A lead magnet is not a coupon, a content upgrade, or a gated PDF you traded for an email. It is a free sample of your judgment — and it works or fails on whether the buyer, having consumed it, thinks: if the free thing is this good and this smart, the paid thing must be worth it. Most lead magnets never provoke that thought, because they are built to capture an address rather than to prove a mind. They are either too thin to earn any trust, or they solve a problem that sits next to your paid offer instead of directly before it. Both failures come from misunderstanding what the buyer is actually sampling.
What Hormozi got right, stated plainly
The sharpest working criteria for a lead magnet come from Alex Hormozi's $100M Leads, and they're worth crediting precisely because they're better than the guru genre they sit in. His test is three parts: a good lead magnet solves one narrow problem completely, it is fast to consume, and it bridges naturally to the paid offer — ideally by solving the problem in a way that reveals a next problem your paid thing is built to handle. He is explicit that a lead magnet which only partially solves the problem is worse than none, because a frustrated non-result trains the buyer to expect frustration from you.
That last point is the one most people skip, and it's the hinge of everything below. "Solve it completely" is not a nicety about being generous. It is the entire mechanism. A lead magnet that half-solves the problem doesn't build half the trust — it builds negative trust, because it demonstrates that your idea of "complete" stops exactly where the invoice starts.
Where the popular version stops is at the what. It tells you to make the thing valuable and fast. It doesn't explain why value at the top of the funnel converts into willingness to pay at the bottom, which means it can't tell you which of two equally valuable magnets will actually pre-sell and which will just harvest tire-kickers. That mechanism is where the real design decisions live.
The mechanism: a costly signal wrapped around a demonstration of taste
Two things happen when a buyer consumes a genuinely good lead magnet, and they are different in kind.
The first is a costly-signal inference. Giving away something that is genuinely valuable — that visibly took real work, real expertise, real specificity — signals there is much more where this came from, and it signals it in a way a low-value provider structurally cannot copy. This is the same logic I've argued runs underneath all trust-scarce marketing: the signals that move a skeptical buyer are the ones a low-quality competitor cannot cheaply fake. A thin operator can't afford to give away a complete solution, because a complete solution is most of what they have. When you hand over something that would be expensive for a bad provider to produce, the buyer reads the expense correctly. The free artifact works like a stott — the gazelle's showy leap in front of a predator that says I have energy to burn, so don't bother chasing me. It tells the buyer you can afford to give this away, which means what you're holding back must be considerable.
The second thing is more specific to why anyone buys expertise at all. As raw competence gets commoditized — as models and tooling let anyone execute what used to require a specialist — what a buyer is actually paying for collapses down to judgment: your taste about what to do. Not the execution, which is increasingly cheap, but the decision about which execution is correct here. A lead magnet is the highest-leverage place to demonstrate that judgment before money changes hands, because it lets the buyer watch you decide on a real instance of their problem. They don't have to take your word that you have taste. They see it operate.
Put the two together and you get the actual design spec. The lead magnet has to be a costly signal (valuable enough that a weak provider couldn't afford to give it away) and a demonstration of the exact judgment your paid offer sells (so that the value they just received is a direct sample of the value they'd be buying). Value alone gets you the first. Value that showcases your specific way of thinking gets you both — and only both pre-sells the paid thing.
This is why a discount fails as a lead magnet and a real tool succeeds. A discount transfers money; it demonstrates nothing about your judgment, so it selects for people who wanted a discount. A tool that embodies how you think selects for people who liked how you think. One taxes you at the door. The other qualifies the buyer for free.
The two failure modes, exactly
Almost every dead lead magnet dies of one of two specific causes. Naming them precisely is what makes them fixable.
Failure one: too thin to signal anything. The "ultimate guide" that is nine hundred words of throat-clearing and a link to book a call. The "free audit" that is a form feeding a sales script. These fail the costly-signal test outright — a bad provider could produce them just as cheaply, so they carry no information, and the buyer's skepticism is the correct Bayesian response, not cynicism. Worse, thin magnets do more than convert poorly. They build negative trust. A guide that pretends to solve a problem and instead pitches has demonstrated your judgment — the judgment that this thin thing was good enough to send you. The buyer now has evidence, not a hunch, that your idea of value is a bait. You've pre-sold the inverse of what you wanted.
Failure two: the wrong bridge. This one is subtler and kills better-intentioned magnets. The lead magnet is genuinely valuable and genuinely smart — but it solves a problem that is adjacent to your paid offer rather than on the path to it. It sits beside the road instead of one step before the toll. So it attracts people with the adjacent problem, delights them, and converts almost none of them, because solving the adjacent problem doesn't create the next-problem that your paid thing answers. A gym that gives away a brilliant meal-prep guide attracts people who wanted a meal-prep guide, not people about to hire a trainer. The magnet was good. The vector was wrong. Value with the wrong bridge is worse than mediocre value with the right one, because it costs you real work to attract exactly the wrong people.
The diagnostic that separates them: a thin magnet fails the question would this be expensive for a bad provider to make? A wrong-bridge magnet passes that but fails does solving this problem put the buyer one step from needing my paid offer? You have to pass both.
What this looks like when I build one
For Velya — where we build AI agents that qualify inbound leads for clinics — the tempting lead magnet is a broad one: "How to get more patients." It's a topic, it's thin by necessity (you cannot solve "get more patients" in a PDF), and it bridges nowhere in particular. Textbook double failure.
The version that actually works is narrow and load-bearing: a qualification script — the exact sequence of questions that sorts a serious cosmetic-surgery inquiry from a price-shopper before anyone burns a consult slot on them. That completely solves one real problem the clinic has the day before they'd want an agent to run it at scale. Critically, it demonstrates the judgment we're actually selling — not "we can build a chatbot," which is now commodity, but we understand which signals in a first message predict a booked, high-value patient and which predict a no-show. The clinic uses the script by hand, sees it work, and arrives at the paid conversation already believing the thing that matters: that we have taste about their specific funnel. The agent is just that judgment, automated and always on. We didn't argue for the paid offer. The free thing did.
The Kommerce version follows the same shape and shows why trust-scarcity raises the stakes. In cash-on-delivery markets, a merchant's single largest cash leak is the refused delivery — the buyer who ordered on impulse and rejects the box, leaving the merchant eating reverse-shipping on both legs. A thin magnet would be "5 tips to grow your store." The real one is a confirmation-message template plus a simple rule for which orders to call before dispatching — a complete solution to the refusal problem that a merchant can run tomorrow and measure by Friday. It works standalone. It also demonstrates precisely the judgment Kommerce sells: that we understand which orders are risky and why, in a market where that judgment is the difference between profit and bleeding cash. With no chargeback protection and no small-claims backstop, a merchant cannot afford to trust a provider on adjectives. The free thing has to carry the whole signal. That constraint is clarifying — it forces the magnet to be genuinely good, because nothing else is holding it up.
The build
Two tests, and a magnet has to pass both before you ship it.
First, does it completely solve one narrow problem the buyer has right before they'd need your paid offer? Narrow enough to actually finish inside the format; positioned so that solving it surfaces the next problem your paid thing answers. If you can't solve it completely, cut the scope until you can — a small problem solved fully beats a large one gestured at.
Second, does it visibly demonstrate the judgment your paid offer is really selling? Not your competence at execution, which is now cheap and unconvincing, but your taste about what to do. When the buyer finishes, they should have watched you make a decision the way you'd make it for them.
Here's the one-line acceptance criterion. After someone consumes it, they should think: if the free thing is this good and this smart, the paid thing must be worth it. If your lead magnet doesn't reliably produce that sentence, it is not undersized or underpromoted. It is built on the wrong theory of what the buyer is sampling. Rebuild it as a sample of your judgment — because that, and not the PDF, is the only thing they were ever deciding whether to buy.