Audience Modeling: An Explainer

Lately, we have been talking about audience modeling at Softcrylic:

• Case study about audience modeling in the health information systems industry
• Case study about audience modeling (which we refer to in the linked document as persona indexing) in the hospitality industry
• Very exciting upcoming webinar (Q3’21) where we will walk through a real-life audience modeling process
• Webinar about audience modeling and data activation
• Webinar about visualizing audience models
• Webinar about technical approaches to audience modeling
• Suite of tools we have built internally to facilitate audience modeling
• Main site page for our audience modeling service

We have been talking about it a lot. We are very excited about the process and tools we have developed to explore audiences and have made it a core focus of our Data Science & Analytics practice. But all of this content and information may leave some wondering “What, exactly, is audience modeling?”. Good question! There are a few ways to answer it…

A brief description of audience modeling

Audience modeling is a data-driven approach for identifying distinct and high value audiences that you can leverage in marketing, both strategically and tactically. We use machine learning to do this, thereby allowing the data to speak for itself and avoiding any human bias in audience design. The resulting audience model is a set of 5-10 audiences, with each audience being characterized by a unique behavioral fingerprint. These behavioral fingerprints tell a marketer who is in each audience and how they interact with messaging. The “who” allows strategists to identify valuable audiences that align with business goals. The “how” directly informs marketing tactics. In every case that we have developed an audience model for a client, we have been able to pinpoint 2-3 audiences that map directly into key business objectives, helping our clients drive incremental conversions.

What we do and don’t mean by audience modeling

What we mean:

• Leveraging all available data to develop unique, valuable and actionable audiences
• Using machine learning to uncover the hidden patterns in customer data that distinguish unique modes of behavior and interaction with a brand
• Formulating strategies for engaging these audiences/personas
• Identifying specific tactics that will resonate best with individual audiences/personas

What we don’t mean:

• Look-alike modeling — audience modeling does not require a pre-defined segment for seeding, unlike look-alike models
• Rule-based segmentation — rule-based approaches are inherently biased and can easily overlook previously unidentified behavioral features, leading to poor quality and low value audiences
• Any process that is not 100% driven by data that you own or have access to, whether 1st or 3rd party

How we approach audience modeling

1. We host a workshop with our clients to identify their marketing objectives and important data sources.
2. Our general approach and toolkit are tailored to match our client’s unique business objectives and KPIs. (To watch us do this in action, sign up for our upcoming webinar here!)
3. We collaborate with stakeholders to define personas based on the data points that describe each audience in the model.
4. In a series of strategic conversations with stakeholders, we identify opportunities and tactics for targeting specific audiences.

You can find examples here.

We hope this helps!

Hopefully this short introductory guide has given you a better sense of what we mean by audience modeling. If you want to dig a little deeper, we have produced a lot of material on various aspects of audience modeling (linked above) and encourage you to have a look! Whether B2B or B2C, regardless of industry or marketing goals, audience modeling can be an immensely powerful tool to help guide your marketing efforts. If you’re interested in learning more, please reach out to us at info@softcrylic.com. We look forward to hearing from you!