What actually happened, read from Meta and Shopify directly. Nothing on this tab is typed in or assumed — it is a report, not a forecast. Days follow the ad account's clock (US Central), so today is always still filling up.

Period buttons

Pick a window. The metrics, the market table and their notes all follow it. Since day one runs from 1 July 2026, when trading started, so it is a real total rather than a rolling month.

Headline metrics

• ROAS — revenue divided by ad spend. Above 1.00× the advertising returned more than it cost on the first order alone, before any repeat purchase. It is not the same as making money: see Retention for the ROAS a customer has to reach before the whole cohort pays for itself.
• AOV — average order value. CPA — what it cost in advertising to win one new customer.
• Subscribe % — how many new customers chose a subscription. Rebills — revenue from customers who were already subscribed, which no advertising was paid for in this window.

The market table

Countries are grouped into regions. The Americas are kept separate from Asia — Mexico built this business and folding it into an “Asia and rest” bucket would hide it.

• Winners & leaks opens each region twice over: the markets earning at or above 1.00×, ranked by revenue, and the markets absorbing real spend below 1.00×, ranked by how much they are absorbing. Ranking by spend alone would only ever show the second group, because we spend most where results are worst.
• All countries opens everything, largest spend first, for when you want the list rather than the decisions.
• Weight follows performance. A market carrying a meaningful share of revenue is set in full ink with a bar behind its revenue; a market with no orders on trivial spend fades back. The bar is scaled to the largest market, not the total.

ROAS is only graded where the spend behind it is big enough to mean something. Below that the number still shows but stays grey — true, not yet a signal. A market with no orders and trivial spend shows a dash rather than 0.00×, and revenue with no ad spend behind it shows no ratio at all: both are absences of information, not findings. CPM turns red above $35.

Day by day

The last eight days. A day is not closed until twelve hours after midnight in Chicago, so today is marked as running and left ungraded, and the day before it may still be settling. Acting on a provisional day is how you misread a scale test.

Advertising buys a customer once; the subscription is supposed to pay for that customer several times. This is the scoreboard on whether that is happening. The adherence half is measured on real orders; the forward half is modelled on the Projection tab's retention curve, so the two tabs cannot disagree.

Rebill adherence

• Only subscribers whose billing window has closed are scored — the cadence implied by the bottles they bought, plus a two-day grace because the biller charges at a fixed time in its own timezone. Someone who bought last week has not failed to rebill; they have not been asked yet. Counting them would make every recent cohort look like a collapse.
• Anyone who has already rebilled counts immediately, whatever the estimated date said.
• Green beats the first-rebill rate the forecast assumes, amber is within a quarter of it, red is below.

Who has not rebilled

The exact people behind that percentage, named and linked to Shopify. Days late runs from the estimated due date, so a few days either side is normal and two weeks or more is a genuine lapse. If this list looks wrong, the percentage is wrong — that is the point of showing it.

Cash per cohort

• Banked has happened; to come has not. Banked is cash that already moved. To come continues each surviving customer along the retention curve, conditional on the rebills they have already made.
• Net is gross revenue less the ad spend of the week that cohort was bought, less product, shipping by region, processing and EU VAT. Fixed running costs are excluded — they belong to the business, not to any one cohort.
• Return on ad spend is contribution over a cohort's life divided by what it cost to acquire. Above 2.00× the money roughly doubles; under 1.00× the cohort never repays what it cost to buy.
• The breakeven ROAS is stated under the table and derived from the actual cost structure, not assumed. ROAS lifetime is graded on it. ROAS now counts only cash already banked, so it stays grey until a cohort reaches its first rebill — before that a low number reflects the calendar, not the customers.
• Doubles by is read per cohort rather than blended: the weeks differ enough that one combined answer would describe none of them.

What this does not cover

Cadence is inferred from bottles bought rather than read from the subscription contract, so a customer on a rescheduled plan can appear due slightly early or late. Cohorts start 1 July 2026 — anyone acquired before then has rebills landing in the data but no first order here, so they are not scored. The running week is marked and will keep filling.

One cohort, bought once and then followed until it dies. Put a sum in, spend all of it on advertising on day one, and watch what comes back. Nothing else is running: no second cohort, no other spend. The question this answers is if I put £10,000 in, when do I have £20,000, and where does it stop.

Why an uplift on the rebill is the lever

• The first order is sold at a discount. The discount does not repeat, so every rebill can be charged closer to list. That is the industry pattern and it is already half-happening here — the free bottle does not ship again.
• A rebill is charged before anyone reacts to its price. The cancellation it provokes lands on the next cycle, not this one. So the money is collected and the cost is deferred.
• That makes it a trade of tail you might never collect for margin you certainly collect — from the largest group you will ever have, on their way out.

What is modelled and what is not

• Everything here is a model. Only ten rebills have happened, which is nowhere near enough to measure a survival curve. Treat the churn dial as the thing you are assuming, not observing.
• The uplift applies to every rebill, not only the first, and the churn dial rises with it through the sensitivity setting — charge more, lose more, sooner.
• Contribution is after payment fees, product and freight. Advertising is only ever paid once, on day one, which is why the curve only ever climbs.

Growth here is financed on the spot: advertising and shipping leave the bank today, and the revenue they produce arrives days later once the processor pays out. This tab walks the bank balance forward day by day so the gap between the two is visible before it bites.

The two ways cash runs out

• Float outrunning cash. Every day of scaling widens the gap between money spent and money received. Float is revenue already earned that has not landed yet — real, but not yet spendable.
• A production payment. Bottles need a deposit on order and the balance a few weeks later, before anything ships. Those two dates are lumps that a smooth daily model hides.

What is counted

• In: revenue net of processing fees, arriving after the payout delay. The first few days come from real past revenue already earned; after that from spend times measured efficiency.
• Out: advertising the day it is spent, fulfilment on every bottle shipped, VAT set aside as it accrues, fixed running costs, and the production deposit and balance on their dates.
• Fulfilment is charged at the rate actually being paid — Europe is not in place, so it is $8–9 a bottle rather than the $3.50 the forecast assumes. That difference is not small.

The number to read first

Sustainable daily spend is solved, not assumed: the highest daily advertising figure that keeps the balance above the floor across the whole horizon, production payments included. Above it you are scaling on money you do not have; below it there is room.

One question: what is the last day we can place a production order and still not run out. Stock and burn are measured; everything else is a dial you can move.

How the date is worked out

Bottles leave at a rate set by ad spend. Stock divided by that rate gives a stockout date. Walk back the production and shipping lead time, then walk back a safety buffer, and you have the order-by date. If that date is in the past, the only remaining options cost money — air freight, or letting a flavour go out of stock.

Why the flavours are separate

Running out of one flavour is running out. Cover is always the binding flavour, never the average — an average hides the flavour that fails first.

What is measured and what is assumed

• Measured: bottles shipped per flavour per day, ad spend per day, and stock on hand in the shipping warehouse. US and OMS stock is excluded because it cannot fulfil these orders.
• Assumed: lead time, safety buffer, batch size, and how burn scales with spend. Each is a dial, and the scenario table below shows what changes if the spend assumption is wrong.
• Deliberately conservative: the bottles-per-ad-dollar rate includes bottles that went to existing subscribers, who needed no new advertising. That overstates burn slightly at higher spend, which errs toward ordering early.

The advertising funnel as Meta reports it, every stage from impression to purchase, with the same window compared against the one before it. This is where a problem shows up days before it reaches revenue.

How to read it

• Every metric is compared to the previous period of equal length, not to a target. The question is always “is this moving, and in which direction”.
• Direction is interpreted, not just shown. A rising CPM is bad, a rising conversion rate is good; the colour reflects that rather than the sign of the change.
• Small numbers are not flagged. A 40% swing on trivial spend is noise, so movements are only called out once there is enough behind them.

Where problems usually appear first

• Frequency rising while CTR falls is creative fatigue — the same people are seeing the same ads again. It shows here well before ROAS moves.
• A gap between link clicks and landing page views is the site, not the ad: people clicked and left before the page finished loading.
• Cart to purchase falling while cart rate holds points at checkout — shipping cost, payment methods, or a broken step.

Meta counts differently from Shopify

Meta claims a purchase when it can attribute one, including views that never got clicked, so its count runs ahead of the orders Shopify actually recorded. Both are shown. The gap between them is not an error — it is the size of Meta's attribution claim, and worth watching in its own right.

Everyone acquired since 1 July 2026, cut three ways: where they are, which pack they chose, and which flavours. Customer counts, order values and subscribe rates are measured. Advertising cannot be attributed to a pack — someone picks the pack after they click — so cost per customer is exact by market and allocated everywhere else.

How the pack is identified

The offer shows in the shape of the order: a 2+1 arrives as quantities of 1 and 2, a 3+2 as 2 and 3, a 5+5 as 5 and 5. Totals alone cannot tell a 2+1 from three separate bottles, so the shape is kept rather than the sum.

The columns

• CPA — advertising per new customer. Exact by market; by pack and flavour each customer carries their own market's cost, which is the most that can honestly be said.
• Kept per order — first-order revenue after processing, product and shipping, before advertising. 1st-order margin is that less CPA: what a customer is worth on day one.
• A negative first-order margin is normal in a subscription business and only a problem if the rebills on the Retention tab do not arrive.

Arithmetic on stated assumptions, not a measurement. Every dial below is editable and everything recalculates instantly — the forecast is only worth what those numbers are worth. It compounds each region forward on its own last-seven-day performance.

How growth is modelled

• Advertising gets harder as it scales. Each doubling of spend costs a set percentage of ROAS, and that penalty itself worsens with every further doubling. This is the single most important assumption here and the one most often left out of forecasts.
• Spend is disciplined, not unlimited. It grows only while the next dollar still clears the ROAS floor in the dials, never takes cash below the floor you set, and past that point the surplus stays as cash instead of being spent. That floor sits below 1.00× deliberately — a customer keeps paying, so a first order that loses a little still adds money over the cohort's life.
• Every cohort runs out. A share of new subscribers buy a second box, a share of those buy again each cycle, and after the cohort life you set they are written to zero. Nobody is assumed to pay forever.
• Rebills ship fewer bottles at similar revenue — the free bottle in a 2+1 never reships, so cost falls and revenue does not.

What is charged against it

• EU VAT at the full rate on every euro of EU revenue, from the first order. The small distance-selling allowance was passed months ago; modelling it would only flatter the early weeks.
• Running costs grow with revenue, sublinearly — stated as a percentage per doubling. None of it is headcount: it is the Shopify tier, the subscription app's per-order fees, email priced by contact count, AI seats and usage, and the accounting and multi-country VAT filing that volume makes unavoidable whether or not anyone is hired.
• Stock is paid for before it ships. Europe and Asia are stocked months ahead, so while growing you are always buying for a bigger month than the one you are in. That gap is real working capital, which is why bottles paid for runs ahead of bottles shipped.

Reading the table

It is grouped deliberately, because the commonest way to misread a forecast is to mistake one month for a total: what a single month looks like, what has accumulated since day one, and where the balance sheet stands. Read the bottles row first — long before cash or ad efficiency becomes the constraint, the factory does.