Segmentation doesn’t have to be a dirty word

Editor’s note: Michelle Castle is chief innovation officer, and Casey Willard is chief analytics offer, Alter Agents, a full-service marketing research consultancy.  

Segmentation research is a powerful tool. It provides a deep understanding of and empathy for real people, identifies those with the greatest potential value and uncovers opportunities to shape the business. It guides strategy and tactics for targeting, messaging, media planning and product and feature development. It engages cross-functional teams and inspires them to action. 

Despite the importance of segmentation research in identifying aggregate unmet needs and building target strategies, researchers often hide behind less loaded terms such as “typologies” or “personas” instead of “segments” when discussing the approach. They aim to disassociate from past negative segmentation experiences and convince themselves that this time will be different.

Why does segmentation research have a bad reputation?  

Segmentation research without a clear purpose, a thoughtful design and stakeholders committed to the process fails to inform decision-making. Too often, segmentation research becomes either oversimplified or unnecessarily complex – a major investment of resources that yields cleverly alliterated names and not much else. Despite a great deal of time and effort by all those involved, the results are not truly relevant nor useful to the strategists, marketers and developers who need to put it to work. While it’s easy to blame the methodology or mechanics of segmentation research for its failure, the fault typically falls to the process itself. 

But “segmentation” doesn’t have to be a dirty word.

In our collective experience of conducting over 100 segmentation studies over the past 20 years, we’ve identified some key principles that have been consistently applied across the most successful studies with the longest-lasting impact. Common challenges surrounding segmentation research – and potential solutions – include:

  • Lack of focus and alignment: Segmentation studies that serve ambiguous or multiple agendas become overly complicated and ineffective. As with any research endeavor, clearly defining goals, scope and expectations up front assists in achieving alignment across all stakeholders (marketing team, product team, agency, insights, leadership, etc.). Stay centered on priority objectives, while also being open to redefine and realign on how best to achieve desired outcomes. Go into the project with an eye not only toward immediate needs, but also long-term utilization and impact (e.g., behavioral data appends, global segmentation implications, typing tool development).
  • Weak ingoing hypotheses: Research designed without fully fleshed-out hypotheses can lead to meandering exploration without business outcomes to anchor the approach. Pursue a flexible design balanced between the validation of ingoing hypotheses and the exploration of new discoveries for potential integration. This sets up a road map for the research and gives it a more intentional tone. Generate strong hypotheses from multiple sources that represent organizational, consumer and expert perspectives.
  • Overgeneralized design inputs: Segmentation projects are frequently handcuffed to the wrong set of inputs that divert from core business issues. This results in segments that are not meaningfully differentiated, and thus difficult to explain and act upon. Employ comprehensive, yet thoughtful and restrained, design-side inputs when exploring potential segmentation solutions. Include a mix of attitudes, needs, occasions and behaviors within a category-specific context to ensure relevancy. For example, someone may not consider themselves to be a “risk taker” in life in general, but they are adventurous when it comes to trying new, esoteric food.
  • Irrelevant metrics: Too often, cluster analyses yield segments that seem intuitive on the surface, but ultimately lack the insight needed for decision-making. Anchor segmentation solutions in pre-identified metrics that are directly tied to desired behaviors and business outcomes. Employ trade-off versus traditional agreement scales to mitigate response bias and force greater differentiation. Evaluate a range of input combinations and clustering methods against specific criteria to create segments that are distinct, recognizable and easily prioritized based on their potential to grow the business.
  • Disengaged stakeholders: One of the greatest dangers to effective segmentation research is a lack of commitment from key stakeholders. There is a misconception that all it takes is the push of a button to yield the optimal solution rather than the time to engage and align at each stage in the process. The most successful outcomes result from a highly iterative process with sleeves-up immersion by all key stakeholders to select the solution that feels most right in the gut and most meaningful to the business.
  • Absence of an advocate: No matter how pretty or practical the results may be, segmentations are useless without acceptance and adoption. An internal advocate is critical to shepherd the research from inception to execution to embedment within the organization. Of course, the research team must support this champion with a clear, compelling narrative and plan for creative socialization that promotes understanding and inspires action across cross-functional teams.

Segmentation research must be rigorously conducted and statistically sound to be useful. When executed through a highly interactive process involving deeply engaged stakeholders, it is transformative.