What is the GP3 Waterfall? It’s Tronvik’s methodology for re-mapping an e-commerce target’s P&L by cost behaviour rather than accounting label: isolating GP1 (product), GP2 (fulfilment), and GP3 (marketing) margin in sequence, down to EBITDA, to show where a blended gross margin figure comes from.
Most e-commerce targets present a single gross margin figure. It tells a PE fund almost nothing about where that margin comes from, or where it breaks under institutional ownership. A target can show a healthy blended gross margin while its product economics, fulfilment economics, and acquisition economics are each degrading in ways that only surface after close.
The Tronvik GP3 Waterfall exists to separate those three questions before the LOI is signed.
Why a Single Gross Margin Number Isn’t Enough
Standard financial reporting aggregates costs into broad functional categories (COGS, opex, marketing) that follow accounting convention rather than operational behaviour. That aggregation is what lets sellers optimise the story ahead of a process: variable costs get reclassified as fixed, marketing spend gets buried in G&A, and founder compensation gets set below market rate. None of this is illegal. All of it inflates the number an acquirer models against.
The GP3 Waterfall re-maps the P&L by cost behaviour, volume-dependent variable cost versus time-dependent fixed overhead, rather than by accounting label. That reclassification is where the real risk becomes visible.
The Waterfall: GP1, GP2, GP3, and EBITDA
Starting from Net Revenue, four cost layers are peeled off in sequence:
- Direct Material (COGS, inbound freight, duties) is subtracted to produce GP1 / GM1
- Fulfilment (outbound logistics, 3PL, warehouse) is subtracted to produce GP2 / GM2
- Marketing (performance spend, growth agency fees) is subtracted to produce Variable GP3 / GM3
- Fixed Costs (G&A, salaries, rent, SaaS) are subtracted to produce EBITDA
Each layer produces two numbers: a GP figure (the absolute margin remaining after that layer’s costs) and a GM figure (that margin expressed as a percentage of net revenue). The GP figure shows scale; the percentage shows trend and comparability across periods and against category benchmarks. Reading GP without GM, or GM without GP, hides half the picture.
For the full walk broken into every cost line inside each layer, with a visual reference and complete definitions, see The Gross Profit Walk.
GP1 / GM1: Product and Material Margin
What remains after direct material costs: product COGS, inbound freight, customs duties and tariffs. GM1 answers whether the target is a price taker or a price maker: whether it can pass sourcing volatility through to the consumer or absorbs it. Single-source suppliers, manufacturing concentrated in volatile geographies, and unhedged commodity inputs all compress GM1 in ways a blended gross margin line will never isolate.
Two regulatory drivers compress GM1 in ways specific enough to isolate on their own: EU product-safety non-compliance (see GPSR Due Diligence) and, for white-label China-sourced portfolios specifically, Digital Product Passport exposure (see You Are the Manufacturer Now).
GP2 / GM2: Fulfilment and Distribution Margin
What remains after outbound freight, 3PL fees, warehouse labour, and transactional customer support (order tracking, returns processing). GM2 measures whether growth requires proportional capital injection into logistics or whether the target has genuine density and infrastructure efficiency. A single-carrier dependency or inefficient multi-hop shipping network shows up here first, usually after volume has already scaled past the point where it’s cheap to fix.
A low-touch support cost line can look like exactly this kind of infrastructure efficiency and be something else entirely. See The Chatbot That Never Lets You Leave for how a support structure engineered to be unreachable manufactures the same GM2 signature as a genuinely efficient one. The same false signal shows up on the returns side of GM2: see Cheaper to Bin Than to Return for how a return-shipping fee engineered to deter use manufactures an artificially low return rate. Where a target sells by subscription, the same obstruction logic sits on the exit and distorts a metric outside the margin line entirely: see Easy In, Hard Out for how cancellation friction suppresses churn, and with it the retention a recurring-revenue multiple is paid on.
GM2 also carries Extended Producer Responsibility exposure, and that liability doesn’t degrade smoothly the way the others do. See One Registration Per Country for why EPR behaves as a contingent liability that multiplies with a target’s cross-border market count rather than a gradually compressing margin.
Platform fees are the visible marketplace cost inside GM2. The invisible one is continuity: marketplaces are now obliged to verify their traders and suspend the ones whose papers don’t hold up. See The Marketplace Enforces First for how DSA trader traceability turns marketplace-heavy revenue into a compliance-triggered concentration risk.
Variable GP3 / GM3: Unit Contribution and Scalability
What remains after performance marketing spend and variable growth agency fees. GM3 is the single most diagnostic number in the waterfall: it shows whether customer acquisition is structurally profitable before any corporate overhead is applied. A target renting its customers via constantly escalating ad spend, rather than owning a genuine brand moat, will show a GM3 that degrades as volume scales, not one that holds or improves.
For targets that sell D2C alongside retailers, that degradation often has a specific, identifiable cause rather than a generic one. See Bidding Against Yourself for how an ungoverned brand-retailer search auction shows up directly in GM3.
A strong GM3 can also be borrowed rather than earned. See The Permanent Sale for how inflated reference prices inflate conversion, and why that GM3 resets once pricing has to comply with the Omnibus 30-day rule. The same borrowed-signal logic applies to the star rating: see Borrowed Trust for how a manufactured review base inflates conversion and corrupts the buyer’s own evidence at the same time. For the general case, where an above-market conversion rate is the aggregate output of manufactured signals rather than of product strength, see Too Good to Be Earned. And beneath all of it sits the audience itself: see Consent Debt for how retargeting pools and analytics built on invalid cookie consent subsidise CAC with data the target may be ordered to erase.
How Sellers Move Costs Across the Waterfall
Three reclassification patterns recur at the general-ledger level:
- The Marketing Camouflage: external agency retainers or growth-focused salaries parked in fixed G&A instead of the Marketing layer, making GM3 look artificially strong.
- The Warehouse Shift: variable 3PL labour or shipping overages moved into fixed overhead to protect the headline gross margin, understating true GM2 pressure.
- Founder Wage Re-baselining: below-market founder compensation or distributions in place of market-rate salary, inflating EBITDA until a market-rate replacement hire is priced in.
Each of these narrows the gap between the reported number and the number that survives institutional ownership. None of them are visible from a single blended gross margin line.
What Outside-In Analysis Can Detect Before the Data Room
Before the LOI, waterfall mapping uses public and semi-public signals to build a directional GP1–GP3 / GM1–GM3 profile:
- Supplier concentration and sourcing geography (product listings, brand registrations, import records)
- Fulfilment footprint and carrier dependency (shipping policy language, delivery time patterns, warehouse job postings)
- Acquisition channel mix and saturation signals (traffic trend proxies, ad transparency libraries, promotional discount frequency)
- Reclassification indicators consistent with the Marketing Camouflage and Warehouse Shift patterns above
None of this replaces formal due diligence. It establishes a directional range for GM1, GM2, and GM3 before the acquirer commits time and capital to a full process, and it flags which waterfall layer the formal DD team should stress-test first.
The Pre-LOI Question Every PE Fund Should Ask
The question is not “what is this target’s gross margin.” It’s: which layer of the waterfall is that margin coming from, and which layer breaks first under institutional ownership?
A target with strong GM1 but collapsing GM3 has a sourcing advantage and a customer-acquisition problem. A target with strong GM3 but weak GM2 has a brand and an infrastructure problem. Both can report the same blended gross margin. Only the waterfall tells them apart, and only telling them apart before the LOI gives the acquirer leverage on price instead of a post-close surprise.
This analysis is part of Tronvik’s GP3 Waterfall methodology. To initiate a GP1–GP3 / GM1–GM3 mapping on a specific acquisition target, contact info@tronvik.com.