Why scaling spend changes the calculator's inputs
The default page imagines a budget built from nothing: a conversion target, an expected cost for each one and a total that funds the whole plan. Scaling begins elsewhere. The channel already spends and delivers, and the question is what more money would add. Entering the full new target at the account's current CPACost of each conversion. answers a different question, and too generously.
The guide to marginal returns and forecasting explains why. Delivery finds the readiest buyers first, so each further share of budget reaches people less inclined to act, the pattern called Diminishing ReturnsEach extra unit of spend or effort buys less extra result than the one before it.. The average blends cheap early results with dearer recent ones, so an increase sized on it promises conversions the money will not buy.
Two changes follow. Current spend and its results become a baseline the calculator leaves alone, and only the increment goes in. The cost entered is the marginal figure, which the guide estimates without a model from periods where spend changed or from planned budget steps, and the answer is judged against the target cost from contribution margin, derived in the contribution margin and break-even guide, rather than against the channel's own average. A marginal CPAThe extra spend divided by the extra acquisitions it produced when spend changed; usually above the average, and the figure scaling and cutting decisions need. measured on the last step describes that step only, so it is a starting point for pricing the next one, not its price.
The inputs for scaling spend
| Calculator input | On the default page | For scaling spend | Where to find it |
|---|---|---|---|
| Conversions needed | The full number of conversions the plan has to produce | Extra conversions wanted above what current spend delivers, with current spend held as the baseline | Current monthly conversions and the new target |
| Target CPA | The cost per conversion the plan aims for | Marginal cost per acquisition from a recent budget step or comparable periods: extra spend ÷ extra conversions | Spend and conversions before and after the last step, in a similar season |
| Testing reserve | A share set aside for new ideas | A cushion for the range around the marginal estimate | How widely marginal estimates have varied across past steps |
| Months | The planning period | The length of the next step, held long enough to read before another | The review point agreed before scaling |
The cost row needs deriving with care. Take two stretches of similar trading either side of the last budget change, with no promotion, price change or new creative between them, and divide the difference in spend by the difference in conversions. That is what the last step paid per extra result. The next step sits further along the curve, so enter that figure raised for the move ahead. Several past steps show how quickly the price climbs and how much the estimates scatter; the scatter is what the reserve covers.
Entering the scaling inputs step by step
Write down the baseline and the target from margin
Record current monthly spend and conversions over a settled period, with the highest price per extra conversion the margin allows beside them. Neither goes into the calculator; they frame the increment.
Price the last step at the margin
Divide the extra spend of the last comparable budget change by the extra conversions it produced, raise the result for the step ahead and enter it as the target CPA. If it already exceeds the margin-based target, stop here.
Enter the increment, the cushion and the step length
Enter only the conversions wanted above the baseline. Size the reserve from the scatter of past marginal estimates, and set the months to this one step, ending on the date it will be read.
Fix the stop rules and the counting method first
Agree the marginal cost that reverses the step, the signals that pause it and how extra results will be counted, then log the step and its expected effect in the marketing decision log template.
A worked scaling step with illustrative inputs
Maezane Tutoring sizes its next paid search step
Maezane Tutoring, an invented online tutoring service, spends £18,000 a month on paid search and records 360 sign-ups, an average of £50. Its last step, from £15,000 a month in a comparable term-time period, added 45 sign-ups for the extra £3,000, a marginal cost of about £67. Its target cost per sign-up, set from contribution margin, is £90.
It wants 150 more sign-ups over the next two months. The next step sits further along the curve, so it assumes a marginal £82, and because past marginal estimates have scattered by about a fifth either way it sets a 20% reserve. The calculator returns £14,760 in total, £7,380 a month and £2,460 in reserve: £25,380 a month in all, the extra money expected to add about 75 sign-ups a month to the baseline 360.
The team agrees to step back if the marginal cost passes £90, frequency climbs or the share of new learners falls, and holds matched regions at the current level. After two months, £13,050 of extra spend had gone out. The platform credited 171 extra sign-ups, about £76 each; against the control regions, enrolment records showed about 138, a marginal cost near £95. The account average was still about £57. Maezane steps back to half the increase and briefs new creative for a wider audience before the next step.
Marketing Budget Calculator
Open the Marketing Budget Calculator →Prefilled with the example's inputs: Conversions needed 150 · Target CPA $82 · Testing reserve 20% · Months 2 → Total budget $14,760 · Monthly budget $7,380 · Testing reserve amount $2,460.
All figures above belong to the invented tutoring service. Put the account average in the embedded calculator's cost field to see how much smaller an increase planned on averages looks.
Interpreting the increment and its monthly pace
Read the total as an increment: money added on top of current spend, buying only the extra conversions entered. Adding the two gives the new monthly level finance will see, but the calculator's figure alone is never the channel's budget. The marginal returns guide's scenario table keeps them apart too: the held baseline is the confident line, and the increase carries a range.
Next, set the cost entered beside the target from contribution margin. The marginal figure has to stay below it for the step to add profit, and the account average will look healthy long after that stops, because every cheap early conversion is still inside it, as the example's average showed. Teams that think in returns apply the same test to Marginal ROASExtra revenue returned by the last increment of ad spend., the return on the next share of spend, against the margin-based floor.
Then treat a large increase as a sequence. The guide's forecasting method breaks a big rise into a series of steps and prices each one dearer than the last, with gains tapering as the channel nears saturation, so run the calculator once per step with a rising cost.
Read the monthly budget as the pace of the step, not an allowance. The guide to testing and scaling wants three things settled before any increase: the marginal cost that ends it, the signals that pause it and the date it is judged. Its typical signals include marginal cost above target, frequency rising steadily, falling new-customer share and a blended cost in the business's own records that drifts away from what the platform reports. The months input ends on that date. The reserve amount lets the step reach its goal if the cost lands at the dear end of its range; drawing on it is a prompt to check the stop rules.
Finally, the calculator cannot tell whether the extra conversions were caused; at the margin many people credited by a platform would have converted anyway. The budget increase incrementality test raises spend in matched test regions while control regions stay put, counts conversions in the business's own records and divides the extra spend by the incremental conversions. That figure decides whether the step stays. If spend has already risen without results, the diagnosis of ad spend up but sales flat separates saturation from the other causes. Sibling pages run the same calculator for a seasonal peak and a product launch, and the CPA calculator checks a step's marginal cost from its extra spend and extra conversions.
Pitfalls when scaling spend numbers go into the calculator
- Entering the account's average CPA. It contains every cheap early conversion the increase will never buy again, so the total is too small for the results promised.
- Entering the whole target instead of the increment. The full goal prices conversions current spend already delivers at the marginal rate, and the total stops meaning anything. Enter only the extra.
- Scaling in one jump rather than in steps. A single large rise skips the readings that show where the curve bends, and can unsettle automated delivery.
- Reading a step taken in a different season. SeasonalityPredictable, repeating changes in demand tied to the calendar, such as months, holidays, weekdays or hours. moves results whether or not spend changed: a rise into a busy month looks cheap and one into a quiet month looks dear. Compare like periods.
- Trusting platform-credited conversions for the extra results. Platforms claim what they touched, not what they caused; count the step in the business's own records against a holdout, which is the point of an IncrementalityConversions caused by marketing. test.
