How to do a content audit without tools: a spreadsheet, 50 pages, and one question

You can do a content audit without tools. A spreadsheet, your top 50 pages by traffic, and one sharp question will land most of the decisions an audit exists to make.

The work happened in a spreadsheet, against one question: does this content help someone make a decision today?

No software made a single one of those calls. People did, and the question is what made the calls consistent.

The tool is not what is missing

The teams I speak to usually say the same thing. They cannot start the audit because there is no licence for a crawler, no analytics integration, and no budget line for either.

That is almost never the real blocker. I have watched teams with excellent tooling produce an inventory of 4,000 URLs and not one decision.

A crawler tells you what exists. It cannot tell you what should. The gap between an inventory and an audit is judgement, and judgement is the part you cannot licence.

A content audit is an assessment of existing content against a question, ending in a decision to keep, change, or remove each page. If the question is missing, no amount of data closes the gap.

What you need to do a content audit without tools

Three things: a spreadsheet, a traffic export, and a question you can answer yes or no.

Your CMS or analytics platform will already give you page-level traffic for the last 12 months. Export it, sort descending, and stop at 50.

That is the whole setup. It takes an afternoon.

The 50-page rule, and why traffic is the right cut

On most government sites a very small number of pages carry the overwhelming majority of visits. The tail is long, and it is mostly content nobody has opened since it was published.

Auditing the top 50 by traffic is not a shortcut. It is the cut that puts your judgement where the public actually is.

One adjustment is worth making. Add any page that carries a legal obligation, a payment, or a safety instruction, even where the traffic is low. Low traffic on a page about how to report a serious incident is a finding, not a reason to skip it.

Traffic is a proxy for attention, not for value. It tells you where to spend limited review time first. It does not tell you whether the page is any good, which is what the question is for.

One question, three columns

Write your question at the top of the sheet before you look at a single page. Then use three columns: keep, change, remove.

The question has to be answerable and specific to the decision you are making. "Does this content help someone make a decision today?" worked for consolidation because everything we assessed existed to support a public decision.

A different purpose needs a different question. After a machinery-of-government change, it might be: can a reader tell who is responsible for this service now? For accessibility remediation: does this page meet WCAG 2.2 AA without a rewrite?

One question, held steady across 50 pages, produces a defensible pattern. Five questions produce five half-finished audits.

Add a fourth column for a one-line reason. You will need it the moment someone senior asks why their page sits in the remove column.

What to measure, and what to ignore

Measure the things that change a decision. Does the page support a real task? Is the information current and correct? Is another page doing the same job? Can someone read it and act without ringing you?

Ignore the things that feel like measurement and are not. Word count, publication date on its own, and readability scores in isolation are all easy to collect and rarely decisive.

A page that runs 200 words, is three years old, and scores badly on a readability tool can still be the best page on your site. The question decides, not the metric.

The Style Manual's guidance on structuring content is a better test of a page than any automated score, because it asks whether a reader can find what they came for.

When you actually need a tool

There are three situations where the spreadsheet stops being enough.

The first is scale. Above a few thousand pages you need a crawler to build the inventory, though you still need the question to audit it.

The second is technical checks no human can do at volume, like finding every broken link or every image without alt text. Automated accessibility checkers catch a portion of WCAG 2.2 failures reliably, and they are worth having for that portion.

The third is recurring measurement. If you are tracking content health quarterly rather than auditing once, the manual version stops being sustainable.

None of those is a reason to wait. Buy the tool when the manual process is straining, not before you have proved the process works.

The question worth asking next

If you opened a spreadsheet tomorrow with your top 50 pages in it, what is the one question you would write at the top?

If you cannot answer that, the tool was never going to help.

Photo by Mostafa Agami on Unsplash