What’s better marketing than anticipating future behavior? Reacting in real time to consumer needs! Look around you. We see it especially with retention marketing among retailers, but more and more with publishers and travel marketers as well – there is more of a dependency on “right now offers” than on deep predictive analysis. Could predictive models be dead?!
I believe past behavior is no longer as reliable a predictor of future behavior. Certainly, it’s getting harder and harder to maintain those 20 percent, 30 percent, or higher lifts in response that marketers had been enjoying over the past five years by using predictive models to improve our targeting and segmentation routines. However, more influential in this trend is the fact that our digital lifestyle is so fragmented, so intertwined with offline behavior, and so enormously spontaneous as we are bombarded by offers that are ever more relevant. Consumers aren’t behaving as they did in the past because the opportunities for new exploration seem endless.
Consider that shopping now happens on the device – while in the store. Consider that GPS overlays have turned commutes into veritable feasts of localized and personalized data. Consider that the more data a marketer has on me, the fewer options I’m offered – relevancy is so refined it’s almost boring.
I’m not saying that predictive modeling has no value. It’s typically used to answer questions like:
- Which one of my customers has the greatest likelihood of becoming a platinum reward member?
- Who has the highest propensity of becoming a customer from a list of prospects?
- What is the probability that a single-purchase customer will become a multiple-product purchaser?
- Who has the greatest likelihood of leaving my company for the competitor?
Essentially, where a descriptive model will tell the story (what is available to buy?), the predictive model identifies potential customers (who will buy?). Predictive modeling typically builds on the descriptive models – where descriptive models tell the story, predictive models identify potential buyers of the story by assigning a score to those most likely to exhibit the buying (or other) behavior in the future.
In the past, that analysis was used to create future experiences – email, mobile messaging, web advertising, and offer or content placement.
Today, however, we are using data at a more granular level. Marketers I’ve spoken to recently are transitioning away from predictive models in favor of becoming more automated in our routines. The models become more real-time analysis: a watch-and-react mode.
The more transactions that a marketer can observe, the stronger this reaction will be in both speed and accuracy. Apparel retailers have an advantage in this regard, and so they are among the first to shift toward automated scoring and offer placement. Publishers and travel marketers are also actively testing.
At the end of the day, data – big or otherwise – is only valuable to marketers when they can use it to make better decisions, and create more satisfying customer experiences. Even when introducing new products – where there is no historical data – marketers can use data to improve the messaging, timing, and placement of the offer.
Consumers seem to benefit from this as well. Frankly, they expect connectivity to “the mother ship” when they are online or in a retail location – or an airport or restaurant or coffee shop. They want to recognize and celebrate no matter the channel or the interaction level. They anticipate that marketers will create touch points for them to lead them to familiar as well as new experiences – and they assume some level of personalized service. There is always the risk that marketers can cross the line between comfort/service and creepy – so the responsible use of data is always paramount.
All business today is data-driven. There are layers in the marketing department of how we use data – and the most progressive and advanced marketers are using automation to reach new levels of customer connectedness.
Consider this continuum of how data can be used in marketing – and you see how access to big data analytics and automation technology can make this happen faster. (Scale is always a different issue, of course.)
- Reporting = “having the data”
- Descriptive analyses = “seeing the data”
- Analytic modeling = “knowing the data”
- Choice models and “what ifs”
- Predictive analytics = “acting on the data”
- Informed decision-making
- Actionable and real-time information engines (match offer to customer)
One caution: the risk of making bad marketing decisions based on weak analysis or poor information is exponentially higher due to the vast volume and diversity of data types. Make it part of your project goals to identify what data should be retained, what can be thrown away or ignored, and how to avoid duplication.
You may not agree that predictive modeling is dead, but it is surely evolving today. Please comment and let me know how your company is creating satisfying customer experiences using analytical models.
Fortune Cookie image on home page via Shutterstock.
The advertising industry, like many other industries, is abuzz about “AI” and “machine learning”. This post covers some of the most promising aspects of the technology, some things to watch for, and finally what that means for your media plans.
Recently, I visited my alma mater, University of Florida in Gainesville, FL, to speak with advertising students about digital marketing, analytics and how to start a career in our field.
Sandy Rubinstein is the CEO of the independently female minority-owned marketing and advertising firm DXagency. ClickZ caught up with her to find out about her role as CEO, and what advice she would give to women who want to work in the digital industry.
Effective app marketing is not about generating app page traffic, but rather about ensuring your app is discovered by targeted and relevant users who will install your app and use it regularly.