the downside is that you end up training customers to anticipate those deals, but perhaps being proactive can afford you to offer lower discounts or maybe even just a “check in” from customer support is enough to save a customer about to churn.
Either way, you can never go wrong with running a controlled experiment to see whether or not these tactics actually improve retention.
Important side note is that not all churn is created equal: some customers may in fact be unprofitable (i.e. high frequency of returns, use of promo codes, or contacting customer service), and it could actually be more profitable to lose those customers while growing your high-value customers (also called “negative churn”).
2 Quotes
#1:
“There is no end to the questions you could ask, and answer, using customer data. The risk is that you do nothing with the data. Just having data alone will not change anything. Only ask questions that will lead to an action, otherwise, it’s all nice to know.”
#2:
“The secret to retaining a customer is to start trying to keep the customer the day you acquire them.”
1 Question
Where is the middle ground between letting your gut versus the data tell the story?
Thanks for reading!
P.S. The link above that drives you to the book’s Amazon page is an affiliate link — meaning that, at no additional cost to you, I’ll earn a commission if you click through and decide to purchase.
