Over this series I have argued three things. Desire is the backbone of commerce. Machines cannot want. And listening, the human kind, hearing with feeling in it, is how wants are found and woken.
Those closest to my work will likely challenge me on this. I have built an app. It asks you to upload transcripts of your market conversations. It then uses AI to process them, with the intention of discerning a unique selling proposition. If machines cannot listen, is my own method a contradiction?
It is not. Let me tell you why.
Imagine the ideal albeit impossible state. What if every single sales transaction could have a dedicated human ear beneath it? One listener per buyer, connecting deeply with each buyer and unearthing each want. This ideal state cannot exist at scale.
For this reason, broad-based marketing has always been a gamble on a message that is mostly right. One message, thousands of readers, and we accepted that there will be major misses. Even in B2B sales, where a human seller is often present, we hand the listening to the seller. And sellers often drop it. They reach for the polished pitch deck, dreamt up by a product marketer who, we hope, did some level of broad listening, so the pitch carries some resonance. Hope. Some level. Some resonance. That is the machinery we have been running the whole time.
The real problem was always simpler: too many humans to understand, too few ears to do it with. Which means machines that can hear at infinite scale is good news. Provided we stay precise about what the machines are actually doing.
Hearing is capturing and processing what real humans said. Recording the call. Transcribing it. Searching it. Coding it. Comparing it. Reading ten thousand verbatims and finding the pattern no single person could hold in their head. Machines now do this at superhuman scale, and they should.
Listening is the live human act. It is hearing with feeling in it, then deciding what the words reveal about what the person wants emotionally, and about the tangible want that would answer it. The feeling driving it often shows up in the same words. AI may suggest an interpretation. A human who has listened to real people must decide whether it is true.
Two jobs, then. And one law that keeps the machine’s job honest: AI hearing earns its place only when it processes real human voices. An AI that reads forty real discovery calls is a useful hearing aid. An avatar that generates feedback is acting for your market. A synthetic respondent can never tell you what a real market feels. More ears, always. Invented mouths, never. Or say it at its simplest, because this is the law beneath the law: AI must never be its own source. Its job is to relay the market’s message, never to contrive one.
Human listening is the scarcest resource in any revenue team. You cannot afford it everywhere. You never could. Use human listening wherever understanding what the market wants emotionally can change what gets built, positioned, or sold.
Marketing has always listened at the level of the target market, not the person. Hit and miss has always been the nature of it. What AI changes is the width of the hearing. Every sales call, every support ticket, every review, every interview transcript. Actually read. Actually connected. No human team has ever come close, and a marketer who refuses this help is choosing to hear less of their real market. Yes, AI also makes hyper-personalisation at scale real: a thousand tailored versions of the message, each sent to the person it fits. Let the machines do all of it.
But watch where the misses come from now. They no longer come from hearing too little. They come from feeling too little at the moments when emotional meaning changes the outcome. This is the heart of the Power Listening method. Consider these three vital aspects.
The first is depth. Some conversations must still be had by a human, because some things only come out in a live exchange with another person. A model can summarise three hundred calls. It cannot sit with a customer while she works out, in real time, why she really left, and catch the pause before the true answer. Breadth belongs to the machines. A floor of live, human-held conversations belongs in every research plan, forever.
The second is meaning. AI can sort five hundred verbatims into tidy themes. A human has to decide which words best reveal what the market wants emotionally. Which sentence carries the fear. Which word people use for the problem when nobody is selling to them. And feelings do not average. Say a third of your market is frightened and a third is frustrated. The answer is not a blend of the two. It is a judgement about which feeling matters in which moment. That judgement is felt, and felt judgements belong to people who can feel.
The third is the bet itself. What we say, and to whom, is a decision about where understanding meets money. The machine drafts. The marketer decides. And the marketer who decides without direct human listening and feeling behind the judgement is signing away a ton of money.
This is the real role of the product marketer in the AI era: the human inserted at the exact points where hearing with feeling changes the outcome, so that one person’s real listening reaches the whole market. Teams that organise around this will get the message right. Teams that do not are in for a long season of expensive trial and error: fluent messages, built on unfelt patterns, converting nobody.
I hope things are starting to take shape in this series. The seventh and final one in the series publishes next week. Let me know what you think.
Make it rAIn, KG