Friction Mapping in Practice: A Strategy Engagement Case Study

I'll be direct: I was skeptical about synthetic users for conversion optimization. My background is in data—I believe in experiments, not simulations. If you can't test it with real users, how do you know it works?

Then I actually tried it. And what I found was that synthetic users don't replace experimentation—they supercharge it. Here's the story.

The Problem

I was consulting for a B2B SaaS company (I'll call them "TechCorp" to protect confidentiality) with a classic funnel problem. Their free trial attracted plenty of signups, but conversion to paid was stuck at 3.2%—well below the 7% benchmark for their category.

They'd tried the obvious stuff: optimizing the signup flow, adding onboarding emails, offering discounts. Nothing moved the needle significantly.

The CEO asked me a simple question: "Why aren't they converting?" And I had to admit: we didn't know. We had plenty of data about what was happening, but no insight into why.

The Traditional Approach (And Its Limits)

My normal playbook would be:

1. Interview churned users to understand objections
2. Develop hypotheses based on interviews
3. Design A/B tests for top hypotheses
4. Run tests, analyze results, iterate

The problem: TechCorp had only 15-20 churns per month—not enough for statistically valid research. And the churned users who agreed to interviews were a self-selected group (probably more engaged than average churners).

We were stuck in a classic growth trap: not enough data to learn, not enough learning to grow.

The Simulation Experiment

Here's what we tried instead. Using SocioLogic, we created synthetic personas matching TechCorp's target customer profile: technical team leads at mid-size companies evaluating project management tools.

Then we did something I'd never done before: we showed these synthetic users TechCorp's actual free trial experience and asked them to narrate their reactions.

Within an hour, we had 50+ "interviews"—more qualitative data than we'd gathered in six months of traditional research.

What We Learned

The insights were surprisingly specific and actionable:

1. Pricing uncertainty: Multiple synthetic personas mentioned anxiety about pricing. TechCorp hid pricing until users were deep in the trial, which synthetic users described as "feeling like a trap."
2. Integration confusion: Synthetic personas wanted to connect their existing tools immediately, but the integration setup was buried three levels deep in settings.
3. Value demonstration: The trial encouraged users to explore features, but synthetic personas said they wanted to see the product working with their actual data—not a generic demo.
4. Team dynamics: Multiple personas mentioned that they couldn't evaluate the tool alone—they needed teammates involved, but there was no easy team invite flow.

The Test Program

Based on these insights, we designed four experiments:

1. Visible pricing: Show pricing on Day 1 of trial instead of hiding it
2. Upfront integrations: Make integration setup part of onboarding, not buried in settings
3. Data import wizard: Prompt users to import their real data within the first session
4. Team invitation flow: Add a prominent "invite your team" step in the first week

In a normal quarter, I might run two experiments. Because synthetic research had given us high-confidence hypotheses, we ran all four simultaneously (to different user segments) in a single month.

The Results

After 30 days:

• Visible pricing: +12% conversion (counterintuitive—we expected worse)
• Upfront integrations: +18% conversion
• Data import wizard: +8% conversion
• Team invitation flow: +15% conversion

Combined and compounded, these changes lifted conversion from 3.2% to 4.7%—a 47% improvement.

The Takeaway

Here's what I learned from this experiment:

1. Simulated buyers identify hypotheses faster: What would have taken months of interviews took hours.
2. More hypotheses mean more experiments: With high-confidence starting points, we could test more ideas simultaneously.
3. They don't replace A/B testing: We still validated everything with real user experiments. But we started with better hypotheses.

I'm still a data person. I still believe in experimentation. But I've added synthetic users to my toolkit—and I don't see myself taking them out.

If your conversion is stuck and you've run out of obvious ideas to test, synthetic users might give you the insight boost you need.

Want to discuss growth strategy? Find me on Twitter or LinkedIn.