Seasonal Ad Budget Plan · October 2026

Halberton Hampers Ltd

Food and drink hampers made in Devon and sold online, for gifts and corporate orders. 38 staff.

Worked example. Halberton Hampers and its figures are fictional. All figures come from the worked-example spreadsheet: ALL CHECKS PASSED (with 4 notes).
The headline

Same £72,000, spent when people buy: £20,700 more a year, but the total may be too high

Suggested split
£20,700
a year more, from 3,308 customers
Owner’s plan
£13,018
a year more; £6,509 if months respond half as differently
Per customer
£25
of gross profit, counting only the truly extra share
Cheapest month
£10
per customer in November, against £75 in July

Verdict: Ad spend: £72,000 last year; £72,000 next. Last year’s even spread brought 2,480 new customers, from £10 each in November to £75 in July. The planned split should bring 3,001, changing the net by £13,018 (£6,509 if the months respond half as differently, £6,509 if only half as many are truly extra); the suggested split, same total, 3,308 customers and £20,700. In 2 months the spend would bring more customers than you could serve: they’re capped. In 12 months the last £100 of the plan loses money: consider a smaller total.

Month by month

Last year, the plan, and the suggested split

JanFebMarAprMayJunJulAugSepOctNovDec

Light: last year’s spend. Mid: the owner’s plan. Dark: the suggested split (same total, within what can be served and no more than three times last year’s spend).

MonthDemandLast yearPer customerPlanSuggested
January20£6,000
110 cust.
£55£3,000
66 cust.
£2,153
51 cust.
February25£6,000
140 cust.
£43£3,000
83 cust.
£4,151
106 cust.
March30£6,000
170 cust.
£35£6,000
171 cust.
£8,123
215 cust.
April20£6,000
110 cust.
£55£3,000
66 cust.
£2,153
51 cust.
May25£6,000
140 cust.
£43£3,000
83 cust.
£4,151
106 cust.
June20£6,000
110 cust.
£55£3,000
66 cust.
£2,153
51 cust.
July15£6,000
80 cust.
£75£2,000
36 cust.
£1,311
26 cust.
August15£6,000
80 cust.
£75£2,000
36 cust.
£1,311
26 cust.
September25£6,000
140 cust.
£43£4,000
103 cust.
£4,151
106 cust.
October50£6,000
290 cust.
£21£9,000
393 cust.
£18,000
669 cust.
November
limit 1,200 customers
100£6,000
590 cust.
£10£20,000
1,200 cust. (capped)
£15,437
1,200 cust.
December
limit 700 customers
90£6,000
520 cust.
£12£14,000
700 cust. (capped)
£8,907
700 cust.

Halberton can pack only so many hampers: 1,200 new ad customers in November and 700 in December (the last posting date cuts December short). Spending £20,000 in November would bring more than that, so the extra is wasted; the suggested split moves it into October, when people start buying for Christmas and corporate orders are placed, and keeps at least £1,000 in every month so the accounts stay alive. The harder finding: once only half of ad customers are counted as truly extra (£25 of gross profit each), the last £100 loses money in every month. The split is right; the total probably isn’t.

How confident are we?

What this rests on

Spend and new customers by month come from the business’s own records for last year. Each month’s response to spend is worked out from those figures, weighted 50% on its own and the rest on its demand (where a month had no spend, from demand alone); customers follow spend on a curve, 30% dearer at the margin. Only 50% of ad customers are treated as ones who wouldn’t have bought anyway, each worth the first order × margin + £20 of later orders. The suggested split keeps the same total, stays within what you could serve, puts no month above 3 times last year’s spend, and none below £1,000. Limits are new customers from ads only. October spend that brings November orders is counted in October. Next year’s months are assumed to behave like last year’s; a half-as-different case is shown.

Last year’s spend was the same every month, so how each month responds to more or less spend is an assumption. Vary spend this year, step October up gradually, and run a held-back test to check the 50%.

The months after

What the agent flagged from November to March, and what people decided

Fictional, continuing the example. At handover the managing director kept the plan as entered on the Months sheet from October, and chose the guided route. The agent was set up in an AI assistant the business already uses, connected read-only to the ad platforms' billing reports and the shop's orders (its IT confirmed this before it committed). On the third working day of each month it enters the month's actual spend and new customers from ads in a new dated copy of the plan, runs the checks, and compares each month with the plan, using the owner's lines on Inputs: spend more than 15% away from the plan either way, new customers more than 30% below the number the plan expects, a month below the £1,000 floor, or a month at what the packing room can serve. Nothing left the business, and no decision changed, without one of its own people approving it.

DateWhat the agent flaggedWhat was decided
3 Nov
No flag
October came in at £9,400 against the plan’s £9,000, with 371 new customers from ads against the 393 the plan expects. Both within the lines.Nothing needed. The managing director read the monthly note.
3 Dec
Flag
November reached the packing room’s limit: the 1,200th new customer from ads ordered on 24 November. Spend was £20,600, 3% above the plan.The managing director capped December’s daily budgets so spend slows as new orders near 700, and stopped the ads after the last posting date. The plan’s total stayed the same.
5 Jan
Flag
December spend was £11,200, 20% below the plan, and December reached its limit of 700 new customers.The underspend was the 3 December decision working. The £2,800 stayed unspent rather than moving to January.
3 Feb
Flag
January showed 41 new customers from ads against the 66 the plan expects, 38% short, on spend 2% above the plan.The marketing assistant found that a checkout update on 12 January had stopped recording the ad source on orders paid by digital wallet. Fixed and recounted: 63 customers, within the line.
3 Mar
Flag
February spend was £640 against £3,000, below the £1,000 floor, with 19 new customers. The ads had stopped on 9 February when the card on the main ad account expired.Card updated and the ads restarted at the plan’s daily budgets. The finance assistant now gets the ad accounts’ billing alerts. February’s shortfall was not made up in March.

At the check-in on 4 December, about day 45 and after the first monthly run, Geoff and the managing director read the log's first two rows against the platforms' invoices. No rule needed changing, and Geoff's upkeep of the set-up ended there. What the agent added. The method is unchanged: same spreadsheet, same checks, same decisions. The agent made sure each month was compared with the plan as soon as it closed. Two of the four flags were problems nobody had noticed (lost tracking and stopped ads), each found within days of the month end, and one made sure the packing limit shaped December's spend. The benefit here is time to act; no £ figure is claimed for it. The plan itself is re-run with this year's figures next autumn.