How to Read a Custom Report

A custom report is only useful if you actually know how to read it. Here’s a step-by-step approach that works regardless of what the specific report is measuring.

Step 1: Start With What Question the Report Answers

Before looking at a single number, be clear on what question this report was actually built to answer. A report built to answer “which job type is most profitable” should be read differently than one built to answer “which marketing channel brings in the best customers” — reading either one through the wrong lens leads to the wrong conclusion.

Step 2: Find the Number That Matters Most First

Every custom report has one or two numbers that matter more than the rest. Find those first, before getting distracted by every data point on the page. Everything else on the report should support or explain that headline number, not compete with it for attention.

Step 3: Compare Against a Baseline, Not in Isolation

A number by itself rarely means much. “Margin per job is 22%” is only useful compared against last month’s 22%, or against a different job type’s margin, or against what the business actually needs to be profitable. Always look for the comparison point, not just the raw figure.

Step 4: Look for the Outlier, Not the Average

The average often hides the real story. A report showing decent average performance can still contain one badly underperforming category dragging everything else down — or one exceptional one pulling the average up and masking a problem elsewhere. Scan for the extremes, not just the middle.

Step 5: Decide What Action the Report Actually Suggests

A report that gets read but doesn’t change any decision wasn’t worth building. After reviewing, name one specific thing that should change — a price, a process, a marketing channel — based on what the report actually showed.

Frequently Asked Questions

How often should a custom report actually be reviewed?

This depends on what it measures — operational reports may warrant weekly review, while broader financial or strategic reports are often more useful reviewed monthly, avoiding the noise of too-frequent checking.

What if a custom report shows a number that seems wrong?

Check the underlying data inputs before assuming the report itself is broken — most “wrong-looking” numbers trace back to an input issue, like a missing cost category, rather than a calculation error.

Do I need a finance background to read a custom report?

No — a well-built custom report should be designed to answer plain-language business questions, not require translating financial jargon to understand what it’s actually saying.

This article was written by STELLA, Intelligent Analytics’ AI Intelligence Layer for business data and unit economics.