toy airplane model being created
PHOTO: superelaks | adobe stock

The B2B customer is a tough person to market to. They prefer to be left alone to research online, and they are picky, looking for a solution that fits complex needs, perhaps for a number of stakeholders. The days of sending a big catalogue or dropping customers into a generic home-page are over. Like B2C customers, a more personalized approach works better.

After years spent testing digital channels — from webinars to LinkedIn, email and programmatic advertising — B2B marketing has gotten complex enough to require real analytics-driven decision making. In fact, 89% percent of B2B sales and marketing executives believe that data quality is what drives the best campaigns. Knowing as much as possible about prospects, their companies and the channels and products that matter to them is essential for a successful approach.

One of the key elements of a smart omnichannel B2B marketing plan is to have data-driven models that can help with budget allocation across channels, scoring and spending on specific ABM targets. But, creating a data-driven marketing program complete with forecasting, predictive models and algorithms isn’t a one-and-done project. It requires committed resources that go beyond the scope of the B2B marketing team, or a single data analyst. If you’re working on developing your own data-driven marketing strategy, consider the following must-haves so that you scope it right.

Data Requires Long-Term Care

Accumulating data into a valuable database that can drive marketing decisions is not a single step process, but rather an ongoing commitment. If you plant a garden, you don’t just throw some seeds in the ground. You have to water, weed and prune in order to get a good harvest. And come next season you have to do it all over again, with lessons learned. Data requires the same constant attention. B2B marketers need to be realistic about having resources that are dedicated to this process. 

IBM engineer George Fuechsel is credited with the data science concept of “garbage in, garbage out.” In other words, bad data equals bad insights, and bad outcomes. For any model to work effectively, the data that it uses needs to be good quality, robust data. Good data is up-to-date, accurate and multi-faceted, and that requires a resource that’s checking on how data is performing, and works to make updates when it gets old or inaccurate. For B2B marketers, not only should data collection and data cleansing be a regular process, but new data should be constantly evaluated to fill in holes with a variety of unique insights from intent data to search data. More data is only more if it helps create a fuller picture of top targets. That means an ongoing process of evaluation and testing against the current suite of customer and prospect models to see if new data is worth the added expense and complexity.

Related Article: Data Ingestion Best Practices

Models Have a Way of Multiplying

A B2B marketing model is never just one model. Consider a global industrial supply B2B brand, using a single high-level propensity to buy model to drive budget allocation decisions. After validating the model on various sub-universes and outcomes, it was clear that different facets of their business could benefit from different models. For example, they promote specific products (e.g., First Aid) across a variety of countries with varying buyer personas and varying purchase timelines across different channels. They also see different behaviors from first time vs. long term buyers. And, their different brands and products have nuances that warrant testing of different models, too. Add that all up, and we’re talking about 80 unique models that need to be created and tested.

From there, these models need dedicated monitoring and maintenance as we often see that buyer behavior can shift over time. New channels become more important to the buyer journey, or new products become more enticing. It is critical to create a process whereby model scores can be efficiently matched to outcomes and model drift can be monitored with the necessary corrective action taken. The data science team can’t assume that models built six months ago will continue to accurately predict future behavior.

Related Article: Stop Torturing Your Data and Other Tips to Reveal True Data Insights

Planning for Innovation

The worst-case scenario is to be overwhelmed just as you’re getting started with data-driven marketing models. Instead, it’s important to build in a good 20% bandwidth for testing and innovation. For example, if a B2B marketer wants to try CTV, creating a plan to test new data, and see how CTV as a channel interacts with the rest of the channel mix shouldn’t mean a 6 month wait for a developer or data scientist to finally have time for it, but rather, it should be part of the normal business operation to expand and keep testing.

That also means that marketers can be empowered to ask smart questions and test long term assumptions. For example, one company had always started remarketing to recently acquired customers after 90 days. But, after additional analysis, they realized that a certain customer profile was much more likely to be a repeat purchaser, and deserved a higher percentage of that budget. These smart tests can continually improve ROI for B2B marketers with the resources to run them. 

Go Big or Grow Big

For the B2B marketer that is ready to make good on data-driven strategies, it’s worth it to consider the entire picture, from acquiring quality data and maintaining it, to building specialized models and innovating for growth. Otherwise, a big project can become a big headache that never seems to live up to its promise.

But even if your company has minor resources to start with, a good data-driven strategy is completely within your grasp, simply start with the right footprint to include all of the best practices. For example, start small and focus only on top tier contacts, get the best data for that small audience, and test one or two models to start. With each new campaign, and each test, the investment in data and modeling should prove itself with higher return for every marketing dollar spent. From there, the program can be incrementally increased as the investment justifies.