Destructive Testing in B2B (and Why LLMs Change It)

B2C brands can do destructive testing. If they destroy a tiny part of their market, so be it. B2B companies generally can't, because they just have fewer customers. The main challenge for a B2C company is too many competitors. The main challenge for a B2B company is too few potential customers.

There's now a solution, and it's really new. Large language models can be very good at predicting audience or prospect responses, even for relatively small customer sets. You can use various companies to test your messaging against simulated AI audiences: simulated prospects and simulated types of prospects, or ICPs. You get a pretty good sense of certainty about how the recipient will react.

You can start asking them as if they were personas in a focus group, or in a conference room. What did you think of this deck? Would this deck get you a meeting? Would this ad get you interested? Does this headline resonate with what you're worried about, even things you don't say aloud? It's really good at that.

So B2C could always do destructive testing, and B2B couldn't. B2B had the more acute need for simulated audiences, but both got the ability at the same time, in the last two or three years.