A 12-month P&L forecast you'll actually maintain
Most forecasts die young: built once as a giant annual spreadsheet, wrong by month two, never opened again. A forecast you'll actually keep is small, driver-based, and updated in 30 minutes a month.
This is the one place AI earns its keep instantly, because a chatbot can draft the whole 12-month model from your last three months of actuals in one prompt. Any tax lines vary by country, so confirm those with a professional in your jurisdiction.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are a DTC finance analyst building me a simple, living 12-month P&L forecast. Keep it driver-based, not line-item fantasy. Do not state any tax rate as fact; leave tax as an input.
Where I'm based: [your country]
My last 3 months actuals: revenue by channel [numbers], units/orders [numbers], average order value [amount], COGS % or per-unit cost [number], fulfillment/shipping cost [amount], marketing spend [amount], fixed costs (rent, software, my pay) [amount].
Growth assumption: [e.g. orders +8%/month, or flat, or seasonal].
Do this:
1. Build a monthly P&L for 12 months with these rows: Revenue (by channel), COGS, Fulfillment, Marketing, Fixed costs, Contribution margin, Net profit. Drive revenue from orders x AOV, not a made-up top line.
2. Show the drivers explicitly (orders, AOV, margin %) as their own rows so I can change one assumption and see the model move.
3. Add a simple month-by-month cash note reminding me that profit is NOT cash (inventory paid upfront, payouts delayed).
4. Give me a 30-minute monthly update ritual: which 3-4 actuals to drop in and which drivers to re-check.
Do not invent my numbers, use only what I gave you; where something is missing, list it and ask. Treat any tax as an input I confirm with an accountant in my jurisdiction.
Output: the 12-month monthly P&L (drivers + rows), a cash caveat, and the monthly update checklist.
Or do it in 4 steps
- Set up simple monthly rows, not a wall of line items. Columns are the 12 months; rows are Revenue by channel, COGS, Fulfillment, Marketing, Fixed costs (rent, software, your pay), Contribution margin, and Net profit. That's it. A forecast you can read in ten seconds is one you'll keep; a 60-row masterpiece is one you'll abandon.
- Drive it from orders x AOV x margin, not a fantasy top line. Don't type a revenue number you wish were true. Build it up from drivers: expected orders x average order value, then your real margin. Now one changed assumption (orders grow 8% not 5%) flows through the whole model, which is the point of a forecast.
- Let AI draft the whole model, then sanity-check it. Hand a chatbot your last three months of actuals and a growth assumption (the prompt above) and it will build the 12-month grid in one pass. Your job is to check the drivers are realistic and the arithmetic ties, not to build it cell by cell.
- Run a 30-minute monthly update, and remember P&L is not cash. Once a month, drop in actuals for orders, AOV, and spend, then re-check next month's assumptions. That ritual keeps it alive. A profitable P&L can still run out of cash, because inventory is paid upfront and payouts land late. That's why cashflow is its companion note.
Worked example (labeled): one founder builds a beautiful 40-tab annual budget in January, never updates it, and by March is flying blind on gut feel.
Another founder has AI draft a 7-row driver-based model from three months of actuals, then spends 30 minutes on the first of each month dropping in real orders and AOV. They can answer "can I afford this ad budget / this hire?" in seconds, because the model shows the impact on contribution and net.
Same business, but the second founder is steering with numbers all year. Keep it small, driver-based, and updated monthly.
Pair this with scenario planning (run a good/base/bad case off the same drivers) and cashflow basics (because a healthy P&L is not the same as money in the bank).
Tax treatment of any line varies by country and changes often; confirm those specifics with an accountant qualified in your jurisdiction.