By David M. Raab, Principal, Raab Associates
Predictive analytics has always seemed like a great tool for business marketers: a way to create better-targeted customer treatments without hours of painstaking research. Indeed, consumer marketers have used predictive models in this way for decades. The main obstacle for business marketers has been lack of data, since predictive models require thousands of observations to produce a reliable result.
{loadposition C2C13IAA}By David M. Raab, Principal, Raab Associates
Predictive analytics has always seemed like a great tool for business marketers: a way to create better-targeted customer treatments without hours of painstaking research. Indeed, consumer marketers have used predictive models in this way for decades. The main obstacle for business marketers has been lack of data, since predictive models require thousands of observations to produce a reliable result.
Now there’s a quaint objection: When was the last time you heard anyone complain about having too little data? Today’s business marketers suffer from a data flood, not a drought. So has the time finally come for B2B predictive analytics?
Vendors seem to think so. I’ve recently heard interesting news about predictive analytics from at least three companies: ReachForce, which acquired predictive capabilities when it purchased SetLogik; Mintigo, which is making a major marketing push for its data aggregation and predictive modeling tools; and Lattice Engines, which has more than tripled the number of clients for its predictive tools in the past eighteen months.
What all of these companies have in common is they combine analytics with data management. I think this is critical to success. B2B marketers don’t have the technical resources or budget to develop their own systems to gather social and behavioral data (the two sources of the data flood) and prepare it for analysis. This separates them from consumer marketers, who often do have those capabilities. As a result, consumer marketers have been able to use analytics-only tools like SAS and KXEN for their predictive modeling projects. B2B marketers need the data gathering, preparation, and modeling in a single package.
It (almost) goes without saying that the B2B predictive modeling systems also automate the predictive modeling itself. Otherwise, they’d require users with advanced statistical skills: a breed even harder to find in B2B marketing departments than database managers. The trade-off is that the systems can only build certain types of models, so there’s some loss of flexibility. But since the alternative for most businesses is no predictive modeling at all, that’s a pretty weak criticism.
The final obstacle that these systems must remove is deployment of the results. Again, the problem is that technical skills are needed to move scores from the modeling system into the execution system, usually marketing automation or CRM. At a minimum, the vendors build connectors to automatically push the scores into the marketing database. It would be even simpler, from the marketer’s viewpoint, for the modeling system to actually tell the execution system what to do: such as, issue an alert to sales, send a specified message, or add selected names to a campaign. These vendors haven’t necessarily achieved that degree of integration, but they’re working on it.
I doubt you’re worried that predictive modeling will make the marketer’s job too easy, but, if you are, you can relax. Predictive modeling systems can only answer the questions they’re asked, which in the marketing case usually means they can predict who will respond to a given promotion or buy a particular product. It’s still up to marketers to create compelling offers and deliver them in well-structured campaigns. In most cases, the marketers will also need to run the programs against a sample audience to accumulate data for the modeling system to make accurate predictions.
Marketers also need to look carefully at system outputs, both to ensure things are running correctly (remember: to err is human, but it takes a computer to really mess things up) and to extract useful insights.
In sum, the current crop of vendors may finally have put predictive analytics within reach for most marketers. With so many of us drowning in data, their helping hand comes just in time.
David Raab is Principal of Raab Associates and a regular contributor to Demand Gen Report.
