Why a fixed monthly number fights the economics
A performance target is a price: the most a new customer or a qualified lead may cost. The marginal returns guide gives the economic answer to how much to spend at that price: spend should rise until the marginal cost per acquisition reaches the target, because below that point each extra amount adds profit and above it each extra amount reduces it. The cross-channel planning guide applies the same rule to demand capture: fund it close to its natural ceiling and stop when the next result costs more than the target allows.
A fixed monthly number ignores both. In a strong month, when demand is high and the next result is still cheap, the fixed line stops spend that would have paid. In a weak month, when the next result costs more than the target, the fixed line pushes the team to spend it anyway, because underspending looks like failure. Neither month is managed on the economics. The budget follows the calendar, and the target becomes a reporting line rather than a purchasing rule.
The answer is not to remove the budget. Finance needs to know the range of spend it is committing to, the cash it needs and when the money will move. The answer is an agreement that both sides sign: the rule spend follows, the limits it follows it within, and the way variances are reported. The testing and scaling guide says that guardrails should be agreed before scaling; this tutorial writes them down as the budget itself.
Before you start
- A target derived from the business's economics. A cost per acquisition or return target built from margin, retention and cash, as the performance economics guide explains. A borrowed or inherited target cannot carry an agreement.
- The approved annual or quarterly budget. The budget-building tutorial separates fixed and variable lines; the agreement changes how the variable paid media lines behave, not the rest.
- A payback figure by channel. From cohort data, on margin, as the retention and payback guide calculates it.
- Evidence of marginal cost. At least one budget step per main channel, or periods where spend changed and conditions were otherwise similar.
- New customers recorded in the business's own data. The stopping rule is read on results the business can verify, not on platform-reported conversions alone.
- A finance business partnerThe person in finance who looks after marketing's numbers and agrees the codes, cut-off dates, accruals, reforecasts and approvals that marketing works to. who will sign the agreement and attend the month-end close.
- A spreadsheet or the marketing budget calculator to model the ceiling, floor and expected range.
The steps
Agree the target and record where it came from
Write the target into the agreement with its derivation: the margin, the retention evidence and the payback that produced it, and the date it was set. State whether it is a cost per new customer, a cost per qualified lead or a return on ad spend, and that it is read in the business's own records. A target with a recorded source can be revisited when an input changes; one without a source gets argued about every month. If growth is the priority for a period and the business will accept a higher marginal cost, the marginal returns guide says that trade-off must be explicit and agreed with finance, so write it here, with its end date.
Write the stopping rule on marginal cost
The stopping rule is the marginal cost rule: each flexible line may keep spending while the estimated marginal cost per result is at or below the target, and stops increasing when it is not. Say how the marginal cost is estimated. The testing and scaling guide describes budget steps: raise the line by a step, hold it, read the extra results against the extra spend, and decide. The marginal returns guide adds that every estimate has noise, so the rule should ask for a sustained reading over an agreed period rather than reacting to a single week. The CPA calculator turns a budget step and its extra results into a marginalExtra revenue returned by the last increment of ad spend. cost. Record what happens when the rule is breached: the line returns to the last level at which the marginal cost met the target.
Set the ceiling from cash and payback
The ceiling is the most the business will spend in a month, however good the results look. It comes from two limits. Cash: every customer acquired is an outlay now against margin that arrives later, which the economics guide describes as a loan the business makes to itself. Payback: the retention and payback guide sets a Payback PeriodMonths of customer margin needed to recover the cost of acquiring that customer. ceiling per channel, agreed with finance and derived from the cash position. Together they say how much acquisition spend the business can fund before the margin from earlier customers returns. Write the ceiling per month and, where channels have different payback, per channel. Above it, spend needs a separate approval, however good the marginal cost.
Set the floor that protects tests and continuity
The floor is the least each protected line will spend, even in a weak month. It exists for two reasons. Tests need their budget to reach an answer, and the planning guide warns that a plan without a protected test reserve can never discover a better plan. And campaigns on automated bidding lose what they have learned if they are paused or cut sharply: the Meta learning phase guide lists large budget changes and extended pauses among the edits that restart learning, and the Smart Bidding guide lists large changes to targets or budgets among the triggers for Google's learning period. Set the floor from the test plan and from the lowest level at which each core campaign has run steadily before, not from a share of the ceiling.
Decide which lines may flex and which stay fixed
Not every line can follow results. Mark each paid media line as flexible, protected or fixed.
Line type Behaviour under the agreement Why Demand capture with marginal evidence (search, Shopping) Flexes between floor and ceiling on the stopping rule Volume follows demand; the marginal result is usually the cheapest in the plan Proven demand creation with marginal evidence Flexes, in budget steps The economics support more spend only as far as the steps show Tests and new channels Protected at their planned budget The answer matters more than the month's efficiency Contracted media, retainers, fees and production Fixed Committed once signed; the budget-building tutorial treats them as fixed costs Remarketing and re-engagement Capped by audience, not flexed The planning guide sizes them by the audience available A line without marginal evidence does not flex until it has some. That rule keeps the agreement honest: money moves towards proof, not towards whichever platform reports the best average.
Name the review points and who may move money
Write down when spend is reviewed and who may change it. The performance reporting guide sets a daily, weekly and monthly cadence: daily checks for breakage, weekly optimisation inside channels, and a monthly review where budget moves between channels. Use the same cadence here. Name the person who may move money within a channel between floor and ceiling, the person who may move it between channels, and the move that needs finance's approval. The tracking spend against budget guide lists the reallocation rules finance usually wants agreed in advance; record each move in a budget reallocation memo so the evidence and approval sit together.
Set platform budgets so spend can follow the rule
Platform budgets are the finer layer beneath the agreement, and they must not quietly cap or force spend. Google documents a campaign budget as an average daily amount: the budgets and pacing guide explains, from Google's help page on average daily budgets, that spend may run above the daily figure on busy days and below it on others, within a documented monthly limit. Set the daily average that implements the month's position between floor and ceiling, and watch the month-to-date total rather than single days. On campaigns bidding to a target, the Smart Bidding guide notes that a campaign held back by its target will not spend more if the budget is raised, while one limited by budget will, at a rising marginal cost. That is the stopping rule working inside the platform. On Meta, follow the budget types guide: daily budgets are averages over the week, and large budget changes can be treated as significant edits, so change budgets in steps.
Report spend against the rule with variance explained
Report each flexible line against the rule rather than against a fixed monthly figure: spend, results in the business's records, estimated marginal cost against target, and position between floor and ceiling. A line that spent above the original plan because the marginal cost stayed under target is following the agreement; a line that spent above the plan while its marginal cost was over target is a breach. Explain each material variance the way the tracking spend guide asks: the cause, whether it is timing or permanent, and what happens next. The monthly paid media review checklist covers the pacing and returns checks that feed this report.
Close each month with finance and reforecast
At month-end, reconcile platform spend with finance's ledgerFinance's record of what the company spent and earned, coded by cost centre and cost type; when marketing's figures disagree with it, the ledger wins., raise accruals for media delivered but not yet invoiced, and explain the variances, as the tracking spend guide describes and the budget month-end close checklist lists. Then reforecast the rest of the quarter as a range: the spend expected if the marginal cost stays where it is, and the spend at floor and ceiling. The scenario planning framework is a useful shape for that range, because finance plans cash against cases rather than a single number. At the company's formal reforecastA rebuild of the full-year view of the budget from actual spend and what marketing now knows, done at the points finance sets; not a fresh budget. points, these quarterly ranges feed the full-year marketing reforecast as the paid media lines' latest estimate.
Review the agreement each quarter
Once a quarter, review the agreement itself rather than the month's results. Has the target's source changed: margin, retention, payback? Has the cash position moved the ceiling? Did any line breach the stopping rule, and why? Did the floor protect the tests it was meant to protect? Has any line earned the marginal evidence to move from fixed to flexible? Record the changes, have finance sign the revised version, and keep the old one for comparison.
A home fitness retailer's budget agreement (illustrative)
Pellavoy Home Fitness replaces a fixed monthly budget
Pellavoy Home Fitness sells rowing machines, adjustable dumbbells and folding benches online. Its paid media budget was a fixed £40,000 a month: £22,000 on Google Search and Shopping, £14,000 on Meta prospecting and £4,000 on remarketing. Demand peaks in January and falls away in early summer, so the business routinely ran out of budget in January and pushed spend into June to use it.
The marketing lead and the finance business partner write an agreement. The target is £95 per new customer: the first order earns £70 of contribution margin, and two years of cohort data show repeat purchases recovering the rest within four months. That payback limit, with the cash position, sets a ceiling of £60,000 a month. The floor is £24,000: £3,000 for a protected Meta test, the £4,000 remarketing cap, and £17,000, the lowest level at which Search and the proven Meta prospecting ad sets have run steadily. Search and proven Meta prospecting flex on the stopping rule; remarketing stays capped.
| Row label | January | June |
|---|---|---|
| Spend on flexible lines | £47,000 | £17,000 |
| Protected test and remarketing | £7,000 | £7,000 |
| Total against old fixed £40,000 | £54,000 | £24,000 |
| New customers in the shop's records | 640 | 230 |
| Average cost per new customer | £84 | £104 |
| Estimated marginal cost at last step | £93 | £118, so held at the floor |
In January, budget steps on Search and Meta prospecting kept the estimated marginal cost under £95 up to £54,000, so spend ran £14,000 above the old fixed line and inside the ceiling. In June, the first step showed a marginal cost of £118, so the flexible lines fell back to the floor; the £16,000 not spent was reported as a permanent variance and the quarter's reforecast was lowered. Across the two months, the agreement spent £78,000 against the £80,000 the fixed budget would have spent, and bought 870 new customers in the shop's records.
Common mistakes
- Calling it uncapped. An agreement without a ceiling from cash and payback is not one finance can sign; the ceiling is what makes the flexibility safe.
- Flexing on average cost. The average stays under target long after the marginal result has passed it; the stopping rule is on marginal cost.
- No floor. In a weak month, cutting every line to nothing ends the tests and resets the learning of the core campaigns, so the next strong month starts from scratch.
- Flexing lines without evidence. A line with no budget steps behind it has no marginal cost to follow; keep it fixed until it earns one.
- Platform budgets that contradict the agreement. A daily cap set to the old fixed line stops the rule from working; a budget raised on a target-limited campaign does nothing.
- Reporting against the old fixed number. A strong month then looks like an overspend and a weak one like thrift; report against the rule.
- Never revisiting the agreement. Margin, retention and cash change; a target or ceiling set a year ago can be wrong in either direction.
