Why dashboard numbers don't match: 9 common causes

Your dashboard says 1,284 orders for March. Finance says 1,251. Both came from "the same data," and the meeting is now about whose number is right instead of what to do next.

In many cases neither number is a mistake. Each is a correct answer to a slightly different question. The job is to find the difference, name it, and write it down so it doesn't come back next month. Below are 9 common causes, roughly in the order they're cheapest to rule out, with a checklist at the end.

How do you start when 2 numbers don't match?

Before you open a single formula, put both numbers side by side and write down 5 things for each: the definition in one sentence, the date range and time zone, the source and when it was pulled, the filters, and the row count behind it.

Then compare row counts before totals. If the row counts match and the totals don't, look at amounts, joins, and rounding. If the row counts differ, look at dates, filters, and duplicates. Finally, find one specific record that appears in one number and not the other. A single concrete row often names the cause faster than any summary.

Are both numbers counting the same thing?

1. Different definitions

The same word can mean different things in 2 reports. Say the dashboard's "active clients" means clients with a paid invoice in the last 90 days, and the sales report counts any client with an invoice this year. Both are right. They just aren't the same metric.

Tools do this too. In Google Analytics 4, Total users counts everyone who triggered any event, while the Users metric you see in reports is active users, the people who engaged with your site or app. Before comparing 2 user counts, check which one each report uses.

2. Different date fields

An order placed on March 31 and paid on April 2 is a March order by order date and an April payment by paid date. If one report filters on order date and the other on paid date, every order that straddles a month end moves. Say 33 orders were placed in the last days of March and paid in April. That alone could explain the gap between 1,284 and 1,251.

Are both numbers covering the same time window?

3. Time zones

A date depends on where the clock is. An order at 9:30 pm on March 31 in New York is 1:30 am on April 1 in UTC. If one system stores times in UTC and the other in local time, orders near midnight land on different days. Google Analytics shows the effect plainly: if a visitor arrives on a Tuesday in their time zone while it's still Monday in yours, the visit is recorded on Monday. Google also notes that changing a property's time zone only affects data going forward.

4. Partial periods and data delays

"This month" on October 17 covers 17 days, and "last month" covers all of it. Compare closed periods only. Delays matter too. Google Analytics says data processing can take 24 to 48 hours, that report data may change during that time, and that some data can arrive up to 7 days late. A Data Studio report on a Google Sheets source checks for fresh data every 15 minutes by default, so a fix you just made in the sheet may not show yet.

Are both numbers using the same rows?

5. Filters and exclusions

One report leaves out test accounts, internal users, refunds, or cancelled orders, and the other keeps them. Filters like these are easy to set once and forget. Say the dashboard includes 21 test orders and finance excludes them: that explains 21 of the 33. List every filter on both sides, including ones hidden in a pivot table or a saved view.

6. Duplicates

When 2 exports overlap, the same rows appear twice. Say you paste a March 1 to 15 file and a March 10 to 31 file into one tab: March 10 to 15 now counts double. Decide what makes a row unique (usually an order or invoice ID), then compare the count of unique IDs with the count of rows. Microsoft warns that Excel's Remove Duplicates permanently deletes the duplicate data, so copy the original first. Google Sheets treats cells with the same value but different letter case as duplicates, so check before you remove anything.

7. Totals and subtotal rows inside the data

Some exports end with a Total row, and some include subtotals per region or month. A SUM over the whole column then counts those amounts twice. Excel's SUBTOTAL function ignores other SUBTOTAL formulas in its range to avoid double counting, but it can't tell that a pasted row labeled Total is a total. To Excel, it's just another number. Search the label column for "Total" and "Subtotal" before you sum, and keep raw rows on a tab with no summary lines.

Did a join or a tool change the math?

8. Joins that multiply amounts (fan-out)

When you combine a table of orders with a table of order items, each order's total repeats on every item row. Say order A is $100 with 2 items, order B is $80 with 1 item, and order C is $120 with 3 items. The real total is $300. Sum the order total after the join and you get $640. Looker's documentation walks through the same problem with orders and order items, and explains that its symmetric aggregates use a primary key so each value is counted once. In a spreadsheet, it happens when you look up the order total onto every line-item row and then sum that column.

9. Sampling and thresholds in analytics tools

Analytics tools sometimes estimate or hold back numbers. Google Analytics 4 applies data thresholds that withhold data so no one can infer a user's identity, for example when a report includes demographics or low user counts. Google says these thresholds are system defined and can't be adjusted. GA4 can also sample: when a report or exploration needs more events than the quota (10 million for standard properties), results may be based on a sample, and the data quality icon shows the percentage used.

How do you diagnose a mismatch quickly?

Work down this checklist. The early rows are the fastest to rule out, and the last row, rounding, is a quick extra check.

Where Every Number fits

Every Number dashboards are built so the causes inside your spreadsheet (definitions, filters, left-out rows, and the date the data runs through) show up instead of hiding. Each number has a receipt: its definition, the filters applied, the rows used, the rows left out with the reason (totals rows, blank amounts, duplicates), the source file, and the date the data runs through. Combining tables never inflates totals, and the same data always gives the same number. See how it works and the anatomy of a receipt.

Today that's a $99 pilot, built by hand from one spreadsheet. It won't reconcile 2 live systems, such as GA4 and a CRM, for you. If that's your mismatch, the checklist above is the faster route.