10 minutes
Your add-on model is only as good as the inputs you start with, and CPG targets almost never show up with clean data. Blended COGS, inconsistent gross-to-net, and cohort data that describes no real customer accurately are the norm. Getting from that starting point to a combined three-statement model that holds up under exit scrutiny is the actual job, so here's how to do it.
TLDR:
Add-ons were 73% of U.S. PE buyouts in 2025: your modeling work starts before the LOI, not after.
Blended gross margin hides real profitability differences across DTC, Amazon, and wholesale channels once fees, fulfillment, returns, and marketing costs are counted.
CPG multiples run from roughly 2x SDE for small deals to 16.4x EBITDA for mega-deals; deal size and valuation basis drive the spread, and clean financials support the higher end.
Lock your base case before building scenarios; bear cases need comparable brand data, not round-number haircuts.
Iris Finance pulls actuals from QuickBooks, Xero, or NetSuite with data-integrity flagging built into the data room output.
What an Add-On Acquisition Actually Means for a CPG Finance Team
Add-on acquisitions made up 73% of U.S. PE buyouts in 2025. If you're running finance at a PE-backed CPG brand, the odds that an add-on lands on your desk are no longer low.
Most CPG finance teams treat the modeling work as something that happens after a letter of intent is signed. It doesn't. By the time an LOI is on the table, the deal thesis is already half-formed, and your job is to confirm it or kill it with numbers before the lawyers get involved.
That means building a CPG financial model, stress-testing unit economics, and pressure-checking the acquisition price against realistic exit scenarios. That's a different muscle than running a monthly close. You're modeling a future company that doesn't exist yet.
The Financial Data a CPG Target Must Have Clean Before Any Conversation
Before a model can be built, the inputs have to be real. That sounds obvious, but CPG targets routinely show up to early conversations with blended gross margins, inconsistent COGS calculation, and revenue figures that include gross sales instead of net settlements. Capstone Partners puts the 2025 consumer median at 9.2x EV/EBITDA, meaning any COGS scope error that understates true landed cost flows directly into a misstated EBITDA base and, by extension, a misstated enterprise value.
Here's what a buyer actually needs clean before any serious conversation starts:
Contribution margin by channel (DTC, Amazon, and wholesale separately, not blended), because a blended average can hide one channel destroying cash while the others look fine.
SKU-level COGS, time-bound to reflect actual landed cost changes across the period being modeled.
Gross-to-net revenue reconciliation per channel, including platform fees, chargebacks, and refunds. Without this, your revenue figure is fictional.
Cohort LTV:CAC with CAC payback periods broken out by acquisition channel.
13-week rolling cash flow with AP/AR visibility.
The LTV:CAC issue is especially costly at acquisition. A brand showing strong top-line growth with a deteriorating cohort curve is a different business than the deck suggests. That difference only surfaces if the cohort data exists and is structured correctly.
Building the Three-Statement Model for a CPG Add-On
Two brands built on different channels, different SKU architectures, and different subscription structures don't merge cleanly without a normalization step first.
Start from the target's existing GL, not a blank spreadsheet. Importing directly from QuickBooks, Xero, or NetSuite collapses the reconciliation window. A blank build forces you to reconstruct period-level actuals from scratch, which introduces transcription errors and version control problems the moment a diligence request changes an assumption.
The sequencing matters:
Normalize COGS first by time-bounding landed costs per SKU, including freight and duties, so the income statement reflects actual margin at each historical period instead of a single blended rate.
Separate revenue by channel before combining. DTC, Amazon, and wholesale behave differently on gross-to-net. Merge them too early and you lose the ability to stress-test channel mix in scenario modeling.
Align subscription economics. If one brand runs Recharge and the other doesn't, cohort structure differs. Subscription revenue carries different refund, churn, and contribution margin profiles than one-time DTC orders and needs separate line treatment before rolling into a combined income statement.
Once the income statement is defensible, the balance sheet follows inventory and AP from both entities. Cash flow ties to the combined operating model. The error rate compounds fast if COGS normalization gets skipped for speed.
Modeling Contribution Margin and Unit Economics Post-Acquisition
After closing, the first question PE operators ask isn't whether revenue grew. It's whether contribution margin per order held, compressed, or improved once the two cost structures merged.
Blended gross margin won't answer that. Contribution margin shows how far profitability actually swings across channels once you account for fees, fulfillment, returns, and marketing costs. That spread is the number you're actually underwriting in an add-on.
Model contribution margin by channel for each entity independently before combining them, since DTC, Amazon, and wholesale carry different fee structures, return rates, and CAC profiles. Stack them too early and acquisition synergies become invisible inside a blended average.
Variable | DTC (Shopify) | Amazon | Wholesale |
|---|---|---|---|
Gross-to-net deductions | Payment processing (~3%), refunds | Referral fees (8 to 15% in most CPG categories; 17% for apparel over $20), FBA fees, chargebacks | Trade spend, slotting fees, chargebacks |
Fulfillment cost driver | 3PL per-order cost | FBA per-unit fee (included in fees above) | Pallet/case shipping, retailer compliance |
CAC profile | Paid social, email (brand-direct) | Sponsored Products, tied to channel | Trade promotion, broker fees |
Return rate exposure | Moderate; brand controls policy | Higher; Amazon return policy applies | Low unit returns; high chargeback risk |
Subscription/repeat dynamics | Subscribe & Save (Recharge); brand-owned cohort | Subscribe & Save (Amazon-owned); churn tied to channel | Reorder driven by buyer, not consumer |
Contribution margin range | Higher end (brand controls cost levers) | Mid-range (fees compress margin) | Lower end (trade spend and terms erode net) |
Channel Contribution Margin Variables: DTC vs. Amazon vs. Wholesale
Synergy assumptions belong at the contribution margin line, not the EBITDA line. Shared fulfillment lowers per-order logistics cost. Overlapping audiences can reduce blended CAC if the acquiree's customer base is adjacent enough to convert on lower spend. Model those separately from one-time integration costs, which are real cash out the door and shouldn't be buried inside operating cost reduction assumptions.
Cohort LTV:CAC Analysis in an Add-On Model
Cohort LTV:CAC analysis is where most add-on models get soft. Revenue is verifiable. Customer economics are modeled, and buyers know it.
LTV:CAC is a valuation input in a diligence context, not a marketing metric. If the target's customer base deteriorates faster than the acquisition price assumes, the return model breaks regardless of what the income statement looks like at close.
Building cohort analysis across two brands requires channel separation first. Amazon, Shopify, and subscription cohorts behave differently enough that merging them produces an average that describes no customer accurately.
Why Channel Mix Changes Everything
Amazon customers repurchase on Amazon terms, with churn dynamics tied to that channel, not the brand.
Subscription cohorts carry a different churn profile and net dollar retention curve than one-time DTC buyers.
If the acquiree is subscription-heavy and the acquirer is not, that mix shift changes combined LTV meaningfully and needs separate treatment before any blended projection gets built.
Young cohorts are where the most diligence pressure appears. A brand with strong recent acquisition volume but short cohort history cannot show mature LTV curves. Projecting them on linear assumptions or by analogy to older cohorts without controlling for channel and acquisition period introduces error that compounds through the exit model. "We extrapolated from our best cohorts" will not hold under buyer scrutiny.
The subscriber-to-one-time mix question matters most when that ratio is actively shifting. Model both states: what LTV:CAC looks like if the current mix holds, and what it looks like under the post-acquisition channel strategy. The gap between those two scenarios is part of what you are acquiring.
Scenario Modeling: Bull, Base, and Bear Cases for a CPG Acquisition
Scenario ranges are only as useful as the assumptions behind them. An arbitrary 10% revenue haircut for a "bear case" tells a buyer nothing about how the business actually breaks. CPG-specific stress variables need to drive the scenarios, and each one should trace back to something the combined business can influence or has historically experienced.
What to Stress-Test
COGS and landed cost volatility, including tariff exposure and co-manufacturing transition risk if the add-on sources differently than the acquirer
Blended CAC movement post-acquisition, especially if the acquiree's growth relied on a single paid channel with limited cohort depth
Subscription opt-in rate changes, which carry outsized impact on LTV curves when a brand is actively migrating one-time buyers to subscribe-and-save
Trade spend ramp in retail if the acquisition thesis includes expanding wholesale distribution
SKU rationalization timing, since pruning underperforming SKUs improves contribution margin but can create short-term revenue drag the model needs to capture explicitly
The Sandbox Principle
Lock the base case before any variant gets built. If your base case moves every time a diligence question changes an assumption, bull and bear cases become meaningless. Scenarios are variants on a base, not rewrites of it.
Calibrate ranges against comparable brand data, not round-number percentages. A 15% CAC increase in a bear case means something if it reflects what similar brands experienced during a comparable channel disruption. Pulled from thin air, it is just a haircut with no defensible floor.
Modeling Exit Scenarios: What Drives Multiple Expansion in CPG
CPG valuation multiples run from roughly 2x to 3x SDE for small Main Street deals up to 16.4x EBITDA for mega-deals. The gap maps to specific financial variables the model either supports or fails to support.
Revenue multiples dominate when EBITDA is thin but growth and category positioning are strong. EBITDA multiples take over at scale with stable margins. Knowing which frame a buyer will apply changes what the model needs to prove.
The variables that move a brand toward the higher end:
Subscriber retention with a mature LTV curve that holds under cohort scrutiny
Channel diversification so no single channel represents a concentration risk to the buyer
Gross-to-net revenue integrity, meaning net revenue after fees, refunds, and chargebacks is what the model is built on, not gross sales
Contribution margin consistency across periods, not a single good quarter propped up by a promotional spike
A buyer discounts multiples when numbers require adjustment during diligence. Clean inputs that survive scrutiny without restatement are worth more than a compelling narrative that breaks on the first data request.
Building the Exit Readiness Financial Package
Audited financials get you to the table. They don't get you to close at the number you want.
Buyers run diligence on a specific set of materials that a standard audit doesn't produce. The package they expect includes cohort retention by vintage, contribution margin by channel, SKU-level profitability, a 13-week cash flow forecast, and a three-statement forward model with assumptions they can interrogate. Each piece needs to arrive in a format that holds up without a guided tour from your finance team.
What Goes in the Package and Why
Cohort retention by acquisition vintage, separated by channel, not blended. Buyers discount LTV projections that can't be traced back to actual cohort behavior per period.
Contribution margin by channel with gross-to-net reconciliation per source. Any blended margin figure that hasn't been decomposed will get decomposed in diligence anyway.
SKU-level profitability data showing margin after variable costs, with COGS time-stamped to reflect actual landed cost at each period. A single blended COGS rate applied across all periods will not survive scrutiny.
A three-statement forward model with locked base case assumptions, so a buyer's scenario questions don't unravel your starting point.
Pre-Empting Data Integrity Issues
The problems that renegotiate valuations are rarely fraud. They're scope errors: a connector offline for two weeks in Q3, trade spend sitting in a suspense account instead of allocated by account, COGS uploaded once and never updated after a co-manufacturer price change. Flag those gaps before the data room opens. A verified package with noted exceptions holds more credibility than a clean-looking package that breaks on the first data request. If your model changes between the management presentation and the final diligence request, buyers assume there's more movement underneath.
The data room due diligence prep is not a document repository. It is a structured presentation of the financial model's underlying assumptions, with the receipts attached.
The 13-Week Cash Flow Model in an Add-On or Exit Context
Cash flow gets more buyer scrutiny than most sellers expect. It surfaces working capital dynamics the income statement never shows: inventory purchase timing, subscription revenue deferrals, AP/AR cycles, and seasonal cash holes that look fine on an accrual P&L but create real liquidity exposure.
Forecast accuracy in 13-week cash flow modeling depends on modeling each input source separately before netting to a weekly cash position. Bank transaction data, PO flows, AP from Bill.com or NetSuite, and subscription deferral schedules all run on different cadences. POs are lumpy. Subscription cash arrives in predictable batches. AP has terms. Modeling them separately is what separates a defensible forecast from a spreadsheet that happens to balance.
Where Forecast Accuracy Actually Comes From
In our experience, an 80/20 pattern holds consistently in CPG: sales timing and inventory purchases account for roughly 80% of forecast accuracy. Those two variables are connectable, automatable, and traceable to source data. The remaining 20% covers irregular items like co-manufacturer true-ups, slotting checks, and unexpected chargeback batches. That portion requires analyst judgment, not automation, and buyers know it. Presenting a cash model without acknowledging that 20% signals overconfidence faster than any variance does.
In an exit context, the 13-week model serves a specific purpose beyond cash management. It shows a buyer the working capital requirements of the business at its current run rate, which feeds directly into purchase price adjustments and normalized working capital targets at close. A cash model that cannot be traced back to bank statements will get rebuilt in diligence at your expense.
How Iris Finance Supports CPG Acquisition and Exit Modeling
Iris Finance was founded by operators: CEO Drew Fallon was CFO of Mad Rabbit, scaling it from $1M to $40M+. She's Birdie, a $30 to $40M top-line CPG brand operated by co-founder Alex Heckmann, runs on Iris as a live production account, built after Iris Finance raised $6.2M to scale its AI-native FP&A solution for CPG brands. Every modeling capability described in this article runs on real operating data, not a demo dataset.
For PE-backed brands and founder-led companies in the $10M to $500M range, the practical workflow looks like this: import your existing Excel model or build from your chart of accounts in QuickBooks, Xero, or NetSuite. Actuals roll forward after month-close in roughly 15 seconds. Iris builds the three-statement model for each entity, but it does not produce a consolidated multi-entity P&L rollup, so the combined add-on view gets assembled from each entity's model by your team and your dedicated analyst, with Excel export available anytime. No manual reentry, and the diligence package and the working model draw from the same data.
How Scenario Ranges Get Calibrated
Scenario ranges are built as Bollinger-band ranges from anonymized data across roughly 500 brands and about $20B in GMV on Iris Finance, the AI-native FP&A solution behind these models. Your bear case reflects what similar brands have actually experienced in comparable conditions, not a round-number haircut someone invented in a spreadsheet.
Data rooms are assembled directly from reconciled Iris data, with data-integrity flagging built in. If a connector was offline for two weeks, that period is shaded and noted in the output instead of silently included. Buyers notice gaps. Flagging them yourself is the better move.
The dedicated analyst layer covers the judgment calls automation will not touch, so your model arrives at the data room with both the receipts and someone who can explain them.
Final Thoughts on Modeling CPG Acquisitions and Exit Scenarios
A deal that renegotiates during diligence almost always traces back to a data problem that was visible before the data room opened. Channel-separated margins, clean cohort history, and a base case that doesn't move every time a new assumption comes in are the inputs that actually protect your exit multiple. Building that foundation takes time, but it's the work that holds the number. If you're getting close to an acquisition or exit conversation, talk to the Iris Finance team before the first diligence request lands.
FAQ
How do PE firms benchmark portfolio CPG brands against peers on retention, CAC, and contribution margin?
PE firms typically benchmark portfolio brands using cohort-level data: subscriber retention by acquisition vintage, CAC payback by channel, and contribution margin per order against category-comparable brands. The challenge is that most CPG brands don't have this data structured at the channel level, so benchmarks get compared against blended averages that describe no single customer accurately. Iris Finance runs anonymized benchmarking against comparable brand data (see the scenario modeling section above), giving PE-backed brands a category-level view of where their retention and CAC payback actually sit relative to similar businesses.
How do I track cohort LTV by channel on Amazon and Shopify without building a custom data warehouse?
You don't need a custom data warehouse. Purpose-built CPG finance tools connect directly to Amazon and Shopify and structure cohort data at the channel level without requiring a data engineer or SQL infrastructure. The bigger issue is that Amazon and Shopify cohorts behave differently enough that merging them produces an average that describes no customer accurately: Amazon repurchase dynamics are tied to that channel, while Shopify DTC cohorts reflect brand-direct behavior. Separating them before any blended projection is what makes the LTV analysis defensible.
What's the right way to model add-on acquisitions and exit scenarios for a CPG brand when the target has a shifting subscriber-to-one-time buyer mix?
Model both states explicitly: LTV:CAC under the current subscriber mix, and LTV:CAC under the post-acquisition channel strategy you're underwriting. A rising subscription opt-in rate changes combined LTV meaningfully, and projecting a blended curve without separating those two populations introduces compounding error through the exit model. The gap between those two scenarios is part of what you're acquiring, and buyers will reconstruct it in diligence if you don't present it first.
How does building an exit readiness financial package for a CPG brand differ from standard audited financials?
Audited financials verify historical accuracy, but they don't produce the materials buyers actually run diligence on. A buyer expects cohort retention by acquisition vintage separated by channel, contribution margin with gross-to-net reconciliation per source, SKU-level profitability with time-stamped COGS, and a three-statement forward model with locked base case assumptions. Gaps that renegotiate valuations are rarely fraud; they're scope errors like a connector offline for two weeks or COGS uploaded once and never updated after a co-manufacturer price change. Flagging those before the data room opens holds more credibility than a clean-looking package that breaks on the first data request.
How do I model the impact of a tariff increase on a CPG brand's gross margin and exit multiple?
Build the tariff exposure at the SKU level first: landed cost per unit changes by sourcing origin and product category, so a single blended COGS adjustment will misstate the actual margin impact across your SKU mix. Then run it through contribution margin by channel, since the same COGS increase hits DTC, Amazon, and wholesale differently once you account for fee structures and return rates. At the exit model level, a persistent COGS increase that the brand cannot offset through pricing or supplier diversification compresses contribution margin consistency across periods, which is one of the specific variables buyers discount when setting multiples.
Related posts
Actionable data, proven strategies, and clear forecasts, empowering you to make smarter decisions and scale with confidence.





