Revenue dipped. Is it real, or just a slow week?

Every business has slow Tuesdays. The expensive mistake is reacting to noise as if it's a trend, or ignoring a trend because it looks like noise. There's real math for telling the two apart.

Guide · FantasyLab ·

Step one: compare like with like

A Tuesday can only be judged against Tuesdays. Compute the typical value for each weekday from the last couple of months (the median again, not the average), and how much that weekday normally swings. A $700 Tuesday against a $900 typical Tuesday with normal swings of $250 is a shrug, not a signal.

The method, plainly

Analysts call the serious version change-point detection: compare the stretch of days before a suspected change with the stretch after, and test whether the difference is bigger than your normal day-to-day variation. If twenty days after March 4 average 32% below the twenty days before, and your normal swing is a fraction of that, the drop is real with high confidence. If the difference sits inside your usual chop, the right verdict is not ready to call.

Three questions to ask first

  1. Is it one day or a run of days? One bad day means nothing. Six days in a row under their weekday norms is a streak worth watching even if no single day broke a record.
  2. Did the mix change or the volume? Fewer sales, smaller baskets, and lower prices are three different diseases with three different treatments. Split revenue into those three drivers before acting.
  3. Did anything real happen? A menu change, a road closure, a platform fee change on a specific date turns a statistics question into a story question. Check the calendar before the math.

What good software should say

When the evidence is thin, the only correct answer is that there is not enough evidence yet. Any tool (or consultant) that hands you a confident story for every wiggle is manufacturing certainty you'll pay for later. Waiting three more days for a clear verdict is cheaper than reacting to noise today.

Common questions

How many days until a drop is confirmable?
It depends on how noisy your business is. A steady shop can confirm a real shift in a week or two; a spiky one needs three or four. The math widens with your noise, which is exactly what it should do.

What is a change-point in plain words?
The day your numbers stopped behaving like the old numbers and started behaving like new ones. That's a regime change, which is more than a dip.

Counsel

Counsel does this math on your real numbers.

Connect Stripe, Square, Shopify, Etsy, QuickBooks or a bank feed, or drop in a CSV. Every answer comes with a receipt that shows the method, the sample size and how confident it is. When the math isn't conclusive, the app says so.

Every figure on this page is computed from Counsel's demo businesses or from the numbers you type in. None of them are made up. This is cash math, not tax or legal advice. Written and kept up to date by the founder.