Marketing Spreadsheet Analysis Readiness Review

The number in the cell is not the truth behind it

A marketing spreadsheet looks authoritative by default. Rows are aligned. Columns have headers. Formulas return numbers. A pivot table summarizes thousands of rows into a clean summary. The human brain sees structure and assumes correctness. But spreadsheets do not arrive pre-validated. They arrive with whatever the export produced, whatever the last person edited, and whatever formula drifted when a range was extended but a reference was not.

The Marketing Spreadsheet Analysis Readiness Review exists because a clean-looking workbook can produce wrong answers just as reliably as a messy one. It checks whether the original export is preserved, the table shape is stable, the formula logic is traceable, and the filters are not hiding rows that would change the recommendation. The output is not a formatted report. It is a decision about whether the spreadsheet can carry the weight of the action it is being asked to support.

• Before trusting any number, trace it back one step. Which cell, which formula, which source export produced it. If you cannot trace it, you cannot trust it.
• A spreadsheet that survives a readiness check produces a decision. A spreadsheet that skips the check produces a gamble dressed as analysis.

Never analyze the original. Always preserve it first.

The most reliable insurance against spreadsheet error is a simple rule. Never work from the original export. Make a copy first and keep the raw data untouched. When the original is preserved, every downstream change is auditable. When the original is overwritten, there is no way to know what the data looked like before someone filtered it, added a column, or corrected a value they assumed was wrong.

This rule sounds obvious. It is violated constantly. A team pulls an export from Google Analytics or a campaign platform, opens it, starts cleaning, and saves over the file. Two weeks later, someone asks why the numbers in the recommendation do not match the platform. There is no way to check, because the original is gone and the working copy is the only artifact left.

• Name your exports with dates and sources. A file called 'data.xlsx' is a liability. A file called 'ga4_landing_pages_20260709.xlsx' is an audit trail.
• After every major transformation, save a new version. Cleaned, analyzed, summarized. Each version captures a decision point that can be traced back.

A formula is a promise. Check whether it kept it.

Formulas are the working logic of a spreadsheet. They calculate the numbers that become recommendations. And they break in ways that are hard to spot. A range that was extended for new rows but the formula was not copied down. A VLOOKUP that returns the wrong column because someone inserted a column to the left. A SUM that accidentally spans into a header row and inflates the total.

A Dartmouth study found that 88 percent of spreadsheets contain at least one error, and after six rounds of revision, more than one percent of cells still had issues. The errors that survive multiple reviews are the most dangerous kind. They are subtle enough that nobody catches them and systematic enough that they distort every number that depends on them.

• Spot-check three formula cells against a manual calculation. If the output does not match, the formula logic is broken somewhere in the chain.
• Check whether any formulas reference external workbooks that are not open or accessible. A broken external link produces errors silently and often goes unnoticed until someone needs the file months later.

Filters and pivots can hide the story, not tell it

Filters and pivot tables are the most powerful tools in a spreadsheet. They are also the easiest way to accidentally exclude data that would change the conclusion. A filter applied to the top row hides rows below. A pivot table that summarizes by campaign hides device-level performance differences. A segment that looks clean on the summary tab masks rows that were excluded by a date range that was never documented.

The reviewer should check what is not visible. Expand every filter. Review every pivot source range. Ask whether the rows that are hidden would change the recommendation if they were included. If the answer is yes or even maybe, the filtered view is not ready for a decision. The spreadsheet should show the full picture before anyone draws a conclusion from a partial one.

• Clear all filters and check the total row count. Compare it to the row count in the summary. If the numbers do not match, find out which rows are excluded and why.
• Document every filter and pivot segment in the reviewer notes. A recommendation built on filtered data must name what was excluded and why it does not change the outcome.

Separate what the spreadsheet shows from what it implies

A spreadsheet displays numbers. It does not display context. The most common error in marketing spreadsheet analysis is treating a data point as a reason to act without checking whether the action follows from the data. A campaign shows rising CPA. The spreadsheet says costs are up. The instinct is to cut the campaign. But the spreadsheet did not say whether the CPA rise is from auction competition, seasonal demand shifts, or a landing page that broke last Tuesday.

The reviewer should separate what the spreadsheet proves from what it only suggests. A number in a cell is observed evidence. A conclusion drawn from that number is an interpretation. The readiness review should label which parts of the recommendation are backed by visible data and which parts depend on assumptions the spreadsheet cannot verify.

• Write the recommendation as two statements. What the spreadsheet shows versus what the team believes it means. If the second statement cannot survive without the first, the analysis is grounded.
• If a key number comes from an external source the spreadsheet does not control, flag it. A CSV export imported three weeks ago is not live data. Treat it accordingly.

Common failure modes this review prevents

The most common failure is treating the spreadsheet as a finished report when it is actually a working draft. Someone applied a filter, added a formula column, and shared the file. The next person sees numbers and assumes they are final. Nobody checked whether the filter was still active, whether the formula referenced the right range, or whether the export was complete.

Another failure is relying on pivot tables without inspecting the source range. Pivots summarize data beautifully, but if rows were added after the pivot was created, those rows are not included. The summary looks complete and is not. A third failure is skipping the formula audit. A broken formula in a visible cell gets noticed. A broken formula in a helper column that feeds a summary tab can go undetected for months and silently corrupt every recommendation based on that sheet.

• Sharing a filtered or partially transformed file as if it were the final source of truth without documenting what was changed.
• Building pivot tables and then adding new rows to the source without extending the pivot range.
• Assuming a formula is correct because the output looks reasonable without spot-checking against a manual calculation.

Sample Review Note

The reviewer confirms the original export is preserved and untouched. The working copy has a stable table structure, consistent headers, and no empty rows or columns breaking formula ranges. Three formula cells have been spot-checked and the outputs match manual calculations. Every filter and pivot segment has been expanded, and no excluded rows would change the recommendation if they were included.

What the spreadsheet shows is clearly separated from what the team interprets it to mean. If any source range changes, column is added, or filter is applied after this review, the affected numbers are flagged as unverified. The recommendation stays approval-gated until the reviewer rechecks the affected cells. The spreadsheet is ready. The discipline around it is what keeps it that way.