How Much Should a Number Change Before You Act?

Every funnel step changes a little every month. Knowing how much is normal tells you which changes deserve a meeting.

A line moving up and down inside a band, with one rust point below the band, cream on charcoal

THE SHORT VERSION

  • Every step in a funnel changes a little every month, even when nothing is wrong. Know the normal range before you react.
  • Teams react to every dip because reacting feels safe. That costs you twice: you chase normal ups and downs, and you miss the real drop.
  • Before you act, ask how much the step usually changes. When you do act, run the fix as a test with a date to check it.

Say applications are down 6% this month.

The first meeting happens that same week. Someone pulls the last three months. Marketing gets asked about lead volume. Admissions gets asked about follow-up. By Friday there’s a plan: more spend on the best channel, a new email to applicants who stalled, a daily report.

Nobody asked the first question. Is 6% a lot?

Every number changes, even when nothing is wrong

Every step in a funnel changes a little every month. Some months more people finish an application. Some months fewer. A holiday falls in a different week. A partner sends a batch of students at once. One counselor is out for two weeks.

Look back two years and most steps have a normal range. Say file completion runs between 56% and 62% in a typical month. A month at 58% is inside that range. A month at 52% isn’t.

File completion rate by month, made-up school Eleven months fall between 56% and 62%. This month is 52%, below the normal range. 62% 56% NORMAL RANGE This month: 52% 12 months ago This month
File completion by month for a made-up school. Eleven months fall between 56% and 62%. This month, at 52%, is the only one outside the normal range.

Without that range, every change looks the same. A 3-point drop and a 9-point drop both show up red on the dashboard. Both get a meeting.

Not knowing the range costs you twice

The first cost is reacting to normal ups and downs. The team changes something every month in response to changes that would have gone away on their own. When the number comes back, the change gets the credit. Now the team believes something worked that did nothing, and they’ll do it again next time.

The second cost is missing the real change. When everything gets a meeting, people stop taking the meetings seriously. A real drop looks like the last five false alarms. It sits for two or three months before anyone treats it differently. In enrollment, two or three months can be most of a cycle.

Why everything gets a meeting

Reacting is safe. If you call the meeting and the number comes back, it looks like you fixed it. If you wait and it doesn’t come back, it looks like you weren’t paying attention.

So people react, even when waiting is the better call. Nobody gets blamed for calling a meeting.

The range is the useful part of the forecast

Most reports show one number. Forecast: 594 starts. Target: 620. That looks precise. It isn’t.

The honest version is a range. Starts will likely land somewhere between 560 and 627, with about a 1 in 6 chance of reaching 620.

Leaders tend to strip the range out before the number goes to the president or the board. I understand why. A range sounds like you don’t know. One number sounds like you’re in control.

But the range tells you how much risk you’re carrying. A forecast of 594 with a narrow range and a forecast of 594 with a wide range call for different decisions. The first says hold steady. The second says find out why the range is so wide while there’s still time to act on it.

Ask how much the step usually changes

Before you react to a change, ask how much that step usually changes.

If this month is inside the normal range, write it down and watch next month. If it’s outside, that’s where the work starts. Find the step that changed. Check every likely cause against the evidence before anyone acts on it.

Small changes still count. Three months in a row at the low end of the range is its own signal. The point is to decide on purpose what’s worth a meeting.

Some changes don’t need a range at all. If the application form stops working, fix it today. You don’t need two years of data for that. The range is for changes where nobody knows the cause yet.

When you act, set a date to find out

Picking a fix doesn’t end the uncertainty. You still don’t know if it will work.

So run the fix as a test. Give the change to half of new applicants. Decide before you start what result would count, and the date you’ll check. Then check on that date, not before.

A test with a date lets you say “we’ll know by April.” That’s a more useful thing to tell a president than one confident number that turns out to be wrong.

Uncertainty is information.

It tells you how much to trust the number and how soon you need to act.

Try it on one number

  1. Pick one number you report every month. It could be applications, a sales pipeline, or renewals.
  2. Pull the last 24 months. Find the highest and lowest months that weren’t caused by something you can name. That’s your normal range.
  3. Look at the last three times that number set off a meeting. How many were outside the range?

If most were inside, you’ve been reacting to changes that would have fixed themselves. If one was outside and it took months to act, you’ve found what the other mistake costs.

Next month, when a number dips, you’ll know whether to call the meeting.

Written by Aaron Edgell

Growth executive. Builder. Writer.

Get new essays and notes by email.

One email when I publish something new.

Previous: