RecommendationsWhat to do nextLive

Of everything you could do,
which is worth doing first.

Ranking needs a common unit and the only honest one is money. Every recommendation carries an expected impact per month, worked out by arithmetic printed on the card — not a score out of ten, not a traffic light.

Measured, not estimated EU-hosted Read-only by default

Ranked byeuros a monthnothing else
Uncostedranked lastnever invented
Confidenceabout the inputsnot optimism
Riskwhat it costsif it goes the other way

01 / The case for it

A list of things you could do is not a plan.

Without it

Advice with no price on it cannot be ranked, so it gets ranked by whatever is easiest to say — a score out of ten, a traffic light, the order somebody typed them in. None of that tells you which one to do on a Tuesday with two hours.

What Profflow does

Ranked in money, with the arithmetic on the card.

Every recommendation carries what it is expected to be worth per month and the sum that produced it. One that cannot be costed carries no number at all and ranks below every one that can — an uncosted suggestion should never outrank a measured one.

02 / What it looks like

What to do next, ranked by what it is worth

On the screenThis week’s list, in euros a month
What to doWorthWhy, and what it rests on
Pause TikTok prospecting€180Returns 3.6× against a 4.19× break-even, four weeks running
Reprice the rain shell€140Best seller, 14% returns, thinnest contribution in the top five
Enter the missing costs€95Three variants have no cost of goods; margin is unknown on 9% of orders
Move the Tuesday send€40Campaign revenue is consistently lower on Tuesdays than Thursdays

Each one carries its confidence and what it would cost you if it is wrong — a recommendation without a downside is a sales pitch.

Ranked by expected impact in money, because it is the only unit that compares an ad decision with a pricing decision. Anything that cannot be costed is ranked last rather than dressed up.

A price rise can lose volume. Cutting ad spend can lose sales the attribution never credited. Renegotiating a courier cannot really go wrong. Those are not the same suggestion with different upside, so the risk is stated separately from the money.

03 / Worth understanding

Why ranking needs a unit

Money is the only honest common unit

A list that mixes “improve your ad targeting” with “add your costs” and ranks them by importance is ranking by opinion. Expected impact in euros a month can be argued with, which is the point.

What it rests on, and what it costs to be wrong

Every item carries the figures it was derived from and a plain statement of the risk. Pausing a campaign has a downside; so does repricing. A recommendation that mentions only the upside is advertising.

Uncosted advice is ranked last

Some good ideas cannot be priced. They stay on the list, at the bottom, marked as uncosted — rather than being given a number so they can compete with the ones that have one.

04 / What it is built from

Recommendations is assembled
from accounts you already run.

01Contribution by product02Discount cost and break-even volume03Payment and platform fees04Ad return against break-even05Refund rate by product06Fixed costs
Recommendations
An impact in moneyConfidence about the inputsRisk stated separatelyThe evidence attached

05 / How it is worked out

From the account
to the figure on the screen.

  1. 01FindsThe changes the figures actually support, not a checklist of good practice.
  2. 02CostsEach one in euros a month, by arithmetic written on the card.
  3. 03QualifiesWith confidence about the inputs and risk about the downside.
  4. 04RanksBy money, with the uncostable below everything costed.

Asked of the figures

Why is this one at the top?

Because it is worth the most per month of everything the product can cost, and the arithmetic that says so is on the card. A suggestion that cannot be costed carries no number and ranks below every one that can.

How Profflow reasons

06 / What it lets you do

What follows
from getting Recommendations right.

An impact in money

Not a priority score. A figure per month, with the sum that produced it visible.

Confidence about the inputs

A store that has entered real product costs and connected its processor is being advised from measurements. One that has not is being advised from estimates, and the card says which.

Risk stated separately

A price rise can lose volume. Cutting ad spend can lose sales the attribution never credited. Renegotiating a courier cannot really go wrong. These are not the same suggestion with different upside.

The evidence attached

The same figures as the screen behind it, so the reasoning can be checked rather than believed.

Uncosted ranks last

A suggestion with no number carries null rather than a made-up one, and never outranks a measured one.

Silence where there is no data

No campaign recommendations, because campaign-level spend is not read yet. No inventory ones, because stock levels are not synced. No retention ones, because that needs a customer history this product deliberately does not hold.

07 / How the figure is kept honest

The same four rules,
on every screen.

Read-only, on every plan

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.

Measured stays measured

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.

Every figure opens

Follow any number to the dated rows underneath it. A figure nobody can check is a figure nobody acts on.

You approve

The product prepares a decision and shows the evidence. Nothing consequential happens until you say so.

Questions

About Recommendations.

Does it act on anything by itself?
No. It prepares the decision and the evidence; the action stays yours. Nothing consequential happens without approval, and nothing is executed on your behalf today.
Where does the impact figure come from?
From the same model as the dashboard. A discount recommendation is priced from the contribution margin it came out of, a fee one from what the processor actually charged.
Why so few recommendations?
Because a gap named is better than a suggestion invented to fill it. The list is as long as the data supports and no longer.
Will it tell me to raise prices?
Where the arithmetic supports it, and with the volume risk stated beside it. It will not tell you what your customers will tolerate, because it cannot know that.

One operating system

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 product

See Recommendations on your own figures.

Ranking needs a common unit and the only honest one is money. Every recommendation carries an expected impact per month, worked out by arithmetic printed on the card — not a score out of ten, not a traffic light.