Darwin hated the peacock's tail. He wrote to a friend that the sight of a feather in it made him sick. He meant it. The tail broke his theory.
Natural selection was supposed to strip away anything that did not help an animal survive. Here was a metre of iridescent luggage that made its owner slower, louder, heavier, and far easier to eat. By every engineering measure, a defect. It should not have existed.
It took a century and a biologist, Amotz Zahavi, to explain why it does. The tail is not useful despite being a burden. It is useful because it is one.
Only a strong, healthy, parasite-free peacock can haul that ridiculous thing around and still survive. A weak bird cannot fake it. The cost would kill him. So the tail is an honest advertisement precisely because it is expensive.
The peahen believes it for the same reason you believe a restaurant that is packed on a wet Tuesday. The signal is costly, and cost is the one thing that cannot be counterfeited.
Economists reached the same idea from the other side. Michael Spence won a Nobel for showing that a job candidate's degree works as a signal less because of what they learned and more because enduring it is costly in a way a weaker candidate could not sustain.
Biology and economics met in the middle on a single law: a signal is believable in exact proportion to how much it would cost a faker.
Hold onto that. Something has happened to your market that turns it into the most important filter you own.
The cost of a claim just fell to zero
For most of history, a fluent, confident, well-argued claim was itself a moderately costly signal. A sharp deck, a credible case study, a polished essay took time, skill, and judgement that was hard to counterfeit. Fluency correlated with competence. Imperfectly, but enough to lean on.
That correlation has snapped. Anyone can now generate a flawless deck, a persuasive narrative, a "we are an AI-native, data-driven, outcome-obsessed growth partner" paragraph in the time it takes to type the prompt.
The production cost of confident language has collapsed to nothing. The law above is merciless about what that means. A signal that costs nothing to produce carries no information. The slickness that once hinted at substance now hints at nothing, because the slickness is free and the substance behind it may be absent.
Most buyers have not adjusted. They still read a fluent pitch and feel the old reflex: this is impressive, they must know what they are doing. The reflex is now wrong. In a market flooded with generated content, eloquence is the cheapest signal there is, which makes it the least trustworthy.
So become a cost detector
The response is not to trust less in some vague, defensive way. It is to get specific about which signals are still expensive to fake, and to weigh only those. Here is the field guide. It works on any go-to-market provider you will ever evaluate, and I want you to use it on us.
A verifiable outcome is costly. A founder who will put you on the phone with a former client, or name a figure you can check yourself, is spending something a fabricator cannot. The lie sits one call away from collapse.
"We drove significant growth" is free. "Ring this person, here is what we changed, here is what it moved" is expensive. Weigh the second and discard the first.
Specificity is costly. Naming the exact recurring decision someone will instrument, the exact mechanism, the exact thing that could go wrong, exposes them to being wrong in public. That is a real cost a vague operator avoids.
Adjectives are cheap. "Leverage", "synergy", "AI-powered" commit to nothing. The provider willing to be concrete is paying a price the hand-waver will not.
Forgoing your own lock-in is costly. A vendor who builds you a black box only they can operate has quietly bought themselves dependency. That is the cheap, self-serving move.
A practice that insists on building you a documented, inheritable system, one you could take in-house and fire them from, is spending its own leverage to earn your trust. Almost nobody fakes that, because faking it would mean surrendering the lock-in for real.
A real portfolio is costly. A pattern library of which go-to-market decisions paid off, under which conditions, cannot be conjured by a solo operator or a week-old entrant. It can only be accumulated by running many motions over real time.
When someone shows you abstracted, cross-account learning, they are showing you something that took years of other people's experiments to build and cannot be shortcut. The cost is the proof.
Trust yourself, not the tail
This is the part I most want you to keep. Your reservation about handing thousands of dollars a month to an outside party is not a weakness to be talked out of. It is a correctly calibrated cost detector, and it is the most valuable instrument you bring to this decision. Keep it. Sharpen it. Point it at everyone.
Point it at us hardest of all. Nothing in any essay we publish, this one included, should move you a millimetre. Words are free, and you now know what free signals are worth.
What should move you is the part that would have been expensive for us to fake. The outcomes you can verify. The specifics we will commit to in writing. The system we will document and hand you. The learning we could only hold if we had done this many times before.
Do not believe the tail. Inspect it. That is not an obstacle to the conversation. It is the whole reason the conversation is worth having, because the practice built to survive that inspection is the one that wants you to run it.
Founder-led. Not founder-limited.
This piece sits alongside The Deal Was Perfectly Qualified. It Still Died. and Why We Instrument Decisions, Not Workflows. If you are weighing any provider, us included, the question to start with is not "do they sound impressive?" That signal is now free. It is: what here would have been expensive to fake?