Why I Don't Trust Pay-to-Rank, and What It Means for Find a Sign

<p>I've spent enough time around marketplaces — building them, using them, watching clients get burned by them — to have a strong opinion on how they should work. My view is that pay-to-rank is one of the more quietly corrosive business models in software, and most people building marketplaces don't even notice they've adopted it until it's load-bearing.</p><p>Here's the mechanism. A marketplace starts with good intentions: connect buyers and sellers, take a cut, everyone wins. But marketplaces need revenue, and the easiest lever to pull is letting suppliers pay for visibility. It feels harmless at first — a small boost, a featured badge, a bit of extra placement. Then it compounds. The suppliers who pay the most rise to the top regardless of whether they're the best fit for the customer, and the ones who don't pay sink, regardless of quality. Within a couple of years the marketplace isn't a discovery tool anymore. It's an auction house wearing a discovery tool's clothes.</p><p>The customer doesn't usually clock this directly. They just notice, vaguely, that the results feel off — too many ads-that-aren't-labelled-ads, too much sameness at the top, a nagging sense that they're being sold to rather than helped. That erosion of trust is slow, but it's real, and once it sets in it's very hard to win back. You can't un-ring that bell with a redesign.</p><p>This is the core reason I built Find a Sign around a different principle: transparent supplier listings, no pay-to-rank. Suppliers get found based on fit to what the customer is actually looking for, not based on who wrote the biggest cheque. It's a harder model to monetize in the short term — pay-to-rank is lucrative precisely because it's easy money that doesn't require you to solve the actual matching problem. But solving the actual matching problem is the point. If the marketplace isn't better than a Google search plus some cold calls, it has no reason to exist.</p><p>I think this matters more now than it did five years ago, because AI is about to make it both easier and more tempting to manipulate ranking in ways customers can't detect. A sufficiently sophisticated recommendation system can make pay-to-rank invisible — dressed up as 'personalization' or 'relevance scoring' while quietly weighting toward whoever's paying. That's a dangerous direction for the industry to drift in, because it removes even the blunt visual cues (obvious ad placements, banner spam) that let a skeptical customer self-correct. If the ranking logic is a black box and the box is for sale, nobody outside the company can tell the difference between genuine relevance and bought relevance.</p><p>My position is that AI should be used to make discovery more honest, not less — surfacing the supplier who's actually the best match for a customer's specific job, not the one who's paid for prominence. That's a harder product to build because it means the AI has to genuinely understand the problem space — signage types, materials, regional suppliers, project scope — rather than just running an auction. But it's the only version of the product I'd want to use myself, and I think that's a reasonable bar: if I wouldn't trust the system as a customer, I shouldn't ship it as a builder.</p><p>None of this is a unique insight — plenty of people have written about marketplace incentive design before me. But it's worth restating plainly, especially as more of these systems get an AI layer bolted on with the implicit promise that it's 'smarter' and therefore more trustworthy. Smarter isn't the same as more honest. The incentive structure underneath the intelligence is what decides whether the system serves the customer or extracts from them. That's the thing worth getting right before anything else.</p>