GMV vs NMV vs PAT vs EBITDA: Financial Metrics, Revenue Realization, and Startup Valuation Frameworks

These Metrics Form a Funnel, and Each Step Loses Something

These metrics are usually introduced as a list of definitions to memorise, which is why they rarely stick. The clearest way to hold them in your head instead is as a single descending sequence, where each step subtracts something real and the gap between steps is itself informative.

At the top sits GMV, the total value of everything transacted. Below it, NMV removes what did not complete — cancellations, returns, undelivered orders. Below that, revenue keeps only what the company itself earns, which for a marketplace is a commission rather than the transaction value. Then EBITDA subtracts operating costs while deliberately ignoring financing, tax and non-cash charges. Finally PAT subtracts everything else and reports what shareholders actually own.

The gaps carry the meaning. A wide GMV-to-NMV gap indicates operational or quality problems — people are ordering and then not receiving, or receiving and returning. A wide NMV-to-revenue gap is simply a low take rate, which may be strategic or may indicate weak pricing power. A wide revenue-to-EBITDA gap means the operating cost base is heavy relative to what the business earns. And a wide EBITDA-to-PAT gap points at debt, capital intensity or tax, none of which are optional.

Reading the sequence rather than any single figure is what makes it useful. A company can be first in its category on GMV and last on PAT, and both statements can appear in the same annual report without contradiction. Neither number is dishonest; they measure different things, and the distance between them is the story.

Why GMV Became the Number Everyone Quotes

GMV dominates marketplace communication, investor updates and internal dashboards for a reason that has considerably more to do with incentives than with analytical merit. It is the largest available number, it grows fastest, and it is the easiest to influence directly.

That last property is the problem. GMV responds immediately to discounting, to subsidised delivery, and to expansion into low-margin categories. A team under pressure to grow GMV can do so reliably by selling more things at worse economics, and the metric will reward them for it. Growth achieved this way is genuine in the sense that the transactions happened, and largely meaningless in the sense that the business is no closer to being viable.

This is why GMV should effectively never be reported alone. Paired with take rate it shows whether growth is translating into revenue. Paired with contribution margin it shows whether that revenue survives the costs of delivering it. Paired with the NMV gap it shows whether the transactions are completing at all. Any one of those three can reveal that impressive GMV growth is a costly way of standing still.

There is a governance version of this worth naming: whatever number leadership repeats in all-hands meetings is the number the organisation will optimise, whether or not that was intended. Choosing GMV as the headline metric is therefore a decision about behaviour, not just about reporting. Teams will find ways to move it, and if the accompanying quality metrics are not equally visible they will move it in the cheapest available way.

The GMV-to-NMV Gap as an Operational Signal

NMV receives the least attention of these four metrics and arguably carries the most operational information, because the gap between it and GMV is a direct measure of how much of what you sold actually stuck.

That gap has three main components, and separating them points at different owners. Cancellations before dispatch usually indicate a pricing, stock accuracy or delivery promise problem — the customer changed their mind before receiving anything, often because the promised date slipped or the item turned out to be unavailable. Failed deliveries point at logistics and address quality. Returns after receipt point at product expectation: sizing, imagery, description accuracy, or genuine quality issues.

Tracking the ratio rather than the absolute figures is what makes this useful. A cancellation rate that moves from a stable baseline to something noticeably higher over a few weeks is a signal that something operational broke, and it will typically be visible in NMV well before it shows up in customer satisfaction scores or support volume. Very few businesses monitor this ratio as a leading indicator, and it is close to free once both numbers exist.

There is also a financial reason to care beyond operations. Every unit of the gap consumed real cost before it reversed — picking, packing, delivery, and frequently reverse logistics on top. Returned orders are not neutral; they are actively expensive, and a business with a large gap is paying to acquire, fulfil and then unwind a meaningful share of its own demand. In categories where the gap is structurally large, contribution margin calculated on GMV rather than NMV is not merely imprecise but wrong by a wide margin.

What EBITDA Hides, and When It Is Still the Right Metric

EBITDA attracts more criticism than any other metric in this set, much of it deserved and some of it misplaced. Understanding both halves is more useful than adopting a position.

The legitimate case for it is comparability. Two companies in the same sector may have entirely different capital structures, tax jurisdictions and asset ages. Interest expense reflects how a business was financed rather than how well it operates; tax reflects where it is domiciled; depreciation reflects when assets were purchased and under what accounting policy. Stripping those out lets you compare operating performance on something closer to like-for-like terms, which is a real analytical need.

The illegitimate use is treating it as a proxy for cash or for profit. Depreciation is a non-cash charge in the current period and a very real one over time — assets wear out and must be replaced, and a business that ignores this is deferring a cost rather than avoiding it. For capital-intensive businesses, EBITDA can look healthy for years while the underlying economics deteriorate.

The more subtle omission is working capital. EBITDA says nothing about cash tied up in inventory or receivables, and a growing commerce business consumes working capital aggressively. It is entirely possible — and common — for EBITDA to improve every quarter while the bank balance falls, because every additional unit of growth requires stock purchased before it is sold. Businesses do not fail because EBITDA turned negative; they fail because they ran out of cash, and those are different events that can occur in either order.

A practical habit for anyone reading these numbers: whenever EBITDA is presented prominently, look for free cash flow. If it is absent from the same document, that absence is usually the more informative signal.

Reading Financial Communication Critically

Nobody presenting these numbers is obliged to present all of them, and which ones appear tells you a great deal before you examine any values. A few patterns recur often enough to be worth recognising.

A metric appearing for the first time usually means it has just become flattering. A company that has reported revenue growth for several quarters and suddenly leads with adjusted EBITDA has generally had something happen to revenue growth. The introduction of a new headline metric is itself information, independent of what the metric says.

Heavy adjustment deserves attention. Adjusted EBITDA excluding one-off items is reasonable when the items are genuinely one-off; when the same category of exception appears every year, it is a recurring cost being presented as an anomaly. Restructuring charges in four consecutive annual reports describe an operating model, not an exception.

Growth rates without absolute figures obscure scale, and percentage improvements on small bases are easy to generate. A category growing by a large multiple may have been negligible to begin with. Conversely, absolute figures without growth rates obscure trajectory. Presenting one consistently and the other only when convenient is a pattern worth noticing.

Changed definitions are the hardest to spot and the most consequential. A company that restates how it calculates active users, or shifts what sits inside a contribution margin layer, breaks the comparability of its own history. Footnotes disclosing a methodology change are usually the most important text on a page of financials and reliably the least read.

None of this implies bad faith is common. Presenting the most favourable accurate framing is ordinary behaviour, and the defence against it is not suspicion but structure: decide which metrics you care about in advance, ask for them consistently, and treat their absence as a data point rather than an oversight.

Matching the Metric to the Question

Each of these figures answers a specific question well and several others badly. The practical skill is matching them rather than ranking them.

Use GMV when the question is about market position and scale — how large is the platform relative to competitors, is the category growing, is supply expanding. It is a legitimate measure of a marketplace's importance in its market, provided quality metrics accompany it.

Use NMV when the question involves operational quality, and always in categories with meaningful returns. The GMV-to-NMV ratio tracked over time is a genuinely useful operational health metric that many businesses never look at.

Use revenue and take rate when the question is whether the business model captures value. Rising GMV with falling take rate is a specific and important pattern, usually indicating expansion into more competitive segments or increased subsidy.

Use contribution margin when the question is about the viability of an additional order or customer, which is the level most operational decisions actually occur at — our guide to the contribution margin layers covers how to decompose it.

Use EBITDA for cross-company operating comparison and for debt covenant purposes, where it is the convention. Use PAT when the question is what shareholders earned. Use free cash flow when the question is whether the business can fund itself — which, for most companies most of the time, is the question that actually matters.

One organising principle ties these together and is worth carrying away even if the individual definitions fade. The higher up the funnel a metric sits, the faster it responds to action and the less it says about viability. GMV moves within days of a promotion; PAT moves over quarters and reflects whether any of that movement was worthwhile. Teams gravitate naturally toward the responsive metrics because they provide feedback quickly, and that pull is precisely why the slower ones have to be kept deliberately in view.

The practical expression of that principle is to pair every headline number with the one directly beneath it. GMV never appears without take rate. Revenue never appears without contribution margin. EBITDA never appears without free cash flow. Each pairing costs one additional column in a report and removes the single most common way that metric misleads the people reading it.