Marketplace liquidity is the probability that a listing finds a match, or that a buyer finds what they came for, inside a window short enough to keep both sides coming back. You diagnose it by measuring match rate and time to match on each side separately, then finding the side that is starving the transaction. The most common mistake is reading a liquidity problem as a demand problem and buying more of the side that was never the constraint.
Every marketplace I have walked into as a fractional operator has a number the team is proud of, and it is almost never the number that matters. At EverQuote we could show you millions of insurance shoppers a month. Impressive, until you asked how many of them, in a given state and a given insurance vertical, actually matched to a carrier that wanted their lead that week. That second number is liquidity, and it is the one that decides whether the business compounds or quietly leaks.
I have started at nine or ten companies now, depending on how you count, and the marketplace ones all share a pattern. The dashboard measures activity. Liquidity measures whether that activity resolved into a transaction. When those two numbers diverge, the team is usually pouring effort into the wrong half of the marketplace, and no amount of that effort moves the business, because the constraint is somewhere else entirely.
What marketplace liquidity actually measures
Marketplace liquidity is the likelihood a transaction happens: the share of listings that get matched, and the share of buyer visits that end in a purchase, inside a set window. Measure it as a match rate on each side separately, seller-side sell-through and buyer-side conversion, because a marketplace can look liquid on one side and be starving on the other.
The methodology is not exotic. As the Point Nine team laid out in their widely cited breakdown of marketplace liquidity, you take the transactions in a period over the listings or the visits in that same period, and you do it for each side. A hundred listings with seventy-five sold in thirty days is a 75 percent seller-side match rate. Ten thousand visits with two hundred purchases is a 2 percent buyer-side rate. The two rarely move together, and the gap between them tells you which side is doing the waiting.
At EverQuote the constrained side was carrier demand for specific lead segments, not shopper supply. National shopper counts looked enormous and hid the fact that in a particular state, for a particular vertical, the carriers who wanted those leads that week were thin. Andrew Chen makes the same point structurally in The Cold Start Problem: every two-sided marketplace has a hard side, and liquidity lives locally rather than nationally. A healthy total across the whole market can sit on top of dozens of dead local ones.
Why liquidity problems get misread as demand problems
A liquidity problem looks like a demand problem because both show up the same way, as too few transactions. The tell is what happens when you add traffic. On a real demand problem, conversions rise with visits. On a liquidity problem, more visits just means more people leaving unmatched.
This is where marketplaces burn money. The instinct when transactions are low is to buy more of the obvious side, usually buyers, because buyer acquisition is a channel you can turn on with a credit card. I wrote a whole piece on where marketplace growth goes to die for this reason. If the match rate is the constraint, paid acquisition does not fix it. It raises the denominator and lowers the very rate you were trying to improve, and it does it while spending real money.
More traffic on a marketplace with a match-rate problem just raises the number of people who leave unmatched.
The counter-narrative I keep returning to is TripAdvisor Instant Booking. We took hotel booking from zero to more than two hundred million dollars in transaction value in eighteen months, and that number was real. The honest part is what it taught me about the difference between inventory and liquidity. Making a hotel transactable by putting a booking button on it is not the same as making it liquid. If travelers were not converting on that property before, a button does not change the match rate; it just gives you a transactable listing that still does not transact. Adding inventory is not the same as adding liquidity, and I learned that on one of the biggest wins I was ever part of.
How I use AI to run the diagnosis in days, not weeks
AI does not diagnose liquidity. It removes the three or four weeks of data archaeology that used to sit between an operator and the diagnosis. I use it to pull cohorts, compute match rates by segment, and read session data, so the judgment call arrives in week one instead of week four.
The contrast is personal. In 2010 I instrumented my own company, EditMe, by hand, writing SQL against the production database to build trial cohorts by signup month and lifetime value by cohort. It took days, and it was the only way to see whether the self-serve funnel was actually converting. That same analysis today is an afternoon. I point an agent from my AI stack at the marketplace's own data and it returns match rate by state, by vertical, by supply cohort, and by time to match, cross-referenced against the pattern I have seen at other marketplaces, before I have finished reading the onboarding docs.
What has not changed is where the judgment sits. The agent produces the match-rate table; deciding that carrier demand in three states is the real constraint, and that everything the team is doing on buyer acquisition is effort spent on the wrong side, is still mine to make. AI compressed the part that was slow and left the part that was hard. That is the honest version of using AI in a first-30-days diagnosis, and it is a long way from the claim that a model hands you the answer.
The first-thirty-days marketplace liquidity diagnosis, in order
Run the diagnosis in one order. Define the transaction, measure match rate and time to match on each side, find the starved side, then decide whether the fix is supply, demand, matching, or price. Skipping straight to the fix is how teams end up solving the wrong side.
The order matters most on the clock. A solar marketplace I worked with matched homeowners to installers, and solar takes about twelve weeks to close. A team measuring weekly listing activity there is reading the wrong clock entirely, because the transaction it cares about resolves a quarter later. The liquidity metric has to match the real cycle of the transaction, or you will call a market dead that is simply slow, and call a market healthy that is churning listings without closing anything.
| Signal | What it measures | What a weak reading means |
|---|---|---|
| Match rate, per side | Share of listings or visits that resolve into a transaction in-window | The side with the lower rate is your real constraint |
| Time to match | How long a listing or a buyer waits before a transaction | A clock longer than the category norm means the two sides are not meeting |
| Concentration | Whether liquidity is local and segment-level or diluted across the whole market | Healthy national totals can sit on top of dead local markets |
| Response to traffic | What conversions do when you add visits | Flat conversions on more traffic means the problem was never demand |
Once the starved side is named, the fix follows from which lever moves the match rate. Sometimes it is supply on the hard side. Sometimes it is a matching or ranking change so the demand that exists actually meets the supply that exists. Sometimes it is price, which is a liquidity lever more often than teams admit, and which fails in a specific way when nobody owns the price. And frequently the highest-value move is not a growth project at all but an organizational one, because the supply and demand sides of a marketplace often need to be run as separate teams with separate metrics before liquidity gets a real owner. That is also why liquidity makes such a good candidate when a team is trying to get a room to agree on one number.
Liquidity is not a stage you clear on the way to scale. It is the diagnosis you re-run every time the mix changes, a new vertical, a new city, a new supply cohort, because each of those is a small cold start hiding inside a business that already works. The teams that keep compounding are the ones that never stop asking which side is starving. The number they are proud of is rarely the answer.
