Dark channels are real. The panic around them isn't.

There is a new thing marketing has decided to be worried about, and it has a suitably ominous name. Dark channels. Dark social. The places buyers go to make up their minds where your analytics cannot follow: the private Slack channel, the WhatsApp group, the podcast, the peer they trust, and now the AI assistant that answers the question before a human ever lands on your site.

Let me concede the real part first, because it is real. Most B2B buying really does happen out of view, in channels that feed the funnel without ever showing up in it. The research everyone cites is sound: at any given moment the large majority of your market is not in the market at all, and the ones who are do most of their deciding before they raise a hand. Google called the churn of it the messy middle. It exists. I am not going to pretend otherwise.

What I want to question is what marketing has decided to do about it. Because the response has followed a very familiar pattern, and it is not the response the evidence actually points to.

The naming, then the gold rush

The move goes like this. First you name the thing. "Dark social" was coined back in 2012, and it sat quietly for years. Then, the moment it became fashionable, it acquired a shadow: a wave of tools promising to illuminate it. Attribution platforms to un-dark the dark. Dashboards for the undashboarded. And now, arriving on cue, the newest members of the family, AEO and GEO, answer-engine and generative-engine optimisation, promising to help you win the box that ChatGPT or Gemini prints instead of a search result.

I want to be careful here, because some of this work is legitimate. Being present and quotable where AI models source their answers is a reasonable thing to attend to. But watch the shape of the conversation. Within a season, "buyers decide where you can't see" has become "buy this to see it," and a whole category has a new line item, a new set of conference talks, and a new acronym to put on a slide. The problem became a product, exactly as it did with the nine-stage funnel and the 47 versions of the four Ps before it.

The tell is always the same: the diagnosis is about the buyer, and the cure is a purchase.

You cannot game what you cannot see

Here is the awkward logic the gold rush skips. The entire premise of dark channels is that they are private, unattributable, human. A recommendation in a Slack group is worth something precisely because nobody bought it. The instant you find a reliable way to manufacture and measure your presence there, you have changed the thing into ordinary, gameable, discountable media, and buyers move on to the next place you are not.

Which means the honest answer to "buyers decide where I can't see" is not a cleverer way to see. It is to be the name they already trust when they get there. That is not a dark-channel tactic. It is the oldest idea in the discipline: a brand strong enough that when the private conversation happens, you are what gets recommended. Binet and Field have spent years showing that the demand built at the top is what the bottom converts. Dark channels do not overturn that. They are the most vivid argument for it anyone has produced in years.

If a channel can be gamed, it stops being trusted. If it can't be gamed, the only lever you have left is being worth recommending.

Follow the money, not the darkness

And this is where it stops being a marketing-fashion argument and becomes a commercial one, which is the only argument that travels to the people who hold the budget. Every hour and pound spent trying to instrument the uninstrumentable is an hour and a pound not spent on the two things that actually move the numbers a CFO reads: lowering what it costs to win a customer, and raising what that customer is worth.

A weak brand shows up in exactly this territory as a tax. If buyers do their private deciding and your name is not the one that surfaces, every deal that finally reaches you arrives colder and more expensive to close. That premium is already sitting inside your acquisition cost. It does not need a dark-channel dashboard to find it. It needs someone to measure whether acquisition is creating or destroying value at all, and then to invest where the maths says the leak is.

So by all means take dark channels seriously. They are telling you something true. Just notice that they are telling you to build a brand and to watch your unit economics, the two least fashionable, most durable things in marketing, and not to buy the torch someone is selling to chase the shadows around.

Dark channels didn't break measurement. They just reminded us that the things worth measuring were never the ones with the newest dashboard.