Where to start
Marketplace liquidity is neither the volume of supply nor the count of transactions. It is the ability of two sides to find each other quickly enough and precisely enough that an exchange happens and both sides return. The definition is always bound to a micro-market and an observation window: in one category a week of waiting is normal, in another it means a lost transaction. A platform-level liquidity aggregate is almost always meaningless, because it averages markets with incomparable rhythms.
A practical requirement follows: liquidity cannot be measured with a single number. Only its separate manifestations are observable — the share of demand that closes, time to match, the share of supply that receives no exposure, and the share of realised matches that survive to a completed transaction. Each describes one part of the flow and can move against the others. Reading these measurements together is what working on liquidity means; collapsing them into one index is a separate methodological problem that current data does not solve.
The most durable practical implication: adding supply and adding liquidity are different operations. Additional supply raises liquidity only when it lands in micro-markets with unmet demand and receives sufficient exposure. Otherwise it competes for the same attention, raises search costs, and can worsen outcomes while headline counts rise. This is a working hypothesis of the research programme rather than an established fact; the conditions under which it holds are precisely what the programme investigates.
How to measure
- Fill rate — The share of eligible demand units for which available supply satisfies a pre-defined successful-outcome criterion within a specified window.
- Time to match — The duration between an eligible demand unit becoming available for matching and its first match that satisfies a pre-defined criterion.
- Unmet demand share — The share of eligible demand units that do not reach a pre-defined successful outcome before a stated observation window ends.
- Unmatched supply share — The share of eligible supply units that receive no qualified demand signal or valid match within a stated observation window.
- Zero exposure share — The share of eligible supply units that received no qualified exposure and no candidate-set inclusion during the observation window.
- Exposure coverage — The share of eligible supply units receiving a preregistered minimum amount of qualified exposure or candidate-set inclusion within the observation window.
- Qualified response rate — The share of supply units that received qualified exposure and then attracted at least one demand-side response meeting a preregistered qualification rule within the window.
- Transaction completion rate — The share of confirmed interactions that reach a preregistered successful completion event within the defined fulfilment window.
- Market thickness — The number of simultaneously active eligible participants on each side within one micro-market during the observation window, reported as a pair of supply and demand counts rather than a single summary figure.
- Supply-to-demand ratio — The ratio of eligible supply units to qualified demand units within the same preregistered opportunity unit: a micro-market and observation window.
Research and publications
- AI in marketplace matching: a conceptual map and research plan — A map of the layers where AI shapes the allocation of supply and demand, separating what an algorithm can improve from what it cannot create.
- Marketplace revenue architecture: from GMV to economic outcome — A framework connecting market activity, monetisation, execution reliability, and variable service cost without treating GMV as revenue.
- How to segment a marketplace into micro-markets — A step-by-step framework for segmenting a marketplace by actual substitutability so aggregates do not hide local scarcity and saturation.
- Seller activation: why registration is not activation — A diagnostic seller-activation framework covering the event chain from registration to eligible supply, first outcome, and repeat activity.
- How public marketplaces define their own metrics — Fifteen metric definitions from the annual reports of seven public marketplaces: identical names describe different quantities.
- State of Marketplace Liquidity: a research protocol — An open protocol for an annual study of marketplace liquidity: units of analysis, measures, data requirements, sensitivity tests, and non-claims.
- Supply without Demand: a research protocol — A protocol for testing when additional supply may weaken matching, supplier economics, or the platform's system outcome.
- Marketplace Liquidity Map — A working diagnostic map separating demand scarcity, unavailable supply, matching failures, and transaction-completion failures.
- Supply is not liquidity — A working hypothesis explaining why supply volume alone does not describe a marketplace’s ability to produce successful matches.
- Marketplace Health Index: a research protocol — A protocol for evaluating whether marketplace health can be summarised without arbitrary aggregation or false precision.
Frequently asked
How does liquidity differ from GMV and transaction count?
GMV and transaction count are outcomes observed after the fact and aggregated across the whole platform. Liquidity describes the conditions under which those outcomes arise and is defined inside a micro-market. Both can rise while matching degrades: it is enough for growth to come from a few dense micro-markets while the rest lose the ability to close demand.
Is there a healthy supply-to-demand ratio?
No universal value exists, and the supply-to-demand ratio card explicitly rejects one. The ratio is interpretable only when opportunity units are aligned on both sides, and only alongside outcome metrics — the share of demand that closes and time to match. The same ratio describes a healthy and a degraded market depending on whether the sides actually find each other.
Which measurements come first when nothing is instrumented yet?
Start with three that cover different parts of the flow: unmet demand share, the share of supply receiving no qualified exposure, and time to match. They are cheap to compute, need no surveys, and immediately localise the problem — a missing side, an allocation failure, or speed. Everything else is worth adding only once these three are computed stably and per micro-market.
How should micro-market boundaries be chosen?
By actual substitutability from the demand side: two supply units belong to the same micro-market if demand considers them alternatives. This is usually the intersection of category, geography, and price band, but the specific partition is validated against data rather than inherited from the catalogue tree. Fix the boundaries before computing and keep them stable across compared periods.
Can liquidity be compared across platforms?
Not directly. Nearly every metric in this cluster depends on platform definitions: what counts as qualified exposure, eligible supply, a completion event, and an observation window. Without a shared measurement protocol, comparing numbers across platforms compares definitions rather than markets. This is why the research programme starts with a protocol rather than a ranking.
Strongest counterargument
Separating liquidity from volume is unnecessary complication. Markets where supply grows usually grow in transactions too, and operating teams rarely have the data for micro-market partitioning. Managing familiar aggregates is simpler.
Response
The objection holds in two cases: in a growing market with unmet demand, and in a homogeneous micro-market where the aggregate matches reality. Outside those conditions the aggregate stops distinguishing scenarios that call for opposite actions — and supply-acquisition decisions are made precisely there. Published research gives grounds to treat the relationship between market thickness and realised match share as non-monotonic, so the assumption that more supply means a better market needs testing rather than presumption. The cost of the added complexity is bounded: the minimum set is three metrics computed per micro-market, not a hundred indicators.