Revenue that carries margin
Every growth decision on this list raises reported revenue. Four of them raise contribution margin. They are not the same four most teams prioritise.
4 of 8
Eight common B2C growth decisions all raise reported revenue. Four of them raise contribution margin.
Key takeaways
- 01Of eight common growth decisions, all eight raise reported revenue and only four raise contribution margin.
- 02The four that carry margin are plan upgrades, contact-pressure reduction, incentive removal on high-propensity buyers, and friction fixes.
- 03Cross-sell is the most misclassified decision: it raises margin only when the second product's margin beats the spend it substitutes.
- 04Contact-pressure reduction reads negative in a four-week window and positive over a long one, so the window has to be fixed before the test starts.
- 05Incentive removal is the cheapest untested experiment on most roadmaps, because it makes a reported number smaller.
Romà Llambés · CEO, Markin · Reviewed 2026-09-10 · Every report is reviewed twice a year, and re-run out of cycle whenever a source it rests on changes. · Method and sources
The question
Which growth decisions raise revenue, and which of them also raise margin?
Upgrades, contact-pressure reduction, incentive removal on high-propensity buyers and friction fixes raise revenue and margin together. Discount-led acquisition, deep save offers, cross-sell into a lower-margin product and bundle-led upgrades usually raise revenue while flattening or shrinking margin, because the delivered cost moves with the revenue.
- Sample
- Eight decision types common to B2C subscription and commerce businesses. Each is evaluated with the same margin ledger: incremental revenue, variable delivery cost, incentive cost and substitution effect, all per treated customer against a randomised holdout. Margin ranges by product category are taken from published gross-margin disclosures of listed consumer businesses across streaming, telecom, marketplaces and subscription retail.
- Period
- Gross-margin disclosures read for financial years ending in 2025 and the first half of 2026.
- Reviewed
- 2026-09-10Reviewed twice a year
- 4 of 8
- Four of the eight decisions raise revenue and margin together: plan upgrades, contact-pressure reduction, incentive removal on high-propensity buyers, and friction fixes.
Findings
What we found
- 014 of 8Derived in this report
Four of the eight decisions raise revenue and margin together: plan upgrades, contact-pressure reduction, incentive removal on high-propensity buyers, and friction fixes.
What they have in common is that the incremental revenue arrives with little or no incremental delivery cost and no incentive. Two of them, contact-pressure reduction and incentive removal, raise margin by doing less rather than more, which is why they rarely appear on a growth roadmap despite being the cheapest wins available.
Do thisPut contact-pressure reduction and incentive removal on the roadmap explicitly. Nobody proposes them, because they consist of doing less.
- 02Derived in this report
Cross-sell is the decision most often misclassified. It raises revenue reliably and raises margin only when the second product's margin exceeds the substituted spend on the first.
In practice that means cross-sell into a higher-margin product is a margin decision and cross-sell into a lower-margin one is a revenue decision wearing a margin costume. Product-line dashboards cannot tell them apart, because each line reports its own growth truthfully while the customer's total spend has barely moved.
Do thisSplit cross-sell reporting by the second product's margin before you scale any cross-sell programme.
- 03Derived in this report
Contact-pressure reduction is the only decision on the list where the expected revenue effect is negative in the short window and positive in the long one.
Sending less lowers immediate response and raises deliverability, list health and long-run engagement. It is therefore the decision most likely to be killed by a four-week read, and the one that most needs a long measurement window fixed before the experiment starts rather than after the first result arrives.
Do thisFix the measurement window for a contact-pressure test before it starts, and make it long enough to catch the reversal.
- 04Derived in this report
Incentive removal is systematically under-tested. Where a discount is applied to a population that includes high-propensity buyers, withholding it from a randomised slice measures the giveaway directly.
The experiment is cheap, the downside is bounded and the result is immediately actionable, which is a rare combination. It is under-run because the framing feels like a cut rather than a growth initiative, and because it makes a previously reported number smaller.
Do thisWithhold the incentive from a randomised slice of high-propensity buyers next quarter. It is the cheapest experiment on the list.
Charts and tables
The data
- Revenue, higher-margin product$14.00
- Contribution margin, higher-margin product$6.65
- Revenue, lower-margin product$14.00
- Contribution margin, lower-margin product$0.50
Delivery cost, incentive and substitution against the primary product are all inside the margin figure and none of them are inside the revenue figure.
| Decision | Incremental revenue | Delivery cost | Incentive | Substitution | Effect on contribution margin |
|---|---|---|---|---|---|
| Plan or tier upgrade | Up | Low | None or small | None | Up |
| Friction fix in the paid journey | Up | None | None | None | Up |
| Incentive removal on high-propensity buyers | Flat to slightly down | None | Removed | None | Up |
| Contact-pressure reduction | Down short, up long | Down | None | None | Up over a long window |
| Cross-sell into a higher-margin product | Up | Medium | Small | Some | Up |
| Cross-sell into a lower-margin product | Up | High | Small | High | Flat to down |
| Deep save offer, untargeted | Up in reported retention | Low | High | None | Down |
| Discount-led acquisition | Up | Medium | High | Some | Down until the second renewal |
| Line | Higher-margin second product | Lower-margin second product |
|---|---|---|
| Incremental revenue | $14.00 | $14.00 |
| Variable delivery cost | -$4.20 | -$8.40 |
| Incentive | -$1.20 | -$1.20 |
| Substitution against the primary product | -$1.95 | -$3.90 |
| Contribution margin | $6.65 | $0.50 |
Run it on your numbers
Run one decision through the margin ledger
Incremental revenue minus delivery cost, incentive and substitution, per treated customer against a holdout. The same ledger as the tables above, on your numbers.
Contribution margin per treated customer
$6.65
Measure each line against a randomised holdout. Substitution is invisible without one.
Margin rate on the incremental revenue: 47.5%
Implications
What this means for the person who has to act
Growth lead
Re-rank the roadmap by contribution margin per treated customer. Two of the four margin-positive decisions, contact-pressure reduction and incentive removal, consist of doing less, so they will not appear unless someone puts them on the list deliberately.
Finance
Supply a margin per action, even as a band. Where it is missing, decisions are ranked on revenue and drift towards discounts and low-margin products while every product-line dashboard still reports growth.
Lifecycle marketing
Split cross-sell reporting by the second product's margin. The same offer into a higher-margin and a lower-margin product produces an identical revenue line and a thirteen-fold difference in contribution margin in the illustrative ledger.
Method
Definitions that change the answer
- Contribution margin
- Incremental revenue minus the variable cost of delivering it, minus the incentive given to obtain it. Fixed platform and headcount cost is excluded because it does not move with the decision.
- Substitution
- Revenue the customer would have spent elsewhere in your own catalogue. It is a cost of the decision and belongs inside the ledger, not in a footnote.
- Revenue-positive, margin-negative
- A decision whose incremental revenue is real and whose contribution margin is not, once delivery cost, incentive and substitution are subtracted. The most common failure mode in growth reporting, because the first number is easy to see and the other three are not.
Estimator
The model, written out
Contribution margin per treated customer for a single growth decision, measured against a randomised holdout over a fixed window. Not revenue, not conversion, and not the product line's own growth.
Incremental revenue
dR = mean(revenue | treated) - mean(revenue | control)
Total customer revenue, not the revenue of the product the decision promoted. That distinction is where substitution hides.
Contribution margin of the decision
CM = dR x m - I - S
m is the finance-supplied margin on the incremental revenue, I the incentive paid per treated customer, S the margin lost to spend the customer moved from elsewhere in your catalogue.
Cross-sell condition
CM > 0 <=> m2 > sigma x m1
Second product margin m2, first product margin m1, sigma the share of the second product's revenue substituted from the first. The whole cross-sell argument reduces to this inequality.
Classification rule
margin-positive <=> lower bound of CM > 0
A decision is classified margin-positive only when the interval excludes zero, not when the point estimate is positive.
What has to hold
- Randomisation at customer level, so substitution inside one customer's basket is captured by the revenue difference rather than needing a separate estimate.
- Margin per product supplied by finance and held constant over the measurement window.
- One decision per treated customer in the window. Overlapping experiments break the ledger before they break the statistics.
- The window is long enough to contain the cost tail: returns, support contacts and downgrades triggered by the decision.
What breaks it
- Substitution outside the measured catalogue, for example a cross-sell that cannibalises an offline purchase you cannot see.
- Product-line attribution instead of customer-level measurement. Every line then reports honest growth while total spend has not moved.
- A window shorter than the cost tail, which makes incentive-heavy decisions look margin-positive for about a quarter.
- Margin taken as a category average rather than product level. Mix inside a category varies more than the gaps between categories.
Method
How to reproduce this
- 01
For each decision you run, write down four numbers per treated customer against the holdout: incremental revenue, variable delivery cost, incentive cost and substitution.
- 02
Use your own finance-supplied margin per product. Category averages are for orientation only; product mix inside a category varies more than the category average does.
- 03
Subtract the last three from the first. That is the decision's contribution margin.
- 04
Rank decisions by contribution margin per treated customer, not by incremental revenue and not by conversion rate.
- 05
Re-rank quarterly. Margin mix moves with pricing and cost, and a decision that cleared last year may not now.
Uncertainty
How wrong each figure could be
The classification is a sign test on a ledger, so the honest statement is not a confidence interval on 4 but the range of inputs over which the count holds. Two of the eight decisions are near the line and move with one input each.
4 of 8 decisions carry margin
Three when cross-sell runs into a lower-margin product and contact-pressure is read on a four-week window; five when the cross-sell target has the higher margin.
3 to 5 of 8
Cross-sell threshold
Across substitution shares sigma of 20% to 70%, which is the range we can defend for consumer catalogues with overlapping products.
m2 > 0.2 x m1 to m2 > 0.7 x m1
Contact-pressure reduction
The sign of this decision is a function of the window, not of the treatment. Fixing the window before the test is the only defence.
negative at 4 weeks, positive at 12 weeks or more
Category margin ranges
Public disclosures are consolidated, so they orient the ledger and must never be the input to a live decision.
blended across product lines
Robustness
What survives a different assumption
Each row changes one input and re-runs the classification across all eight decisions.
| Change one assumption | What changes | Result | Conclusion |
|---|---|---|---|
| Substitution set to zero | S = 0 for every decision | Six of eight carry margin. Cross-sell and bundle-led upgrade both flip positive. | Flips |
| Measurement window cut to four weeks | Cost tail and long-run effects excluded | Contact-pressure reduction flips negative; incentive-led decisions flip positive. | Flips |
| Category margin replaced by product-level margin | Finance-supplied m per product | Count unchanged at four. Two decisions change rank. | Holds |
| Incentive cost halved | I x 0.5 | Count unchanged. Deep save offers remain negative. | Holds |
| Measured on revenue rather than contribution margin | CM replaced by dR | Eight of eight pass. This is the failure mode the report exists to name. | Flips |
The count of four is not the durable result; the ledger is. Three of the five scenarios move the count, and every one of them moves it by dropping a term from CM rather than by disagreeing about a number. A reader who keeps all four terms will reach a defensible classification on their own business even where it is not ours.
Limits
What this cannot tell you
- The classification is directional. A decision that is margin-negative in one business can be margin-positive in another with a different cost structure, and the ledger, not the label, is what settles it.
- Category margin ranges from public disclosure are blended across product lines, so they are a starting point rather than an input to a real decision.
- The ledger measures the decision, not the strategy. A margin-negative decision can still be right where it buys a market position, provided that is the stated reason rather than an accident of reporting.
- Second-order effects, brand damage from over-discounting or support load from a poorly matched cross-sell, are named in the table but not quantified here.
Falsifiability
What would change our mind
- 01
A holdout-measured cross-sell into a lower-margin product that clears contribution margin after substitution. That would move the row from margin-negative to conditional.
- 02
Long-window evidence that contact-pressure reduction stays revenue-negative past two quarters. The margin case would then rest on delivery cost alone.
- 03
A discount-led acquisition cohort that reaches parity with organic contribution margin before the second renewal.
Provenance
Sources, and what each one was used for
- Public company disclosureGross-margin disclosures, listed consumer businesses (FY2025 to H1 2026)
Category margin ranges used to orient the ledger. Not used to produce any per-decision figure.
- Derived in this reportMargin ledger
Both tables. The four-column structure is the reproducible part.
- Independent researchBCG, Measuring Incrementality in Next-Best Action Programs (July 2026)
The requirement that each line be read against a randomised holdout rather than against the previous period.
Questions this report answers
- Which growth decisions actually improve margin?
- Does cross-sell increase profit or just revenue?
- How do I measure contribution margin on a marketing decision?
- Why does revenue grow while margin stays flat?
- Should we send fewer marketing messages?
Frequently asked
- Which growth decisions raise margin as well as revenue?
- Plan upgrades, friction fixes in the paid journey, removing incentives from high-propensity buyers, and reducing contact pressure. All four add revenue with little or no incremental delivery cost and no incentive, which is what separates them from the rest.
- Why is cross-sell classified two ways?
- Because its margin effect depends entirely on the second product. Into a higher-margin product it adds contribution margin; into a lower-margin one the delivery cost and the substituted spend on the primary product consume most of the gain, while both product lines still report growth.
- How do I put substitution in the ledger?
- Measure the treated group's spend on the primary product against the holdout's, over the same window. The difference is substitution. Without a holdout it is invisible, which is why cross-sell programmes so often report success while total customer spend is flat.
- Is contact-pressure reduction really a growth decision?
- It raises contribution margin by lowering delivery cost and protecting deliverability, and over a long window it usually raises revenue too. It reads as a cut in a four-week window, which is why the measurement window has to be fixed before the experiment rather than after the first result.
Act on this
- PlaybookCross-sell a second product without cannibalising the plan the customer already pays forSecond-product offers valued on incremental margin, not on attach rate.
- PlaybookAdd next best action to journeys you already run in BrazePer-customer arbitration written into the canvases your team already operates.
- ResearchThe retention discount lineMost save offers are evaluated on how many customers stayed. The only number that decides is how many stayed who would otherwise have left.
- ResearchARPU and churn, measured consistentlyMost cross-sector ARPU comparisons are comparing denominators. Once the definitions are held constant, the level differences shrink and the growth spreads do not.
- GuideHow to increase ARPU for B2C companies in 2026Five ARPU levers ranked by durability, the diagnostic to run first, and why sustained ARPU growth is a decision throughput problem rather than an ideas problem.
- GuideCustomer action arbitration: choosing one action per customerHow competing actions are resolved per customer: eligibility, expected value, contact policy, conflicts between teams, and when the right answer is no action.