Your first 30 days of profit tracking

Week 1 — Establish the truth about one order

Nothing else is meaningful until you know your contribution margin. Spend this week on cost of goods and nothing else. It is the least interesting week and the one that determines whether the other three produce anything.

1. 1Build landed cost for your top 20 SKUs by revenue. They will cover most of your volume.
2. 2Pull real payment fees from a settlement report, not the rate card.
3. 3Get fulfilment cost per order, including shipping you subsidise.
4. 4Compute return rate by category from order data, and what a return actually costs.
5. 5Derive break-even ROAS: 1 ÷ contribution margin.

Week 2 — Find out how wrong your current numbers are

Now size the gap between what the platforms claim and what actually happened. Do not change any budgets this week. You are measuring, not fixing.

1. 1Total real orders for last month from your shop or processor.
2. 2Add up reported conversions across every ad platform for the same period.
3. 3Divide the second by the first. That ratio is your over-attribution.
4. 4Set every platform to the same click-based attribution window, and turn off view-through.
5. 5Re-read your campaign performance with those settings. Note which campaigns fell hardest.

Week 3 — Find the single worst leak

One leak, not a list. A list produces no action; one clear loser produces a decision.

1. 1Export the last 90 days at ad or keyword level, with spend and orders.
2. 2Compute profit per campaign: orders × contribution margin − spend.
3. 3Sort ascending. Read the top five losses.
4. 4For the worst one, check whether it is a genuine loss or an attribution artefact — is another campaign claiming the same orders?
5. 5Check the first-time-customer share on it. A campaign buying only strangers must clear break-even on the first order alone.

Week 4 — Change exactly one thing

Resist the urge to restructure. One change, measured properly, teaches you more than ten changes made at once — because ten simultaneous changes cannot be attributed to anything.

1. 1Pause the worst leak, or run a two-week pause test on it if you are not yet convinced.
2. 2Watch total business profit, not the campaign's own reporting.
3. 3If total profit is unchanged or improves, the campaign was not incremental. Keep it off.
4. 4If total profit falls by more than the spend you saved, turn it back on — it was doing more than it could prove.
5. 5Write down what happened. This is the first entry in a decision log that will be worth more than any dashboard within a year.

What month two looks like

By week five you will have a defensible margin, an honest view of your attribution gap, one campaign decision you can justify, and a habit of judging on profit. What you will not have is any of it staying current on its own — every number above decays, and rebuilding the spreadsheet monthly is how this quietly stops happening.