Our monthly report is one page. Four numbers, what moved, what we did that relates to the movement, and what happens next. It takes a couple of minutes to read and it is meant to be argued with.
Most reporting is not built that way. A long deck of charts is easier to produce than a short page with a conclusion on it, and length is the oldest way to make a thin month look substantial. If you have ever finished a forty-slide report unsure whether the month went well, that was the format working as designed.
But a short report puts more work on the reader, because there is nowhere to hide and nothing to skim past. So this is how to read ours: what each part is for, what to check before believing a movement, which line should always be there, and the questions that get the most out of us.
The same checks apply to any report you receive, from anyone.
Read the direction before the decimal
Start with the four numbers and ask only whether each went up, down or stayed flat. Resist the second decimal place. Month to month movement in a store's numbers includes a lot of ordinary variation, and precision invites you to react to noise.
What you are looking for is a shape. One number up and the rest flat is a clean result and usually means a specific fix landed. Everything up together is more often a demand story: a good season, a promotion, a mention somewhere. Everything down together is the same story in reverse and rarely means the work stopped functioning.
One number up and another down is the most interesting case, and it is the one to ask about. Conversion up and average order value down can mean a discount did the work. Order value up and orders down can mean a threshold pushed small baskets away.
The report should already say when it sees one of those. If it does not, ask, because the interaction between the four numbers is where the real story usually is.
A movement only counts as evidence when the work is the only thing that changed.
Ask what else changed at the same time
Before crediting any movement to the work, check what else was different about the month. Three things account for most false readings.
Traffic mix. If the share of visitors from branded search or email went up, conversion rate rises without anything on the site improving, because those visitors were always more likely to buy. Our report notes mix shifts, and you should ask if it does not.
Promotions. A discount period lifts conversion and usually lowers average order value and margin. A month containing a sale is not comparable with one that does not, and comparing them is the most common way a report flatters itself.
Season and calendar. Different numbers of weekends, a holiday, a pay cycle, the run-up to a peak. Comparing the same month against the previous year is often more honest than against last month, provided the business has not changed shape in between.
The general rule: a movement is only evidence of the work when the work is the only thing that changed. Where that is not true, the report should say so plainly, and the finding is weaker.
Match the change to the number it should have moved
Every fix in the plan names the lever it belongs to and what we expect it to move. The report should connect those, and you should check the connection.
If we changed the checkout, the number to look at is cart to checkout and checkout to order completion, not store-wide revenue. If we rebuilt a product feed, look at impressions and clicks on the products affected and at the cost of a click, not at total sales. If we built a post-purchase flow, look at the revenue attributed to that flow and at the second-order rate for customers who received it.
Store-wide revenue is affected by everything, which makes it the worst place to evaluate any single change. It belongs in the report as context, not as evidence.
When a change was aimed at a step and the step did not move, that is a clear result and it should be reported as one. When the step moved but revenue did not, that is worth a conversation, because it usually means a different step absorbed the gain.
Look for the part that says what did not work
Every month contains something that did not do what we expected. If a report has no such line, the reporting is being edited or the work was too cautious to learn anything.
So find that section first. It should say what we changed, what we thought it would do, what actually happened, and what we concluded. A failed change with a conclusion attached is worth more than a successful one with no explanation, because it narrows the next decision.
It should also be specific about confidence. Some months the honest answer is that the sample was too small to tell yet, and the fix stays running for another month before judgement. That is a legitimate line and it is different from a failure.
What should not appear is a failure quietly disappearing. If something was in last month's report as in progress and is not mentioned this month, ask what happened to it. The month to month arrangement means the report is the document you renew on, and a report that only carries wins is not a document you can decide with.
The questions that get the most out of us
Four questions will make any monthly conversation more useful, and they work on any supplier.
What would have to happen for you to change your mind about the current priority. This surfaces whether there is a real hypothesis or a preference.
How many orders is this conclusion based on. Small samples produce confident nonsense, and the honest answer is sometimes not enough yet.
If we stopped this piece of work today, what would decay and how fast. Campaign management decays quickly, a flow keeps running, a page change keeps working. This tells you what you are actually paying for.
And what is the next thing you would do if the budget were the same. It is a way of asking whether the plan still has a direction or has become maintenance.
We would rather be asked all four every month. A report that never gets questioned tends to drift towards whatever is easiest to describe, and the questions keep it pointed at the business.
What the report leaves out on purpose
Some numbers are missing from our reporting deliberately, and you should know which.
Impressions, reach and follower counts do not appear as headline figures. They can be useful diagnostically, and they will appear when they explain something, but as monthly headlines they measure exposure rather than outcome.
Ad platform revenue totals do not stand alone, because each platform reports the sales it believes it caused and the sum of those claims usually exceeds what the store actually took. Where we quote platform figures, the report says which platform is claiming what.
Industry benchmarks do not appear at all. We compare your store with itself, before and after, because a comparison against an average of businesses with different products, prices and markets tells you nothing about whether your month was good.
And there are no scores or grades. A number out of a hundred is a summary of somebody's opinion dressed as a measurement.
If you want any of this added, ask. The default is to keep the page to things that can be acted on.
Ten minutes, once a month
Here is a routine that gets almost everything out of a short report.
Read the four numbers and note the direction of each. Read the what changed section and check that each change is matched to the number it should have moved. Read the section on what did not work. Then read what is next and ask yourself whether it follows from the rest of the page. If the next step could have been written before the month started, something is wrong.
Keep the reports. Three or four of them side by side tell you something no single month can: whether the store is trending, which fixes had lasting effects, and whether the plan is still being followed or has quietly become a maintenance routine.
That archive is also what makes the month to month decision real. Every month you can look at what was promised, what happened and what it cost, and decide whether the next one is worth it.