Metric trees for marketplace businesses

A standard SaaS or e-commerce metric tree flows in one direction: from a single top-level revenue number down through a funnel the company controls end to end. Marketplaces break this model. Revenue depends on two groups of participants whose behaviours are interdependent but separately motivated. Suppliers join because there are buyers. Buyers join because there are suppliers. Neither side is fully under the platform's control, and the health of the whole system depends on the balance between them.

This creates three structural challenges for measurement. First, you need separate branches for supply and demand because the levers, owners, and leading indicators are entirely different on each side. Second, you need a layer that captures the interaction between supply and demand, which is what marketplace operators call liquidity and match quality. Third, you need to track the platform's economics separately from the total economic activity it facilitates, because GMV and revenue are very different numbers.

A metric tree designed for a marketplace addresses all three. At the top sits GMV or Net Revenue. Below that, the tree forks into a supply branch, a demand branch, and a transaction quality branch. Each fork decomposes further into the specific metrics that teams can act on. The result is a single structure that shows how supplier onboarding, buyer acquisition, matching algorithms, pricing, and trust mechanisms all connect to the same top-level outcome.

Gross Merchandise Volume is the total value of goods or services transacted through the marketplace over a given period. It is the number that captures scale. But GMV alone tells you nothing about the platform's financial health. A marketplace with £100 million in GMV and a 5% take rate earns £5 million in revenue. A marketplace with £20 million in GMV and a 25% take rate also earns £5 million. The metric tree must decompose both the volume and the economics.

The core equation is:

Net Revenue = GMV x Take Rate

GMV itself decomposes further. The most useful decomposition for a marketplace is:

GMV = Active Buyers x Transactions per Buyer x Average Transaction Value

This mirrors the e-commerce equation (Sessions x Conversion Rate x AOV) but adapts it for a marketplace context. Active Buyers replaces Sessions because marketplace health depends on the size of the engaged buyer base, not just visit volume. Transactions per Buyer captures repeat usage, which is more important in a marketplace than in a one-off e-commerce purchase. Average Transaction Value functions like AOV but can vary enormously depending on the category mix within the marketplace.

Take Rate deserves its own branch because it is rarely a single number. Most marketplaces earn revenue from multiple streams: a commission on each transaction, payment processing fees, promoted listings or advertising, insurance or trust and safety products, and value-added services like logistics, financing, or analytics. Decomposing Take Rate into its components reveals which revenue streams are growing and which are under pressure. A marketplace that appears to have a stable Take Rate might actually be seeing commission revenue decline while advertising revenue compensates. The metric tree makes this visible.

Benchmarks vary significantly by category. Physical goods marketplaces typically operate at 5-20% take rates. Service marketplaces range from 10-30%. Luxury or niche verticals can command 25% or more. The right take rate for your marketplace depends on the value you add, the competitive alternatives available to suppliers, and the price sensitivity of buyers. The metric tree helps you track whether changes in take rate affect supplier retention or buyer behaviour, which are the early warning signs of pricing pressure.

The most common mistake in marketplace measurement is treating supply and demand as a single funnel. They are not. Suppliers and buyers have different motivations, different acquisition channels, different onboarding journeys, and different retention dynamics. The metric tree must reflect this by maintaining two parallel branches below GMV.

The supply-side branch tracks the health of your seller, provider, or inventory base. It answers the question: do we have enough of the right supply, in the right locations or categories, at the right quality level, to satisfy demand? The demand-side branch tracks the health of your buyer base. It answers the question: are we attracting enough buyers, converting them efficiently, and retaining them over time?

Both branches follow a similar structure: acquisition, activation, engagement, and retention. But the specific metrics within each stage are different.

On the supply side, the metrics that matter most depend on your marketplace type. For a goods marketplace like Etsy or Amazon Marketplace, the key supply metrics are the number of active listings, listing quality, pricing competitiveness, and fulfilment reliability. For a services marketplace like Upwork or Thumbtack, the key supply metrics are provider availability, response time, service quality ratings, and geographic or skill coverage. For a rental marketplace like Airbnb, supply metrics focus on inventory utilisation, availability calendar accuracy, and host responsiveness.

On the demand side, the critical distinction is between first-time buyers and repeat buyers. First-time buyers test whether your marketplace can deliver value. Repeat buyers prove it. The metric tree should segment the demand branch by cohort so you can see whether buyer quality is improving over time. A marketplace that is growing GMV purely through new buyer acquisition without improving repeat rates is on an unsustainable path.

The connection between the two branches is what makes the marketplace model powerful and what makes it hard to measure. When you improve supplier coverage in a category, buyer conversion in that category rises. When buyer demand grows, more suppliers are attracted to the platform. The metric tree must track both branches independently while also capturing the interaction effects that connect them.

Liquidity is the defining metric for a marketplace. It measures how reliably and quickly buyers can find what they want and suppliers can find buyers. A marketplace with high liquidity feels effortless to use. A marketplace with low liquidity feels like a ghost town, no matter how many registered users it has.

The simplest definition of liquidity is the percentage of searches or requests that result in a completed transaction. This is sometimes called the search-to-fill rate. If a buyer searches for a product or service and finds a suitable match that leads to a purchase, the marketplace is liquid for that query. If the buyer searches and finds nothing relevant, or finds options but none that meet their quality or price expectations, the marketplace is illiquid for that query.

Liquidity decomposes into four components: supply density, demand density, match quality, and trust. Each one can be measured and improved independently, which makes them ideal branches in a metric tree.

1. 1

   Supply density

   The number of relevant listings or providers available for a given search, location, or category. Density matters more than total supply count. Having 10,000 listings nationally is meaningless if a buyer searching in a specific city finds only two. Track density at the level of granularity that matches how buyers search: by category, by location, by price range, by availability window.

2. 2

   Demand density

   The volume of buyer interest concentrated within a given category, location, or time period. High demand density gives suppliers confidence that listing on your platform will generate business. Track demand density alongside supply density to identify imbalances: categories where demand outstrips supply (opportunity) and categories where supply outstrips demand (risk of supplier churn).

3. 3

   Match quality

   How well the marketplace connects the right buyer with the right supplier. Measured through transaction completion rate, post-transaction satisfaction scores, return or dispute rates, and repeat transaction rates with the same supplier. Poor match quality erodes trust even when supply and demand are both abundant.

4. 4

   Trust and safety

   The infrastructure that gives both sides confidence to transact. Reviews and ratings, identity verification, payment protection, dispute resolution, and insurance all contribute. Trust is a threshold metric: below a certain level, transactions simply do not happen regardless of supply and demand balance.

The practical challenge with liquidity is that it varies enormously across the marketplace. Your overall search-to-fill rate might be 40%, but that aggregate hides categories at 80% and categories at 5%. The metric tree should decompose liquidity by category, geography, and time to expose these pockets of illiquidity. A marketplace growth strategy often involves systematically identifying illiquid segments and investing in supply acquisition or demand generation to bring them above the threshold where the flywheel begins to spin.

Time-to-match is another critical liquidity metric, particularly for services and labour marketplaces. A buyer who posts a job request and receives a qualified response within minutes has a fundamentally different experience from one who waits days. In ride-sharing, the equivalent is wait time. In food delivery, it is delivery time. Whatever form it takes, reducing time-to-match improves conversion, satisfaction, and repeat usage simultaneously.

Track utilisation rate alongside liquidity. Utilisation measures what percentage of available supply is being consumed by demand. Low utilisation signals that you have more supply than you need in a segment, which is wasteful and may cause suppliers to churn. High utilisation signals that you may not have enough supply, which causes buyers to see "sold out" or "unavailable" states. Neither extreme is healthy. The metric tree should make utilisation visible per segment so you can rebalance supply and demand proactively.

Every marketplace founder knows the chicken-and-egg problem: you need supply to attract demand, but you need demand to attract supply. What is less discussed is how this dynamic affects your metric tree and which metrics to prioritise at each stage of marketplace maturity.

In the earliest stage, before the marketplace has reached critical mass, traditional metrics like GMV and take rate are nearly meaningless. The numbers are too small to be diagnostic. What matters instead are leading indicators of liquidity: are you building enough supply density in your initial categories or geographies to give early buyers a good experience? Are those early buyers converting and coming back? The metric tree at this stage should be heavily weighted toward supply health and early demand signals.

The chicken-and-egg problem also creates a measurement trap. If you track supply and demand metrics independently without tracking the interaction between them, you can convince yourself the marketplace is healthy when it is not. Imagine a marketplace that is growing supplier sign-ups by 20% per month and buyer sign-ups by 15% per month. Both numbers look strong. But if search-to-fill rate is declining, it means the new supply and new demand are not matching well. Perhaps you are adding suppliers in categories where demand is already saturated, while neglecting categories where buyers are searching and finding nothing.

The metric tree solves this by placing liquidity metrics between the supply and demand branches. When supply grows but liquidity does not improve, the tree immediately surfaces the disconnect. This is why the interaction layer is not optional. It is the layer that tells you whether growth on each side is translating into marketplace value.

Another common pitfall is optimising for one side at the expense of the other. A marketplace that subsidises buyers with aggressive discounts will drive transaction volume but may erode supplier economics to the point where quality suppliers leave. Conversely, a marketplace that offers generous supplier terms but charges high buyer fees may maintain supply quality but struggle with buyer acquisition. The metric tree should track unit economics on both sides, including supplier lifetime value and buyer lifetime value, so that teams can see when one side is being subsidised unsustainably.

A marketplace metric tree is more complex than a single-sided business because it must capture two funnels and the interaction between them. But the principles of good tree design still apply: start from the top, decompose using real equations, assign ownership, and connect to live data.

1. 1

   Choose the right top-level metric

   For most marketplaces, Net Revenue (GMV x Take Rate) is the right North Star because it reflects both scale and monetisation. Early-stage marketplaces may use GMV or even liquidity (search-to-fill rate) as the top-level metric if monetisation is not yet the priority.

2. 2

   Fork into supply, demand, and interaction branches

   Below the top-level metric, create three parallel branches. The supply branch tracks supplier acquisition, activation, listing quality, and retention. The demand branch tracks buyer acquisition, conversion, and retention. The interaction branch tracks liquidity, match quality, and time-to-match.

3. 3

   Decompose by segment, not just in aggregate

   A blended liquidity number hides the categories and geographies where your marketplace is thriving and those where it is failing. Decompose every metric by the segments that matter to your business: category, geography, price tier, or customer type.

4. 4

   Track both sides of unit economics

   Calculate LTV:CAC for suppliers and buyers independently. A healthy marketplace typically needs a 3:1 ratio or better on both sides. If one side has a ratio below 1:1, you are paying more to acquire participants than they are worth, which signals a leaky bucket somewhere in the tree.

5. 5

   Assign dual ownership at the interaction layer

   Supply metrics have supply team owners. Demand metrics have demand team owners. But liquidity and match quality sit between the two. Assign these metrics to a marketplace operations or growth team that has visibility into both sides and the authority to rebalance resources.

The most common structural mistake in marketplace metric trees is building a single funnel that looks like a standard e-commerce tree. This works for the demand side in isolation, but it ignores the supply side entirely and misses the interaction dynamics that determine marketplace health. If your metric tree does not have a supply branch, you are flying half-blind.

The second most common mistake is treating GMV as revenue. A marketplace that reports "50% GMV growth" sounds impressive, but if take rate declined from 15% to 10% over the same period, net revenue grew by only 10%. The metric tree must keep GMV and take rate as separate branches so that growth in volume and growth in monetisation are tracked independently.

KPI Tree is designed to handle the structural complexity that marketplace businesses require. You can build parallel supply and demand branches, add an interaction layer for liquidity metrics, decompose by segment, and connect every node to live data from your marketplace platform, payment processor, and analytics tools. Each node can have a dedicated owner, targets, and linked actions so that your weekly marketplace review is structured around the tree rather than a collection of disconnected dashboards.