Root-cause analysisExplanationLive
Not “profit moved.”
Which line moved it, and by how much.
“Profit is down because of the ads” is a sentence any product can print. It is worth something only if the ads account for the fall — so the change is split exactly into the changes in the lines that make it up, and those add back to the cent.
Measured, not estimated EU-hosted Read-only by default
01 / The case for it
"Because of the ads" is a sentence, not an answer.
Without it
Any product can print a cause beside a figure that moved. It is worth something only if the thing it names accounts for the movement — and almost nothing checks, because checking means the arithmetic has to close.
What Profflow does
The change split into the lines that made it.
Every line beneath profit gets its own change, and those changes add back to the change in profit to the cent. Whatever moved most is named with its share beside it. That is a decomposition, and unlike a story it can be disagreed with.
02 / What it looks like
Where the change actually came from
Profit change€61
Revenue€30
Advertising−€45
Cost of goods−€9
Refunds and fees−€5
VAT€90
Every line is a measured difference between the two periods, and they add up to the change exactly. If they did not, the explanation would be a story.
45 + 30 − 14 = 61, which is the whole of the change. Revenue barely moved; the month was made by spending £45 less on advertising, and no amount of looking at a revenue chart would have shown that.
03 / Worth understanding
The difference between a reason and a sentence
“Because of the ads” is worth nothing on its own
Any product can print that sentence. It is worth something only when the advertising line is measured against the same period, in euros, and the lines together add up to the change — otherwise it is a plausible sentence attached to a real number.
The parts have to reconcile with the whole
The decomposition here is exact: every contributing line, including the ones that helped, summed to the movement you are asking about. A reason that explains €45 of a €61 change has not explained it.
Down to the day and the product
A month-level cause is a starting point. The same decomposition runs on a day, a channel or a product, so “advertising” becomes one campaign on three days — which is a thing somebody can actually do something about.
04 / What it is built from
Root-cause analysis is assembled
from accounts you already run.
01This period’s P&L02The period before it03Daily revenue and costs04Refunds by date05Ad spend by day06Cost lines by kind
Root-cause analysis
Exact decompositionThe largest mover namedDays that do not fitCosts that moved05 / How it is worked out
From the account
to the figure on the screen.
- 01SplitsThe change in profit into the change in every line beneath it.
- 02RanksWhatever moved most, with its share of the total movement.
- 03ChecksThat the parts add back to the whole. If they do not, nothing is shown.
- 04StopsAt the arithmetic. It names what moved, not why you moved it.
Asked of the figures
Why is profit up £61 when revenue barely moved?
Advertising fell £45 and revenue added £30; goods, tax, refunds and fees took £14 back. 45 + 30 − 14 = 61. That is a decomposition, not a story — and it is checkable, which a story is not.
How Profflow reasons06 / What it lets you do
What follows
from getting Root-cause analysis right.
Exact decomposition
Revenue, goods, tax, refunds, fees, ads and fixed costs — each with its own change, adding to the change in profit to the cent.
The largest mover named
With its share of the movement beside it, so a £45 line in a £61 change reads as most of it rather than as a cause.
Days that do not fit
Found with the median and the median absolute deviation, not the mean — one Black Friday wrecks a mean and makes the day that matters look ordinary.
Costs that moved
A supplier rise and a fee change look identical on a revenue chart and nothing alike on this one.
No cause invented
“6 July was unusually low” is a fact. “Ads caused it” is a guess, and this product does not print guesses next to measurements.
Opens to the rows
Every component of the decomposition opens to the orders and charges it was computed from.
07 / How the figure is kept honest
The same four rules,
on every screen.
Profflow observes the accounts it is connected to. It does not write an order, a price, a campaign or a payout back to any of them.
A charge read from an account is labelled measured. Anything modelled — a forecast, a projection, a filled gap — is labelled as such on the screen it appears on.
Follow any number to the dated rows underneath it. A figure nobody can check is a figure nobody acts on.
The product prepares a decision and shows the evidence. Nothing consequential happens until you say so.
Questions
About Root-cause analysis.
- Why is the mean not used?
- Because it is wrecked by exactly the value you are looking for. One enormous day pulls the mean up and inflates the deviation, so the unusual day stops looking unusual and three ordinary days start to. The median and the MAD are unmoved by a handful of extremes.
- Does it tell me the cause?
- It tells you which line moved and by how much. Whether the ad spend caused the revenue is a question about your business that arithmetic cannot answer, and the product does not pretend otherwise.
- What if the parts do not add up?
- Then something is wrong and the reconciliation check says so as a fault in Profflow. A decomposition that does not reconcile is not shown as though it did.
- How far back does it compare?
- The equally long stretch immediately before the period on screen. A week against the month before it is not a change, it is two different questions.
One operating system
The rest of the picture,
one view away.
Each part is useful on its own. Profflow holds them in one model, so a figure here carries the context from everywhere else.
See the whole productSee Root-cause analysis on your own figures.
“Profit is down because of the ads” is a sentence any product can print. It is worth something only if the ads account for the fall — so the change is split exactly into the changes in the lines that make it up, and those add back to the cent.