Pricing a market where neither side knows the price

A functioning marketplace needs a price that both sides can look at and accept. We don't have one. Nobody in this category has one.

Ask a creator what a post costs and you get a number arrived at by a process I'd charitably describe as vibes: what someone paid them once, what a friend claims to charge, or what feels like enough money to be worth the hassle. Ask a brand what it's worth and you get a CPM calculation borrowed from display advertising, applied to a thing that is not display advertising.

Those two numbers are rarely close. Building a self-serve marketplace between them has been the central problem of this year.

The three options and why I picked the worst one#

Set the prices myself. Fastest path to a functioning market. I have more data than anyone on either side, so my guess would be the best guess available. The problem is that my guess would become the market — every subsequent price would anchor to it, including the ones that were wrong, and I'd never find out what the real price was because I'd have permanently replaced it with my opinion. Also, the moment I set prices I'm a party to the transaction rather than a venue for it, and that changes what I am legally and what I am to the people using me.

Let sellers name it. What we do. Every creator sets their own price. It's honest and it produces an enormous amount of garbage, because most sellers have genuinely never thought about it and the ones who have are calibrated against a market that doesn't exist yet. I've watched people with nearly identical audiences post prices 10x apart in the same week.

Auction it. Theoretically correct and requires liquidity I don't have. An auction with two bidders isn't price discovery, it's a coin flip with extra steps. Maybe later.

We do the second one. It's the worst option except for what it produces.

What the garbage is actually doing#

I've come around on the spread. For a while I saw it as a failure of the product — the market looks broken, therefore the market is broken. Now I think the spread is the product doing its job in public.

Every transaction at a bad price is a data point. When a wildly overpriced listing sits there for four months and never sells, that's information, and it's information the seller receives directly rather than from me. When something sells instantly, the seller raises the price next time without me telling them to. The market is teaching itself, slowly and embarrassingly, in a way that would not happen if I'd just published a rate card in year one.

The cost is that it's teaching itself in front of the customers.

The thing I got wrong#

I assumed sellers wanted to name their price. Autonomy, control, the whole framing I'd built the product on.

Most of them don't. Most of them want to be told, and the blank field is a source of anxiety rather than freedom. I built for the version of a seller who is me — someone who had to figure out what their audience was worth because nobody would tell them — and the median person signing up would much rather have a suggestion to accept or reject.

So we're adding a suggested range. Not a set price, a range, based on what comparable listings have actually transacted at. It's a smaller intervention than setting the price and it does most of the work.

I resisted this for about eight months on principle. The principle was fine. It was also costing us listings from exactly the people we most wanted, which is the sort of trade you should notice you're making rather than discover a year later.

Where I've landed#

Price discovery in a new category isn't a problem you solve, it's a process you either host or short-circuit. Hosting it is slower and produces a market that's actually yours. Short-circuiting it gets you revenue this quarter and leaves you owning a rate card instead of a marketplace.

I'm reasonably confident about this. Ask me again if we're still explaining the spread to brands in two years.