A dashboard of many gauges with a few central dials glowing and several peripheral ones dimmed and covered.

A marketplace generates an intimidating number of metrics, and most dashboards fill up with the ones that are easy to count rather than the ones that predict whether the business survives. Pageviews look reassuring; they tell you almost nothing. This article is a practitioner's shortlist: the KPIs that actually reflect marketplace health across demand, supply and discovery, the vanity metrics safe to ignore, and how to instrument the whole thing without lying to yourself. It pairs with the money mechanics in the marketplace business model guide — those explain the revenue; these tell you if it is healthy.

The core commercial KPIs

Start with the five numbers that describe the marketplace as a business.

KPIWhat it measuresWhy it matters
GMVTotal value of goods sold through the marketplaceThe top-line size of the network; the base your revenue is a share of
Take rateYour revenue as a % of GMVTurns GMV into your revenue; must stay defensible to vendors
AOVAverage order valueThe biggest lever in a niche marketplace — relevance and basket-building raise it
Repeat rateShare of customers who buy againThe engine of niche economics; repeat buyers cost nothing to reacquire
Vendor concentrationShare of GMV from your top vendorsA risk metric — heavy concentration means one vendor leaving can gut the marketplace

GMV and take rate give you revenue, but AOV and repeat rate are where a niche marketplace actually wins or loses. You cannot beat a horizontal giant on volume, so you beat it on relevance that lifts basket size and trust that brings people back. Vendor concentration is the one most operators forget to watch: a marketplace where two vendors drive most of GMV is fragile, and the fix is deliberate supply diversification, not more traffic.

Discovery KPIs: where sales are quietly lost

On a multi-vendor catalog, discovery is where most revenue leaks, and the metrics that expose it rarely make it onto executive dashboards. Three earn their place:

  • Search-to-cart rate. Of searches that return results, how many lead to an add-to-cart? A low rate points at relevance problems — the right products exist but are not surfacing.
  • Zero-result rate. The share of searches returning nothing is one of the highest-signal KPIs a store has: every zero result is a shopper who wanted something and was told you do not have it. Some of those are genuine gaps; many are synonym or catalog problems you can fix. We treat this as its own surface in the search work.
  • Skipped position-one rate. How often shoppers ignore the top result and click lower is a quiet indictment of ranking. Persistent skipping means your best slot is showing the wrong thing.

These are actionable in a way GMV is not: a rising zero-result rate is a to-do list, not just a number. Reviewing them weekly turns search failures into concrete catalog and synonym decisions.

The vanity metrics to ignore

Some numbers feel like progress and measure nothing you can act on. Treat these with suspicion as primary KPIs:

  • Raw pageviews and sessions. Traffic without conversion is cost, not health. A niche marketplace with modest traffic and high intent beats a busy, cold one.
  • Total registered users. An account count says nothing about whether people buy. Repeat rate and active buyers do.
  • Total SKUs listed. A huge catalog full of poorly enriched, un-findable products is worse than a smaller clean one. Findable SKUs matter; listed ones do not.
  • Social vanity numbers. Followers and likes rarely correlate with GMV in commerce. Watch attributed sales instead.

The test for any metric is simple: if this number moves, do I know what to do? If not, it is a vanity metric — interesting, not operational.

How to instrument it honestly

Good KPIs depend on a data pipeline you can trust, and there are two ways teams routinely fool themselves. The first is relying on a single analytics tool that lags, samples, and quietly excludes bots and staff — useful for trends, dangerous for exact figures. The second is trusting the browser to report events the browser has every reason to get wrong. The honest setup is a first-party event pipeline you own, joined at the seams:

  • Join query to click to cart with a per-request identifier, so you can see not just that a search happened but whether it led anywhere.
  • Join order to session so revenue attributes back to the discovery path and the content that drove it — the backbone of content-to-commerce attribution.
  • Verify counts server-side. Confirm impressions and events in your own database, not by trusting a number rendered in the browser.
From production

We instrument search with a per-request id that joins each query to the clicks it produced, review zero-result queries as a standing loop, and verify impressions in the database rather than trusting the browser. Order data joins back to session data so a sale attributes to the path that produced it — the difference between a dashboard that looks busy and one you can actually act on.

Reporting cadence

Metrics only help if someone looks at them on a rhythm and acts. Discovery KPIs — zero-result and search-to-cart — reward a weekly review, because their fixes (synonyms, catalog gaps, ranking tweaks) are small and frequent. Commercial KPIs — GMV, AOV, repeat rate, vendor concentration — move slower and suit a monthly view, ideally beside the bookkeeping figures your VAT and invoicing process already produces. The goal is not a prettier dashboard; it is a short list of numbers that each come with a clear action when they move.

One caution on cadence: resist the urge to react to daily noise on slow-moving commercial metrics. GMV and repeat rate wobble day to day for reasons that mean nothing — a single large order, a quiet weekend — and chasing that noise leads to whiplash decisions. Look at commercial KPIs as trends over weeks and months, and reserve the fast weekly loop for the discovery metrics whose fixes are genuinely small and frequent. Matching the review interval to how fast a metric actually moves is as important as picking the right metric in the first place.

Key takeaways

  • AOV and repeat rate win niche marketplaces, not volume — relevance lifts basket size and trust brings buyers back.
  • Watch vendor concentration as a risk metric: if a few vendors drive most GMV, one departure can gut the marketplace.
  • Discovery KPIs are the actionable ones — zero-result rate, search-to-cart and skipped-position-one turn lost sales into a to-do list.
  • Ignore vanity metrics like raw pageviews, account counts and total SKUs listed; the test is whether you know what to do when the number moves.
  • Instrument first-party and verify server-side: join query→click→cart and order→session yourself rather than trusting a lagging tool or the browser.

Frequently asked questions

What are the most important marketplace KPIs?
The commercial core is GMV, take rate, average order value, repeat rate and vendor concentration. Alongside those, the discovery metrics — zero-result rate, search-to-cart rate and skipped-position-one — are often more actionable because they point directly at fixable relevance and catalog problems. For a niche marketplace, AOV and repeat rate matter more than raw traffic, and vendor concentration is a risk metric worth watching closely.
What is a good search-to-cart rate?
There is no universal number, because it depends on catalog, category and how intent-heavy your traffic is — so the useful benchmark is your own trend over time. What matters is watching it move: a falling search-to-cart rate on searches that do return results signals a relevance problem, meaning the right products exist but are not surfacing. Pair it with zero-result rate to separate ranking issues from genuine catalog gaps.
Which marketplace metrics are vanity metrics?
Raw pageviews and sessions, total registered users, total SKUs listed, and social follower counts are the usual culprits. They feel like progress but rarely tell you what to do. A large catalog of un-findable products or a big account list that does not buy are worse than smaller, healthier equivalents. The test for any metric is: if it moves, do I know what action to take? If not, treat it as context, not a KPI.
How do you track sales attribution in a marketplace?
Own a first-party event pipeline and join it at the seams: link query to click to cart with a per-request identifier, and link each order back to its session so revenue attributes to the discovery path and the content that drove it. Verify impressions and events in your own database rather than trusting browser-reported counts, and treat single third-party analytics tools as trend indicators rather than exact figures.

A marketplace instrumented to tell you the truth.

We run the first-party event pipeline and discovery metrics so you see which searches, articles and vendors actually drive sales.

Request early access See the live marketplace →