Everything in moderation. Including moderation.

Anthropic announces they’ve conducted the largest qualitative research project in history with their chatbot. Qualitative institutes fire back that we are talking about two different things. Harvard Business Review then announces the age of AI qualitative is upon us. LinkedIn goes bananas. What’s a researcher to believe?

Let’s say you are a savant at designing custom market research. And let’s say you need to figure out which concept from the many you were handed is the customer’s favorite and why, but you don’t have a handle yet on all the possible reasons why – you want the customers to tell you that – so a traditional survey isn’t the best fit because you don’t know the response choices to write. You need qual, or something like it. You need to know this across multiple markets with different languages, but the turnaround time is fast since the decision date is right around the corner. Plus, your deciders are sticklers for robust base sizes, yet as luck would have it, penny pinchers on budget. You’re hearing whispers like “early and wrong is better than late and right” and “80% is good enough here.”

Would you design human-moderated qualitative or use an AI qual tool instead? AI qual. The insight objective is simple enough that AI qual’s branched logic and follow-up probing can do the job sufficiently. Maybe toss in a week’s worth of 5-20 native-language human interviews in parallel to drill down on the crux insights, and you’re in great shape.

Now imagine those same deciders come with a deeper question where the nature of the exploration is nuanced, and the style of insight desired is especially strategic in order to provide them with a true decision advantage. Instead of sticklers for N, they are suckers for why – why this, why that, why yes, why not. The crux of the ask, and your path to greatness, is a leading-edge insight that creates an elite competitive advantage that almost surely will require seeing around corners and stakeholder alignment on the fly based on experience, intuition, empathy, teamwork, brainstorming, reflection, and iteration. Sound familiar? It should. That’s often the ask from leaders who know what they are doing and demand excellence.

That kind of insight is tough to wring out of an AI qual tool, and it’s even tougher to deliver in a couple of weeks on the cheap. Not impossible, but exceedingly unlikely. Those instances still need human moderators to hit the mark.

What you have heard is true. AI conversational tools are blurring the line between qualitative and quantitative by enabling speedy scale, converging the distinction between qual and quant to a finer point. Sometimes speed does matter most, and sometimes maximizing the number of impressions is required to generate the necessary rigor or segments to cut. Extending and amplifying human moderating with AI qual in those times is an innovation. When the particulars of the situation call for it, we are putting AI qual to work.

But let’s not call it “moderating” quite yet. We prefer to think of a moderator as personifying the kind of emotionally intelligent and bespoke conversations that sharp and nimble practitioners are known to cultivate. There is a ton of VC money going into this tool, and perhaps there is a future where AI conversational surveys are as dynamic as expert moderators applying a proven craft, and if so, we’ll be on it. Partly, it will depend on respondents’ eagerness to engage with AI qual in such a subtle way for a long enough session to reveal the kind of deep and triangulated insights that emerge when qual is at its best. That is not quite what these tools are today, so much as relatively brief and simple Q&A. Sometimes that does the trick, oftentimes it can’t.

It is understandably frustrating in the age of elevated speed and cost pressures, when leadership has a general notion that insights ought to be AI-able, that the magic of masterful human moderating is usually still essential to produce the kinds of sharp insight that moves markets by deeply knowing the customer. Chasing fast and cheap has its tradeoffs; pretending otherwise is risky. Our duty to clients is to think through all design options, including the very best AI tools of the moment, and present the fittest design for the unique context and critical needs of each qualitative project. More often than not, today, that is still an elite Linker moderator.

Moderation is an art as it ever was. Socrates had a pretty good bead on it, and trust is still a biological, not a technological phenomenon, at least for now, as TLG sees it.