Getting Started with Predictive Analytics Applications
No doubt you've heard about a new generation of sales and marketing applications that use machine learning to predict buyer behavior. Analysts are writing about them, marketing and sales leaders are boasting about their results, and venture capital firms are making big bets on the space. However, not all marketing and sales teams are ready to take the plunge into predictive models just yet.
Marketers need to look before they leap when it comes to getting started with predictive applications. While predictive analytics can have a huge impact on revenue, marketers should keep the following in mind when evaluating whether it's time for them to take on predictive marketing:
Think about the problem you are trying to solve
Predictive models are all built with some set of assumptions. If you don't have a particular marketing problem you're trying to solve, predictive marketing will not be of value.
Many companies have begun their predictive marketing journeys by focusing on the common problem of scoring and segmenting their leads with a predictive lead-scoring approach. Predictive lead scoring works for companies that have a high volume of leads, a large-capacity sales team, or both.
There are three common challenges relating to lead volume and sales capacity.
- The Curse of Abundance: The curse of abundance occurs when a company has a high volume of leads and a limited sales team. Predictive lead scoring can identify which leads have the highest revenue opportunities so that your sales team knows where to make at least one call.
- Forest for the Trees: Even with a large sales force, it is sometimes difficult to determine how to consistently engage your large lead database. Predictive lead scoring can help identify the higher-propensity leads so your sales team knows how to allocate more follow-up time.
- Feed the Beast: A low volume of leads for a larger sales team can cause issues that a predictive approach could solve. Specifically, it can identify new attractive prospects so you can allocate capacity to the best leads and prospects.
Predictive marketing applications can also help you focus on the hidden opportunities many marketers ignore—those within your existing customer base. If you sell multiple products, you could consider a solution that predicts which products your customers want to buy that they aren't buying already. Many refer to this as predictive cross-sell and upsell. Similarly, predictive marketing can be used to retain your existing customers. Predictive apps can help identify patterns within your customer base so you can identify which accounts are most likely to attrite—giving you what amounts to a data-driven "at risk" list. Think how much more effective you'd be at reducing churn with that knowledge.
Ensure that you have the right data assets
Organizing data can be daunting.You want to ensure you have enough data so results aren't misleading. As a quick rule of thumb, more data often leads to better results.
From an internal data perspective, you should think about incorporating the following data: marketing automation, CRM, customer service, usage, and transaction history. Many vendors in the predictive apps space will also incorporate external data from the Web, social media, third-party data providers, and public Web sites. This will help uncover such insights as hiring trends, funding announcements, technology usage, etc., which you might have missed by solely relying on a rules-based scoring approach.
Understand what success looks like
From the start, it is important to identify what you consider success. Say you define success as "lift in the top 10 percent of scored leads." This definition, along with others, will factor into the model developed within predictive applications, and help ensure you choose the right vendor to partner with. Many users of predictive marketing measure success as a lift in conversion, average deal size, and revenue growth.
Think through what you'll do with the output
Knowing what you want to do with the output is critical. You may uncover insights that ultimately change a well-established process within your organization. It may make sense to engage your counterparts in sales early on to help educate them on the new predictive marketing and sales approach.
As with any new marketing technology purchase, it's important to do your homework before making the final decision. These four best practices should set you on the right track.
Brian Kardon is chief marketing officer at Lattice Engines.
SalesPredict and Marketo Partner
SalesPredict for Marketo help marketers target more effectively and increase conversion rates using predictive analytics.
LeadiD Adds Consumer Intelligence Applications
Expanded AppMarketplace features apps for scoring leads based on particular behaviors.
HG Data and EverString Join Forces to Offer Improved Customer Analytics
Predictive algorithms will work with data sets to help with customer conversion.