Power Law Performance Subscribe
Articles

Forecasting Next Year, One Layer at a Time

A forecast can be built in an afternoon. Each layer you add after that gets the months and quarters closer, and shows where a miss comes from.

By Joseph Brady · November 16, 2026

You’ve just been made responsible for next year’s forecast. How do you build it, and how far should you trust it?

The Forecasting: Revenue guide answers both on one book, an invented payments portfolio with sample data and no client data. It builds next year’s new accounts four ways, builds the accounts already on the book in four layers, then back tests every method against what was billed. This is the short version.

Here’s where the building ends up: next year for the whole book, by month, four ways.

Sample data: an invented payments portfolio, no client data.
Sample data: an invented payments portfolio, no client data.

Chapter 1: Net New In Year

Net new in year is the revenue from accounts approved during the year, in the months they bill that year. The chapter builds it four ways, each ending in a complete forecast.

  • Straight Line. Start at the end. This year’s new accounts will bill about $3.76M; add a 10% goal and that’s $4.14M. Work back to a count with this year’s revenue per account, 3,595 accounts, price it with this year’s cost per account, and draw the year as a straight line: January at this year’s January share, every month after up by the same step.
  • Seasonal. The same number, with each month carrying its average share of the year.
  • Cohort. Each month’s new accounts on their own line, followed as they bill. Next year is the count, times what a new account bills by its age, times the time of year: $4.42M from the same 3,595 accounts.
  • Power Law. The cohort build, split into four tiers by size. On this book the top two tiers are 16% of the new accounts and 72% of their revenue. A year ahead the tiers don’t change the total. They change what the plan says it needs: about 135 A accounts carry $1.63M of it.
Sample data: an invented payments portfolio, no client data.

Each section ends with a Reforecast: the same build made again at the 3+9 (three months closed, nine forecast), the 6+6, the 9+3 and every month.

Chapter 1 in full: Net New In Year

Chapter 2: Existing Book

New accounts are about 12% of a year’s revenue. The rest comes from the book you already have, built as a layer cake. Straight Line works it the way Chapter 1 does: close out this year’s book, add this year’s new accounts, grow this year’s total by how much bigger the book comes in, and draw the months from January’s share. Its year came within 1.4% of what billed every year, but a straight line can’t see the time of year: its months were off by 4.0% on average. Seasonal keeps the year and fills it by the time of year, taking the months to within 0.9%. Cohort runs it on every approval cohort at its own age’s ratio, and Power Law on every tier inside them: Tier A is 8% of the accounts and 53% of December’s revenue.

Sample data: an invented payments portfolio, no client data.

Chapter 2 in full: Existing Book

Chapter 3: Back Testing

Add the two, and next year for the whole book comes to $38.15M by Straight Line, $38.15M by Seasonal, $38.32M by Cohort and $38.38M by Power Law. Then the guide stands at every month-end close from December 2021, makes each method’s forecast for the whole book with only what was known that day, and sets every month of it against what billed. A fifth method joins them: the trend line most teams reach for during the year.

Sample data: an invented payments portfolio, no client data.

Twelve months ahead, the typical miss (how far off a forecast landed, either way, on average) on the whole book’s month is 6.0% for Straight Line, 7.5% for the trend line, 5.0% for Seasonal and 4.1% for Power Law, which stays at or under 4.1% at every distance out to two years on this book. The quarters tell the same story: Straight Line’s quarters missed by as much as 9.0%, Power Law’s by no more than 4.3%.

On the whole book’s year, all four missed by 2.0% to 2.1% on average, and the trend line by 4.8%. On next year’s new accounts no method wins, because every one of them rests on the plan’s count of accounts nobody has signed yet. Fed the latest cohorts’ pace instead of the plan, Power Law runs low: 7.3% under on average at twenty-four months.

Chapter 3 in full: Back Testing

What the layers buy

Straight Line gets you a number in an afternoon. Each layer after it adds something the number can’t give you: the shape of the year, a forecast built from the accounts and read again every month, and a tighter band for the months and quarters ahead.

The guide has every step, with the grids filled in so you can lay your own book out the same way. The workbook, the CSV files and the engine behind it are on GitHub: BroadEyeSolutions/power-law-forecasting-revenue.

Want the next one in your inbox?

Every new article lands in your inbox the week it comes out, with the research behind it.

New articles and research, straight to your inbox.

Your choice

Consent Preferences

Choose what this site can store in your browser. Your choice is kept in this browser for 6 months.

Strictly necessaryAlways on

Your cookie choice and the bot check on the sign-up forms. The site can’t work without them.

Cookie Policy Privacy Policy