Power Law Performance Subscribe

ResourcesForecasting: Revenue

Existing Book

The accounts already on the book, the same four ways, each reforecast as the months close.

Chapter 2 of 3 · 10 min read · A guide by Broad Eye Solutions · Version 1.0 · Updated November 14, 2026

Get the whole guide as a PDF

Chapter 1 built next year’s new accounts four ways. On this book they’re about 12% of a year’s revenue. The other 88% comes from accounts already billing when the year begins: the existing book.

This chapter builds the existing book in four layers, the way Chapter 1 builds net new. Each starts from the one before. Straight Line closes out this year’s book, adds this year’s new accounts, and draws next year on a straight line from January’s share. Seasonal fills that year by the time of year. Cohort runs it on every approval cohort, and Power Law on every tier inside them. Each layer is a full year ahead, and each ends with its Reforecast. It’s the same invented payments portfolio, and the numbers are sample data, no client data. Revenue, as in Chapter 1, is what each account billed in the month, as on its statement.

Here’s what you’ll be able to build by the end: the existing book next year, by month, four layers. Switch between them under the title.

Sample data: an invented payments portfolio, no client data. In the repository: CSV: Straight Line, Seasonal, Cohort · made by the engine

What you’ll need

  • Every account’s approval month, and what it billed in every month, back as far as you have. This book goes back to January 2020.
  • A December for every year: the month each year’s book is read from.
  • Chapter 1’s numbers for this year’s new accounts: they join the book heading into January.

Straight Line reads last year, as in Chapter 1. Seasonal, Cohort and Power Law read every earlier year: the existing book is one book all the way back, and the more years, the steadier the read.

Straight Line

The forecast you take into January, made the October before.

Starts from: this year’s book, closed out at the October close, and this year’s new accounts from Chapter 1.

Step 1: Read Every Year The Book Came In

The existing book in any year is every account approved before January. It comes in at what it billed in December. Each January it takes in last year’s new accounts, the net new in year from Chapter 1, so the book you carry in grows every year. Here’s every year on this book: what came in, and what was still billing the December after.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.1 Straight Line · CSV

Read the kept columns down. Across 5 years, the share of December’s revenue still billing a year later stayed between 84.6% and 87.2%. The share of billing accounts stayed between 74.5% and 76.1%. Revenue keeps more than the count because the accounts that leave are smaller: the December before, they billed 36% to 45% less than the ones that stayed. This year the book came in at $3.22M and 8,760 billing accounts, and $639K and 2,581 of that were last year’s new accounts. Its own next December hasn’t billed yet: Step 2 puts it at $2.74M.

Step 2: Close Out This Year, Add The New Accounts, Draw The Line

These are the three moves of Chapter 1’s Step 1, on the book you carry in.

Close out this year. This year’s book has billed $27.60M through October. Last year’s book billed 16.3% of its year in November and December, so this year’s finishes at about $32.96M, with $2.74M in December.

Add the new accounts. This year’s, from Chapter 1, bill about $578K in December and join heading into January. Next year’s book comes in at $3.32M, 3.2% bigger than this year’s came in at. Where Chapter 1 has a goal, the existing book has that: next year’s total is this year’s $32.96M grown by the same 3.2%, $34.01M.

Draw the line. This year’s January was 8.11% of this year. Next year’s January starts at that share of the total, $2.76M, and every month after adds $14K, so the twelve months add up to the total.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.1 Straight Line · CSV: Revenue, Billing accounts, The math

Billing accounts come in the same way. This year’s book ends the year at 6,634: October’s count times last year’s December over October. Chapter 1’s 3,268 new accounts join at the 79.4% of last year’s new accounts still billing in December, 2,595. From the 9,229 the book comes in with, they go down one equal step a month at the average kept, to 6,954 next December.

Here’s the same build for every year so far, each from its actual December and the year before it:

Sample data: an invented payments portfolio, no client data. In the repository: CSV · made by the engine
Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.1 Straight Line · CSV

That’s Straight Line’s existing book for next year: $34.01M.

Reforecast

Chapter 1’s way. At each close the months closed are what billed. Divide them by last year’s share of those months, and that’s the year. What’s left goes on the straight line through the months still open, from January’s share of this year. Billing accounts carry on down the line from the latest close.

Sample data: an invented payments portfolio, no client data. In the repository: CSV: 3+9, 6+6, 9+3, Every month: October · made by the engine

What This Build Can’t Tell You

A straight line can’t see the time of year. Its year came within 1.4% of what billed in every year from 2022 to 2025, but its months were off by 4.0% on average: over in February and the fall, under in the spring and summer. The next layer puts the time of year in. And one rate for the whole book can’t see which accounts are leaving.

Seasonal

Same year, same accounts. What changes is which months carry it.

Starts from: Straight Line’s year, $34.01M.

What You’ll Need

  • Every year’s existing book by month, the same grid as Straight Line’s Step 1.

Step 1: Read Each Month As A Share Of Its Year

Line up every year the book has, and read each month as a share of its year, as Chapter 1’s Seasonal does. This year joins for the months it has closed, against the year it’s on course for.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.2 Seasonal · CSV

The same months sit low every year: February and the fall. On average February runs lowest, 7.9% of the year, and May, the highest in every year, 8.8%. Processing volume, read the same way, follows the same shape: February at 7.9%. So the shape is merchants processing more or less with the season.

Step 2: Fill Straight Line’s Year By The Time Of Year

Multiply Straight Line’s year by each month’s average share. The year stays the same. Only the months move.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.2 Seasonal · CSV
Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.2 Seasonal · CSV

That’s Seasonal’s existing book for next year: $34.01M, the same year as Straight Line, with each month carrying its usual share.

Reforecast

The same year as Straight Line at each close: the months closed, divided by last year’s share of them. The months still open each take their average share of the year, not the straight line’s.

Sample data: an invented payments portfolio, no client data. In the repository: CSV: 3+9, 6+6, 9+3, Every month: October · made by the engine

What This Build Can’t Tell You

The months now have the right shape. Forecast from each year’s own December, this layer came within 1.4% of what billed every year, and its months within 0.9% on average (the grid at the end of the chapter). But it’s still one rate for the whole book. A book whose mix of old and new accounts shifts can’t be seen here.

Cohort

Starts from: the book Straight Line carries in, split by approval cohort, in Seasonal’s shape.

What You’ll Need

  • Each account’s approval month, and what it billed in every month, back as far as you have.

Step 1: Put Every Cohort On Its Own Line

Split the book you carry in by when each account was approved. Each approval month is a cohort, and each comes in at its share of Step 2’s $3.32M: what its accounts billed at the October close, read forward to December. Its year is its December times what cohorts of the same age billed in a year against their December, in earlier years: month by month for the first two years, then year by year. Its months take Seasonal’s shares. Switch to the second tab to open the two youngest years into their months.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.3 Cohort · CSV: By approval year, By approval month

The youngest cohorts bill the least against their December. This year’s new accounts are expected to bill 9.8 times their December next year. Every older year’s cohort bills between 10.2 and 10.5 times.

Step 2: Add Them Up

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.3 Cohort · CSV

That’s the cohort layer’s existing book for next year: $33.90M.

Reforecast

At each close every cohort is read against its own path: its months closed, against the path’s months, rescale its months still open.

Sample data: an invented payments portfolio, no client data. In the repository: CSV: 3+9, 6+6, 9+3, Every month: October · made by the engine

What This Build Can’t Tell You

Each cohort is still hundreds of accounts read as one. It can’t tell a large account from a small one, so it can’t show how much of the year rests on a few large accounts.

Power Law

Starts from: the cohort layer, every cohort split by tier.

What You’ll Need

  • A tier for every account on the book, cut the way Chapter 1 cuts it: Power Law Nested Means on monthly processing volume, each account at the best tier it has reached.

Step 1: Split Every Cohort By Tier

Each tier inside each cohort runs at the ratio of year to December that tier billed at that age in earlier years. Accounts approved but not billing yet, and the ones still to be approved, run at the ratio every tier billed at their age.

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

Tier A is 738 accounts, 8% of the book, and it bills 53% of December’s revenue. Next year it carries $18.39M, 54% of the layer’s year. The ten largest accounts alone billed $2.83M over the last twelve months, 7.7% of the book. Lose them in January, and the year comes in short by far more than any layer has ever missed.

Step 2: Add Them Up

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

That’s the Power Law layer’s existing book for next year: $33.97M. On the total it lands within a whisker of the cohort layer. What it adds is where the year comes from.

Reforecast

The same as the cohort layer, with every tier inside every cohort read against its own path.

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

What This Build Can’t Tell You

The tiers say where the year comes from, not whether to believe it. Every line runs at the rates of earlier years, and a book can always break from its own history. In Chapter 1, Power Law’s total is the cohort build’s, because a new account’s tier isn’t known a year ahead. Here every account already has one, so the two can differ.

Where the existing book lands you

Four layers of the existing book’s year. Each starts from the one before.

Sample data: an invented payments portfolio, no client data. In the repository: Excel, tab 2.5 Three layers · CSV

The grid that matters is this one: every year so far, forecast from its own December with the rates of the years before it, against what billed. Every layer’s year came within 1.4% of what billed. Straight Line’s months were off by 4.0% on average; the time of year takes that to 0.9%, and the cohort and tier layers stay there.

Sample data: an invented payments portfolio, no client data. In the repository: CSV · made by the engine

With the rates of every year through 2024 and December 2025 still an estimate, each layer’s forecast of 2026’s months so far ran against what billed by +5.0% for Straight Line, +6.1% for Seasonal, +6.1% for Cohort and +5.0% for Power Law. Most of that is the start. Estimated at the October 2025 close, the book was to come in at $3.40M, with $731K from that year’s new accounts. It came in at $3.22M, and the new accounts billed $639K.

And 2026 as each layer stood at each close:

Sample data: an invented payments portfolio, no client data. In the repository: CSV · made by the engine

Here’s the chart from the top of the chapter again. You can now build every bar, four layers.

Sample data: an invented payments portfolio, no client data. In the repository: CSV: Straight Line, Seasonal, Cohort · made by the engine

What This Chapter Can’t Tell You

You now have the existing book in four layers and net new in year four ways. The check above is on the existing book alone. Chapter 3 sets every method against what the whole book billed, close by close.

Get the whole guide as a PDF

Every chapter, version 1.0, in one file, sent to your inbox. The newsletter follows with the articles that build on it.

One email a week after the PDF.

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