Forecasting Residual Income and Setting a Monthly Goal

Most agents forecast residual income by taking last month and adding optimism. It produces a number that is wrong in a predictable direction and, worse, is wrong in a way that hides the two things a forecast is supposed to surface: whether the book is growing on its own, and whether the pipeline is large enough to hit the goal.

A forecast that separates those two questions is more work for about twenty minutes a month and is the difference between knowing you are behind in week two and discovering it in week five.

Forecast from margin, not volume

Volume is the number agents quote to each other and it is nearly useless for forecasting income, because the relationship between volume and your commission is not constant across your book. Two merchants processing identical volume can produce very different margin depending on card mix, average ticket, the template they are on and how much of their spend is bank payments.

A merchant moving volume from cards to ACH is a merchant whose volume is flat and whose margin contribution changed materially. If you forecast from volume you will not see it. Forecast from the margin line, whatever your program calls it, and volume becomes what it should be: a diagnostic you look at when margin moves unexpectedly.

Split the book into three cohorts

A single trailing average conflates three populations that behave completely differently.

• Mature accounts, live more than six months. These are predictable and their trailing three-month average is a genuinely good estimate of next month.
• Ramping accounts, live under six months. These are still growing into their run rate and a trailing average systematically understates them. They also carry most of your attrition risk.
• Seasonal accounts. Any trade with a season. Their trailing average is actively misleading and should be replaced with the same month last year where you have it.

Forecast each cohort with the method that fits it and add them. The mature cohort is your floor, and knowing your floor is the most useful single output of the exercise, because it tells you what you earn if you sell nothing next month.

Attrition is not optional in the model

Every book loses accounts, and a forecast that does not subtract is not a forecast. The mistake is applying a percentage to the count. Attrition is not evenly distributed across your merchants: it concentrates in the first months, in accounts that never reached the volume they projected, and in accounts you have not spoken to.

The practical version is to look at your actual losses over the last six months and characterise them rather than count them. If four of your five losses were under ninety days live, your attrition assumption belongs on the ramping cohort and not on the mature one. Subtracting a flat percentage from a stable mature book while ignoring a churning new one produces a number that is wrong twice.

Weight the pipeline honestly

The pipeline half is where forecasts go badly wrong, because agents weight by enthusiasm. A deal is not seventy percent likely because the conversation went well. Weight by observable stage, and let the stages carry fixed weights you do not adjust per deal.

The other half of pipeline weighting, and the one almost nobody models, is timing. A merchant who signs on the twenty-fifth contributes a few days of processing to that month, and their first full month is the one after. An agent who forecasts a signed deal at full monthly value in the month it closes will overstate every good month and then be confused by the shortfall.

Two fields fix this: expected close date and expected first full month. The second is the one that goes in the forecast.

Set the goal in accounts, not dollars

A monthly income goal is not actionable, because you cannot do a dollar. Convert it once and then work the converted number.

Take your target monthly increase, divide by the average monthly margin of a new account in your book, and you have how many accounts you need to write. Divide that by your close rate and you have how many proposals. Divide by your proposal rate and you have how many statements you need to collect. That last number is the only one you control directly, and it is the number to put on a wall.

Doing this arithmetic honestly is uncomfortable the first time, because it usually reveals that the goal requires substantially more prospecting activity than is currently happening. That is the point of doing it.

Review monthly, revise quarterly

Look at the forecast against actual every month and note where the gap came from: mature attrition, ramping slower than modelled, pipeline slipping, or a closed deal landing in a later month than assumed. Do not adjust the model for one month's noise.

Revise the assumptions quarterly, with a season's worth of evidence. A forecasting model that is retuned every time it is wrong stops being a model and becomes a description of the past, which you already had.