Verbatim, Issue 05: A Very Human Issue

Keeping research human with AI: the ERGO NEXT webinar replay, how Prose tested a new brand in 8 days, Genevieve's TMRE talk with Cint, BevNET Live, and October product updates.

Melissa Tovin
October 28, 2026

It feels like every conversation about research today ends up as a debate around keeping things human and integrating AI. We are eager to increase efficiency but don’t want to lose control or, worst case, produce inaccurate insights.

So how do you do it?

We are constantly evolving our approach, so I figured, this month, we could share a few answers. From the Strella perspective, from our customers, ERGO NEXT Insurance and Prose, a Research Ops leader and from the participants themselves.

Watch the replay: research beyond the researcher

Last week Jordan Brinkman, Lead UX Researcher at ERGO NEXT, sat down with Kate Towsey of The Research Ops Review to talk about reaching a whole product org with a lean research team: what goes to PMs and designers, what stays researcher-led, and where AI moderation works best.

My favorite takeaway (that I’m stealing from Priya): AI isn’t eliminating the need for researchers! It’s creating an opportunity for research to become much more embedded in how companies operate.

In their words: Prose brand and creative testing

Prose had a new skincare campaign already in motion and wanted real proof before committing. Surveys could tell them which headline scored highest, but not why.

So they interviewed 42 existing customers and people new to Prose on Strella, testing the manifesto, mood board and three headlines, integrating Strella through all steps in the process:

  • Before launch, the AI assistant flagged overlapping questions and messy wording in the guide, and the Prose team alongside the Strella team reviewed every edit.
  • During fieldwork, "ask verbatim" kept the moderator reading each headline exactly as written.
  • After, they watched interviews and read transcripts for the questions that mattered most, then used AI chat to compare customers vs. newcomers.

The verdict: people loved bold copy, as long as it came with proof. Warmth beat technical language, and specifics beat abstractions.

“Creative instincts can tell you which headline should win, but Strella told us which one actually earned trust, and why. It was a good reminder: Great copy is an art. But it's also a science.”

Jack Ball-Smith · Senior Director of Copy & Editorial Strategy, Prose

Read the full Prose case study on the Strella North Star blog.

From the field: TMRE & BevNET Live

Looking back: our fireside chat in Denver

Genevieve Becker took the stage at TMRE with Nik Blach from Cint, one of our panel partners, to tackle a question we hear all the time: will participants actually be comfortable talking with AI?

They ran a concept study with Cint participants and Strella's AI moderator, then shared what participants really said.

My favorite insight from the session?

Shorter doesn’t mean thinner. AI-moderated interviews tend to run shorter (the AI moderator talks less), and the B2B incentive math often nets out cheaper, but open-ended responses are getting longer, not shorter.

Coming soon: see you in LA this December

We'll be in person at Nosh (Dec 3-4) and BevNET Live (Dec 6-8) in Marina del Rey. If you're there, come find us.

Going to BevNET? Join our workshop:

From Question to Consumer Insight, Overnight: A Live Study for Beverage Brands

Email us at hello@strella.io and we’ll sign you up!

In case you missed it: product updates

You can always check out our changelog for the latest updates but here’s a few highlights from the last month.

Say-do gaps on tasks. When a participant says one thing but their screen shows another, Strella now flags it, with the quote and what the screen showed side by side. Every Passed/Failed verdict also comes with a short reason.

Say-do gap flagged on a usability task in Strella

Build a study from one sentence. Describe the study or upload a brief, and Strella drafts the guide and screener. Researchers can also copy a guide or screener from an earlier project.

Building a Strella study from one sentence

Tree Testing. Test whether people can find things in a proposed navigation before it's built. Works alongside card sorting.

A participant completing a tree test in Strella

Coming Nov 10: Humans of Strella

It was never AI vs. human. On Nov 10 we're officially launching something big that will make the Strella product a little more human.

Want an early look before launch day? Email us at hello@strella.io.

Talk soon,
Melissa