Look-Alike Targeting Can Be Deceiving
Customer segmentation is the bread-and-butter of B2B marketing. Yet marketers are often at a serious disadvantage when performing segmentation because they rely on inaccurate or incomplete data.
In fact, Forrester found that the top three data sources that marketers use for customer segmentation are manual inputs from sales reps, static list purchases, and manual inputs from contact forms.
With the dawn of predictive analytics, there is a better way. Predictive can break down customers and prospects into segments based on very granular attributes such as technology stacks, recent news and events, social and web activity, and buying intent that can be used for creating personalized messages and buying experiences.
Marketers setting out to use predictive for segmentation will encounter a variety of different approaches that may appear to accomplish the same outcome, but there is a key distinction that divides segmentation features into two categories: those that produce lookalike audiences and those that generate predictive segments.
What’s Lacking with Lookalikes
The term “lookalike audience” became prominent when Facebook made it the cornerstone of its social advertising platform, but the concept is the same regardless of what platform is used to create the lookalike audience:
- A marketer produces a list of existing prospects or customers that represent a sample of their target audience
- After the marketer uploads the list, the marketing platform (social network, data vendor, predictive vendor, etc.) finds those individuals or companies among its database and analyzes the data it has about them to identify shared attributes
- The marketing platform locates other individuals or companies within its database who possess attributes that are similar to those of the prospects and customers that the marketer uploaded
This approach is very useful for campaigns where you know what your message or offer will be and the audience you want to target with that content.
B2B marketers don’t always execute campaigns this way, however. Especially with the growing popularity of account-based marketing, many marketers want to identify a high propensity audience and then craft messages and offers that will appeal to that group.
Unfortunately, lookalike audiences are not sufficient for these scenarios because marketers may not be able to generate a list of existing customers and prospects that can serve as the basis for an optimal audience.
For example, let’s say a marketer wants to build an audience that shares attributes with the customers who have bought from her company previously. She may select 50 closed-won accounts to serve as the basis of a lookalike audience, but these might not be a representative sample of the company’s historical customer base — maybe they’re just the most recent 50 customers. As a result, she may end up with an audience comprised of prospects who look similar to 50 closed-won accounts but aren’t actually a good fit for the company’s offering.
Marketers attempting to run audience-first or ABM campaigns instead need a way to build segments with shared attributes that are predictive of success based on sales and marketing outcomes recorded throughout their CRM and marketing automation (MAT) systems.
Picking the Best Audience with Predictive Segmentation
This is where predictive segmentation can make a difference. Rather than building an audience based on a defined set of customers or prospects, a predictive segment is created using data from your CRM and marketing automation platform that indicates what types of accounts are most likely to be successful.
At Radius, we call these Recommended Segments. Radius generates a set of ten Recommended Segments every week and ranks those segments according to their predicted likelihood of success so that marketers always have the most up-to-date insights when they go to launch a campaign. In addition, Radius surfaces the signals and attributes characterizing each of the segments so that marketers (and their sales teams) can personalize campaigns and outreach.
Predictive (or recommended) segments offer an advantage to marketers because they are high propensity groups of accounts created using combinations of attributes and signals that marketers may never have anticipated.
For example, a predictive (or recommended) segment might include companies that use cloud-based data storage and received funding or financing within the past 60 days. This segment would be generated by analyzing the historical sales and marketing outcomes in a CRM and combining that information with external data about prospects’ technology stacks and recent news.
This may turn out to be the optimal segment for inclusion in a new outbound campaign or a target account list, but it would be very difficult for a marketer to build this segment using a lookalike audience. Marketers would need to know which of their customers and prospects use specific technologies and recently received funding, and they would also need to be able to determine that this is the group of accounts that converts most often among all possible segments within their CRM.
Wrapping It Up
Understanding the difference between lookalike audiences and predictive segments is just one requirement for marketers seeking to improve their campaign strategy and execution using segmentation. Marketers also need to know how they can personalize content for their segments and run campaigns that are targeted at the right groups.
To learn more about the impact of segmentation, best practices, and how to create targeted segments for your marketing needs, check out our predictive playbook by clicking the banner below: