Choosing Platform Metrics

Strategic intent: Choose a small mix of output metrics that span the three dimensions of platform-marketplace performance, then break each one down into manageable input metrics that the team can actually move.

#Overview

There are countless ways to analyze the growth performance of a platform-marketplace. The risk is to either get lost in dashboards or to over-collapse to a single "north-star" metric that hides too much. The framework presented here covers three macro-layers that capture all the relevant aspects:

1. Reaching liquidity and delivering engagement
2. Increasing retention
3. Platform economics and sustainable business growth

The output is a mix of output metrics (3+ of them, one per layer) and a derivation of input metrics that the team can act on.

"Don't be fooled to pick only One Metric That Matters, nor the so-called North Star metric, as many would suggest." — Boundaryless Research

#When to use it

• After Growth Loops — the loops dictate which input metrics matter
• When the team is overwhelmed by dashboards and unclear on priorities
• When pivoting strategy and old metrics no longer fit
• Quarterly, as a discipline

#Layer 1 · Liquidity and engagement

Every metric in this layer is a variation on the buyer-to-supplier ratio (or Producer/Consumer ratio) — the main enabler of liquidity.

#From the producer perspective

• Minimum Order Flow (MOF) — the number of customers a supplier needs in a given time frame to stay engaged. Low MOF leads to disengagement and multi-tenanting (suppliers using competing marketplaces simultaneously).
• Average Order Value (AOV) and Frequency of Transactions (FOT) combine into MOF: MOF = f(AOV × FOT). They distinguish high-frequency/low-value marketplaces (e.g., gigs) from low-frequency/high-value ones (e.g., real estate).
• Utilization Rate (UR) — for asset-provider marketplaces (rentals, car-sharing): the % of available supply actually used in a time window.
• Zeros — unsuccessful matches (Li Jin & D'Arcy Coolican). Each zero is a clue about user-experience flaws or core-experience friction.
• Saturation Index (SI) — for marketplaces on Underutilized Fixed Assets (UFAs): how much the platform reduces the time the UFA sits unused. Inspired by Kevin Kwok's UFA framework.

#From the consumer perspective

• Search-to-fill (S2F) — % of searches that result in a transaction. Most relevant for marketplaces with curated supply (e.g., Fiverr).
• Time-to-fill (T2F) — time needed to fulfill a request. Most relevant for commoditized supply (e.g., Uber).

#Marketplace type matters (Josh Breinlinger)

The right liquidity metric depends on the marketplace type:

• Double-commit marketplaces (e.g., Upwork) — liquidity is harder; search-to-fill is the key metric
• Buyer-picks marketplaces (e.g., Airbnb with Instant Book) — supply is curated; time-to-fill is the key metric
• Marketplace-picks (e.g., Lingoda, Uber) — supply is componentized, marketplace assigns; MOF on the supply side is the key metric

In all three, number of full-time sellers is also worth tracking.

#Layer 2 · Retention

Retention metrics must be measured on a cohort basis — clusters of users onboarded in the same time window. Aggregate retention hides the cohort dynamics that matter.

#Concentration check (whale curves)

Before retention metrics, plot the whale curve (Andrei Brasoveanu): customers vs % of business they generate. Excessive concentration on one side — too few suppliers carrying too much volume, or too few categories, or too few geographies — creates fragility and pricing pressure.

#Behavioral retention metrics (Li Jin & D'Arcy Coolican)

• User Retention (UR) — % of a cohort still active after N periods. Older cohorts often retain better because they were the original target audience.
• Dollar Retention ($R) — for subscription/paid products: revenue retention per cohort. Healthy platforms see increasing dollar retention from newer cohorts as network effects compound.
• Core Experience Retention (CER) — % of users who perform and value the core experience.

#Direct feedback retention

• Net Promoter Score (NPS) — direct or indirect user feedback. Negative feedback is especially useful when it reveals technical/flow problems.

#Layer 3 · Platform economics

The economic layer captures sustainable business growth.

#Aggregate metrics

• Gross Merchandise Value (GMV) — total goods/services exchanged. Distinguish delivered GMV from contracted GMV (especially for service marketplaces, where the lag inflates contracted figures). Use GMV post-cancellations and returns.
• Net Revenues (NR) — what the platform actually keeps. From take rate, subscriptions, listing fees.

#Unit economics

• Customer Acquisition Cost (CAC) — costs to acquire a new customer. Fully loaded CAC charges everything: sales, onboarding, support, plus costs of attracting prospects who didn't convert.
• Customer Lifetime Value (LTV) — expected revenue from a single user over time. Measure by cohort.

#Fabrice Grinda's rules of thumb

• Fully loaded CAC recouped on a net contribution margin basis in 6 months
• 3× the CAC in 18 months
• Decreasing CAC trends with growing LTV (network effects kicking in)

#How to choose your mix

1. #1. Pick a mix of output metrics

   Choose at least 3 output metrics, one from each layer (liquidity/engagement, retention, economics). A single OMTM (One Metric That Matters) or NSM (North Star Metric) misses too much because it can't capture trade-offs between sides or layers.

2. #2. Find your input metrics

   Break down each output metric into manageable, actionable input metrics that the team can directly impact. The Reforge framing:

   "Output metrics represent results and input metrics represent actions."

   Be willing to experiment: discard input metrics that don't actually move the output, test new ones until you find the leading indicators that work.

3. #3. Account for tradeoff metrics

   Many metrics depend on each other. Improving one often degrades another. Identify the trade-offs explicitly so you can balance the system rather than over-optimize one variable.

4. #4. Distribute metrics across teams

   As the company grows, assign input metrics to specific teams. Each team contributes to growing one output metric by managing its leading actions. Amazon's OP-1 / OP-2 process is the canonical example of orchestrating this distribution.

#Inputs

• Required: active growth loops with their bottleneck steps identified
• Required: marketplace type classification (double-commit / buyer-picks / marketplace-picks)
• Required: cohort data (at least 3–6 months of usage history)
• Recommended: existing dashboard inventory (to deprecate vanity metrics)

#Outputs

• 3+ output metrics, one from each macro-layer
• Input metrics broken down per output, owned by specific teams
• Trade-off map — which metrics constrain which
• Cohort definitions — by acquisition window, geography, or category
• Whale-curve concentration check for the supply side

#Process heuristics

• The right metric depends on marketplace type — search-to-fill for double-commit, time-to-fill for buyer-picks, MOF for marketplace-picks
• Vanity metrics are seductive — sign-ups, page views, downloads feel good but don't predict value
• Negative feedback is valuable — especially worded NPS feedback that reveals friction
• Beware single OMTM/NSM — output metrics alone are too slow to drive day-to-day decisions
• Concentration is a fragility signal — whale curves shouldn't be too steep on either side

#Validation criteria

• One output metric chosen per layer (liquidity, retention, economics)
• Input metrics derived from each output, each with an owning team
• Cohort definitions are explicit (acquisition window, segmentation)
• Trade-off relationships between metrics are mapped
• Concentration check (whale curves) performed for supply side
• Marketplace type identified and metrics adjusted accordingly

#Common mistakes

• Single-metric thinking — OMTM or NSM alone hides too much
• Aggregate-only views — cohort dynamics are invisible
• Wrong metric for marketplace type — using time-to-fill on a double-commit marketplace gives noise
• Confusing contracted with delivered GMV — for service marketplaces this can overstate dramatically
• Ignoring concentration — too few whales on either side is fragility, not strength

#Used in pipelines

• Achieving Growth — as Phase 2

#Connections

• Requires: Growth Loops — input metrics come from the loop bottlenecks
• Feeds: Growth Model — these metrics become the model's variables
• Connects to: Building Liquidity — liquidity thresholds become guardrails post-liquidity

#Related reading

• Key Metrics for Platform-Marketplaces — the canonical Boundaryless essay (by Simone Cicero, Andrea Valeri, Manfredi Sassoli de Bianchi)
• The PDT Growth & Product Guide — full chapter "Measuring Improvement: Platform Metrics" in Legacy PDT Growth