Somewhere in your company there’s a slide deck with four personas on it. Each has a stock photo, a name, a bulleted list of pain points, and a messaging recommendation. The deck was built in a workshop. It represents what your team believed about your buyers at the time they made it.
When’s the last time anyone updated it?
Your email program sends thousands of messages a month to real contacts who demonstrate, through clicking and not clicking, exactly what resonates. That data is more specific and more current than anything produced in a conference room. And it costs nothing to collect because you’re already collecting it.
Why the Workshop Deck Drifts
Three biases compound over time. Sales anecdotes overweight the memorable — the huge win, the ugly loss — and underweight the boring middle where most deals actually live. Customer interviews capture what buyers say they care about, which correlates loosely with what drives their behavior. Market research describes the category. Your audience? Not so much.
Then nothing updates. The product evolves. The market shifts. New buyer profiles emerge. But refreshing the persona deck means scheduling another workshop — and the last one took three weeks of calendar negotiation and a full day locked in a room with Post-it notes. Nobody’s volunteering for that again.
The Method (Manual Version)
Pull your top-performing emails.
Export twelve months of campaign data. Identify the top performing quartile by click-through rate or conversion rate — whatever reflects real engagement in your business.
For each top performer, tag the content attributes: tone (formal, conversational, technical), structure (long-form, scannable, single-CTA), hook (pain point, aspiration, social proof, urgency), depth (surface, moderate, deep technical), and topic.
This takes a few hours of manual work. No shortcut. You’re building the dataset everything else depends on.
Find the clusters.
Cross-reference content attributes with contact-level engagement. You’re looking for groups of contacts who consistently respond to similar content types.
This is the hard part. Not a “pull a report” operation. You need click-level data joined to content tags joined to contact records. In most MAPs, that means exports, SQL or pivot tables, and data cleaning. If you have an analyst, this is their project. If you don’t, budget a week and expect the first pass to be rough.
What you’re looking for: clusters of contacts with consistent content preferences across multiple campaigns. One group engages with deep technical content. Another with concise business-case framing. Another with social proof and peer comparisons.
What if the clusters don’t emerge? Sometimes they won’t — especially if your content mix is narrow. If you’ve been sending similar content to everyone, there isn’t enough signal variation for clusters to appear. That’s a finding in itself: diversify your content before you can discover your segments.
Overlay firmographics.
Once you have behavioral clusters, map them against job titles, industries, and company sizes. Sometimes the mapping is clean. Often it isn’t, and the surprises are the point.
Your “IT Decision Maker” persona might split into two groups with the same title but opposite content preferences. Your messaging has been averaging across them, which means it’s been optimal for neither.
Or contacts from completely different industries might cluster together because they share a behavioral pattern that cuts across verticals — segments your persona workshop never identified because they don’t map to any obvious demographic category.
Or Skip the Manual Work Entirely
Everything above — the content tagging, the cluster analysis, the behavioral profiling — is what we built Persona Reporting to do. NLP and LLMs analyze the content of every email you send — type, topics, tone, structure — and connect that metadata to the persona’s engagement behavior over time.
A week of analyst time produces a snapshot. Content Profiling runs continuously. Your personas update as behavior changes — which it does, faster than most teams realize.
The Persona Report lets you isolate specific groups and track their content response patterns. Persona Voices take it further — generating content guidelines for each segment based on what’s actually performing. Not what you think should work. What does work, measured from your data, updated as your audience shifts.
Close the Loop Between Data & Action
Persona decks are stuck in the cloud and only work if content specialists refer to them. Close the loop by using Persona Voices in Motiva Generator. The persona insights automatically gathered can be directly injected into your content generation process and adapt any content you input to be more engaging to your audience.
The ultimately promise of AI is automated, hyper personalization where your tech understands your audience and adapts to them to achieve better business results. Persona Voices combined with Motiva Generator gets you one step closer to making full use of your data to personalize your content in ways that move the needle.
The Limitation
This method — manual or automated — discovers what your audience responds to within the content you’ve already sent. It can’t tell you about preferences you’ve never tested. If you’ve never sent a deeply technical email, you won’t discover a cluster that prefers deep technical content.
Data-derived personas are more accurate than workshop guesses, but they’re bounded by your content history. Use them for immediate targeting while expanding your content variety to discover segments you might be missing.
Learn more about Persona Voice and Persona Reporting as well as the future of segmentation.
