AI Moderation Is a New Research Methodology. Let's Treat It Like One.

The trade-off every researcher knows by heart

For as long as insights teams have existed, they've faced the same forced choice. Run interviews and get depth but accept that you'll only hear from a dozen people. Or run a survey and get scale, but accept that every answer is final: no follow-ups, no clarification, no ability to hear a participant ramble, no "tell me more about that."

Every research method has its trade-offs and, in the past, every research roadmap was shaped by these tradeoffs.

What AI moderation actually is

AI moderation doesn't sit on the spectrum between surveys and interviews. It breaks the axis.

In an AI-moderated study, every participant has a real conversation. The moderator asks a question, listens to the answer, and probes — in the moment, the way a skilled interviewer would. When a participant says the onboarding felt overwhelming, the next question isn't the next item on a script. It's "what was the first moment it felt that way?"

And because the moderator is software, those conversations run in parallel. Three hundred interviews can happen in a weekend, across time zones and languages, without a scheduling spreadsheet.

Depth at the scale of a survey isn't a faster version of either method. It's a different method.

Why AI moderation isn't a survey

A survey is a fixed instrument

The questions you wrote in your survey are the questions you get answered, and the question you forgot stays unasked. One of the defining feature of AI moderation is that the instrument adapts.

When you launch a study, you don’t always know how the participant will respond and what direction the response will take. The follow-up is built into every session, and dynamically adjusts to capture this. After each session you are able to review the results and make adjustments based on how the AI moderator has responded. Your question set is no longer fixed.

Strella’s format is conversational

Some assume when they see AI moderation that it is just a fancy video survey. Strella’s AI moderator is different and creates a dynamic conversation with the participant, who will open up and feel comfortable like they’re on a zoom call. With video surveys a participant clicks to record themselves, presses done and then waits for the next question to appear on the screen. It is a task.

With a Strella AI moderated study, the participant is in a dynamic conversation with the moderator for 45 minutes or more.

Why AI moderation isn't replacing researchers

The parts of research that were never the bottleneck are exactly the parts AI moderation doesn't touch: deciding which question matters to the business, moderating studies where relationship building or sensitivity is essential, designing a guide that gets past rehearsed answers, noticing the finding that should change the roadmap.

What it removes is the constraint that kept that judgment trapped in small studies: the hours of scheduling, moderating, and transcribing that capped most teams at eight to twelve sessions. Researcher judgment doesn't disappear. It moves to where it is needed most upstream, into design and synthesis, into studies where human judgement is needed, where it compounds.

New methods deserve new standards

When a method is genuinely new, borrowed standards fail it. Judging AI moderation by survey standards undervalues the conversations; judging it by human-moderation standards misses the benefits that taking the judgement of another person out of the conversation creates.

Researchers should hold AI moderation to high standards. We'd just argue they should be its own: Did the probing go deep enough? Did the scale surface patterns a dozen interviews would have missed? Did the researcher stay in control of what was asked and why?

Where this fits in your toolkit

AI moderation won't replace In Depth Interviews (IDIs), and it shouldn't replace all of your surveys. It earns its place on a third shelf: when you need the why at a sample size that used to mean settling for the what. Concept tests with real probing. Churn studies where every lost customer gets interviewed. Discovery work across five markets in one week.

The teams getting the most from Strella’s AI moderation aren't the ones replacing an old method. They're the ones asking a question they previously had no way to answer.

‍