Pott Candles: new customers, CAC, and what each line costs

30 October 2026, rebuilt after Ailis's comments. All figures include VAT. Draft for discussion.

The forecast is now built in three layers, in the order Ailis asked for: subscriptions from the charges Recharge already has scheduled, other returning customers from last year's days and this year's bigger base, and new customers bought with ad budgets that stay flat within each week and ramp only into Black Friday. The rule has not changed: as a business, the ordinary days before Black Friday break even after ads and overhead, and the profit comes from early access onwards.

1. Where we are

New and returning are on the dashboard's first-order basis (a customer is new on the day of their first ever order). Shopify's returning flag puts about £1.8k a day more into returning and the same amount less into new, which is why Ailis's split of the same week reads differently. Both are right; the page uses one basis throughout.

2. New customers and CAC, by phase

Each table shows last year on the same days, our forecast, Ailis's model as she wrote it, and what her ad budgets would buy if new customers cost what our curve says they cost at that spend.

Before Black Friday, 1 to 25 November

The one real disagreement is here. Her sheet buys 2,849 customers at a flat £55 on £6,268 a day. The last week ran at £60 on £5,339 a day, and at £6,268 a day our curve puts CAC near £75. Her days still balance because her returning is £9,108 a day against our £5,953; on our returning the same budgets lose about £2,171 a day.

Early access to Cyber Monday, 26 to 30 November

Here we agree. Her £32 CAC is close to our £30, and her £84,224 would buy about 2,511 customers on our curve against her 2,632. The gap in revenue is again returning: £236,014 in her sheet, £184,905 in ours, £115,342 last year.

November

December

December's new-customer line is at the 3.5x floor every week. The month is 58% returning and subscriptions, so the December lever is the send plan and the 12 December and Boxing Day offers, not ads. Her £768k would need 7,600 new customers against our 2,409.

3. The spend path

Budgets are set per block and held flat inside it, so a launch day does not pull ad spend up and a quiet day does not pull it down. Each pre-BF block is set at the level where its ordinary days break even after ads and overhead; the worst single day in the plan is about −£792. The step down from today's £4,850 happens on 1 November. The sharp rise is on 26 November, where it belongs.

The Gold launch (10 Nov) and the birthday (14 Nov) carry the same budget as the rest of their week. They are email and SMS days, and they show up as profit: £7,893 and £4,421 after overhead. The earlier version of this page spent £14,605 on Gold day because it let the launch margin fund more ads; Ailis was right that it should not.

4. Returning customers and subscriptions

Subscriptions are settled. Recharge already holds every November renewal by day, and all three numbers agree:

The page now uses the Recharge ledger day by day, so 1 November reads £6,160 of returning including £3,597 of subs, not the £5,094 of the earlier version. December is not scheduled in Recharge yet (it only holds the next cycle), so December subs are the renewal book, £91,280, on November's day-of-month shape.

Other returning is the number in dispute, and it is the bigger of the two gaps:

Last year other returning fell about 18% from October into early November. Ours carries that dip on a base that is 70% bigger; hers holds the current run-rate flat and then some. The 30 October cohort check, which worked last year's spend per customer by cohort age against this year's cohort sizes, landed at £330k to £360k of returning for November against our £334k and her £464k. Neither side can prove it today. The first week of November will: if other returning runs at £5k a day or more on the ordinary days, the break-even budget rises and we step up.

5. What a new customer is worth

From the Shopify order history, revenue per new customer after the first order, by when they were acquired. Pre-Black Friday customers are worth almost twice a Black Friday customer over the following year, and the 2025 cohorts are tracking well above 2024 because more of them subscribe.

That gives the CAC the business can afford before Black Friday, depending on how long it is prepared to wait for the money back:

Ailis's £55 sits between the daily break-even and the 12-month payback. It is a reasonable price for a customer; it is not a price the business breaks even on in the month. The forecast runs on the daily rule because that is what we agreed. If Pott would rather carry a November loss for a bigger 2027 base, the 12-month line says how far that can go, and the daily control below is how we would run it either way.

6. Daily control through November

Ailis is right that November should be run day by day on new customers and CAC. The forecast gives the numbers to run it against.

1. Every morning, three reads from the dashboard for yesterday: subscriptions against the Recharge ledger (should match within a few percent), other returning against the forecast line, and new customers with their CAC on the first-order basis.
2. Budgets change in steps, not jumps. At most 10% a day up or down, and only on a three-day read, never on one day.
3. Step up when three-day CAC is under £51 at the current budget, or when other returning runs 20% or more above forecast for three days (the break-even budget rises with it; the block is re-solved and the new level reached in 10% steps).
4. Step down when three-day CAC is above £66, or other returning runs 20% below forecast for three days.
5. Re-solve the blocks weekly with the actuals in, so the 9-15 and 16-25 November budgets are set on what 1-8 November actually did rather than on the model.
6. Black Friday itself is run on the first hours of early access: above 4.4x, the 26 and 27 November budgets take more; below 3.5x, they hold.

7. Day by day

Sign-off

If this is the forecast we work to, approve it here and Casey will publish it to the Kova Forecast as the Q4 plan. This signs off our forecast on page 2, not the £900k / £768k scenario on this page. If it is not right, reject it and tell us why; your comments go straight to Casey.

Draft, 30 October 2026. Our forecast as a whole is on page 2. Nothing changes in the ad accounts or the Kova Forecast until it is approved and Casey publishes it.