All articles
Guides11 min read

The anatomy of a documentation audit: what to look for

A useful documentation audit looks for four things: gaps, where a process has no documented answer; contradictions, where two documents disagree; outdated content; and documents with no reachable owner. It agrees what "current" means before anyone opens a file, ranks what it finds by how many people each issue reaches, and gives every finding a named owner and a next step.

Anchoris teamNotes from the people building Anchoris

Most documentation audits start the same way. Someone notices a problem, usually because a customer got the wrong answer, a new hire got lost in week one, or an AI tool repeated something that hasn't been true for two years. Leadership asks for an audit. Someone volunteers, or gets volunteered, and opens a spreadsheet.

Six weeks later, that spreadsheet has four hundred rows and "needs review" next to most of them. A plan is presented, a handful of the most obvious problems get patched, and everyone moves on as though the job is done.

This isn't a failure of effort. It's a failure of structure. Most people running their first documentation audit don't know exactly what they're looking for, so they look for everything, which in practice means nothing in particular. A useful audit has a shape. It knows in advance what counts as a problem, how to weigh one problem against another, and what happens to a finding once someone has written it down.

Decide what "current" means before opening documents

The instinct is to dive straight into the shared drive. Resist it. Before anyone touches a document, the audit needs a working definition of good, because "outdated", "wrong" and "incomplete" are all judgement calls, and different reviewers will make them differently without a shared standard.

That standard doesn't need to be elaborate. It might be as simple as: does this page reflect the current process, does it name a real and reachable owner, and could a new starter act on it today without checking with someone? Whatever the criteria, they need to exist before the review starts. An audit built on inconsistent judgement calls produces a spreadsheet that looks thorough and means very little.

It also helps to decide early which places are in scope. The average company now uses 101 apps, according to Okta's Businesses at Work 2025 report, so documentation rarely lives in one place. It's spread across a wiki, a shared drive, a folder inside a folder inside a folder, a few pinned chat messages, and at least one page someone built during a sprint and never told anyone about. Naming the places in scope, and being honest about the ones left out for now, keeps the audit from sprawling indefinitely or missing the systems where the real risk sits.

Coverage: where the gaps are

The first thing to look for is missing information. A gap is any process, policy or piece of company knowledge that should have a documented answer and doesn't.

Gaps are harder to spot than errors, because there's nothing on the page to flag. The most reliable way to find them is to work backwards from the questions people actually ask, rather than forwards from what already exists. Support tickets, onboarding questions from the last few hires, and the things people ask colleagues instead of searching for are all better signs of a gap than scrolling through a folder looking for something that feels thin.

A useful discipline: imagine someone starting tomorrow with no company memory. Could they find out how expenses get approved, who to escalate a particular kind of customer complaint to, or which tool is still in use and which has been retired? If the honest answer is "they'd have to ask someone", that's a gap.

Consistency: where the documents disagree

The second thing to look for is contradiction, which differs from a gap in an important way. A contradiction is two different answers to the same question, both of which look equally official.

Contradictions build up quietly. A policy is updated in one system, but the original in another system is never touched. A process changes after a reorganisation, and the old version stays live because nobody was asked to retire it. Neither document looks wrong on its own. The problem only shows when they're read side by side, or when an AI tool retrieves both, can't tell which is authoritative, and picks one to present as fact.

Finding contradictions usually means deliberately comparing documents that cover the same ground, rather than reviewing each one on its own. Pay particular attention to policy areas where acting on the wrong version has real consequences, especially HR and legal documentation. That's where a contradiction becomes a liability rather than an inconvenience.

Currency: where the documents have gone out of date

The third thing to look for is content that was accurate once and hasn't been touched since. It's often the easiest category to find and the easiest to underestimate, because an outdated document looks exactly as credible as a current one. Nothing about its formatting signals age.

A few practical tells are worth checking systematically: the last edited date; whether it mentions a tool, supplier or team structure that no longer exists; whether its screenshots show an interface that has since changed; and whether the named owner still works at the company. None of these checks needs deep subject knowledge. They just need someone to look, which is exactly the step that gets skipped when an audit tries to cover too much too quickly.

It's worth keeping a specific category for content that still exists, isn't tied to any current process, belongs to nobody, and keeps turning up in search as though it were written last week.

Ownership: where nobody is responsible

The fourth thing to look for is the absence of a clear, reachable owner. This is arguably the category that decides whether the other three ever get fixed. A gap, contradiction or outdated page with a named owner is a to-do item. The same issue with no owner is just a fact everyone works around.

Ownership gaps show up in predictable ways: a document whose listed owner has left, a page created for a one-off project and never handed to anyone afterwards, or content that spans two departments so closely that each assumes the other is responsible. This is the ordinary residue of teams changing, projects ending, and organisations growing faster than their documentation habits.

This is also where most audits reveal that keeping documentation accurate has never actually been anybody's job. It's been an occasional favour from whoever noticed a problem and had a spare afternoon. That's a reasonable way to get by for a while, and not a lasting way to keep information reliable at scale.

What a useful audit produces

The output of an audit shouldn't be a list of problems. It should be a ranked set of actions, each with an owner, because four hundred flagged documents with no ranking and no owner is no better than no audit at all. It just takes longer to produce and feels more thorough.

A workable structure scores each finding on two things: how much it matters if left unfixed, and how many people or systems are likely to come across it. An outdated wiki page nobody has opened in eighteen months matters less than a contradicting HR policy that an AI assistant is serving to every employee who asks a related question.

Every finding also needs somewhere to go. Not a note that "someone should look at this", but a named owner and a specific next step: update the page, merge two contradicting ones, or retire content that no longer serves a purpose. An audit that stops at finding problems and never gets to fixing them has done half the job, and in practice none of the good.

Use this order to run it

  1. Agree what "current" means, and which places are in scope.
  2. Work backwards from real questions to find the gaps.
  3. Compare documents that cover the same ground to find contradictions.
  4. Check dates, references, screenshots and owners to find outdated content.
  5. List documents with no reachable owner.
  6. Rank every finding by how much it matters and how many people it reaches.
  7. Give each one a named owner and a next step, and check again in a few weeks.

Why most audits don't last

Even a well-structured audit tends to have a short shelf life, and the reason isn't effort. Knowledge doesn't stay still. Teams update processes, products change, and new documents are written every day, so the drift an audit fixes starts again the moment it ends.

AI is speeding that up. In GitBook's State of Docs 2026 survey of more than 1,100 documentation practitioners, 76% said they now use AI regularly in their documentation work, up 16 points in a year. More content is being written faster, which is exactly what turns a one-off clean-up into a recurring chore.

The organisations that get lasting value from an audit treat it as the first pass of an ongoing process, not a finished project. The audit sets the baseline. What happens afterwards decides whether the knowledge stays reliable.

How Anchoris helps

Anchoris turns the structure of a good audit into something that keeps running. It connects to the shared drives you choose in Google Drive, or the sites you choose in SharePoint, and checks the documents for three of the four categories directly: contradictions between documents, outdated and expired content, and people named as owners or contacts who have left. It also flags duplicates, unfinished placeholders and broken links. For coverage, it groups your documents into topics automatically, so you can see where coverage is thin.

Where a manual audit ends at a list, Anchoris goes further. Every issue is ranked by how many people can open the document. It suggests who should fix each one, and approved workflows fix issues at the source (move, rename or move to trash), with every run logged. A quality score from 0 to 10 shows where you stand, what's holding the score back, and how it changes over time. First results arrive within an hour, you choose which drives Anchoris can reach, and nothing changes without your approval.

The takeaway

A documentation audit is only as useful as the structure behind it. Looking for gaps, contradictions, outdated content and ownership problems gives reviewers a shared standard instead of a vague sense of what feels off. Turning findings into ranked, owned actions is what separates an audit that changes something from one that documents the mess more thoroughly. And treating the audit as a starting point rather than a finish line is what keeps knowledge reliable long after the review closes.

Want to see what an audit of your own drive would find?

Book a demo

Frequently asked questions

What should you define before starting a documentation audit?

Before opening a single document, agree what "current" means, because outdated, wrong and incomplete are judgement calls that different reviewers make differently without a shared standard. Also decide which places are in scope, since documentation is usually spread across many tools, wikis and folders.

What are the four things a documentation audit should look for?

Coverage: gaps where a process or policy has no documented answer. Consistency: contradictions where two documents give different answers to the same question. Currency: content that was accurate once but has gone out of date. Ownership: documents with no clear, reachable person responsible for keeping them right.

Why are content gaps harder to find than outdated or contradicting documents?

A gap is silence, not an error, so there's nothing on the page to flag. The most reliable way to find one is to work backwards from questions people actually ask, such as support tickets or new joiners' onboarding questions.

Why do contradictions matter most in HR and legal documentation?

Two contradicting documents look equally official on their own, and the problem only shows when they're compared, or when an AI tool retrieves both and picks one. In HR and legal documentation, acting on the wrong version turns a contradiction into a real liability.

How can you tell a document has gone out of date if it still looks credible?

An outdated document looks exactly as credible as a current one. Check the last edited date, whether it mentions a tool, supplier or team structure that no longer exists, whether its screenshots show an old interface, and whether the named owner still works at the company.

What should the output of a documentation audit look like?

Not just a list of flagged documents. Rank each finding by how much it matters and how many people are likely to come across it, then give it a named owner and a specific next step: update, merge or retire.

Why don't documentation audits last, even when they're done well?

Knowledge keeps changing after the audit closes. Processes change, products change and new documents are written every day, so the drift an audit fixes starts again almost immediately unless something keeps checking.

How does Anchoris keep an audit's findings from decaying?

Anchoris checks your documents for contradictions, outdated content, people who have left, duplicates, placeholders and broken links, and groups them into topics so you can see where coverage is thin. It ranks every issue by reach, suggests who should fix it, and fixes issues at the source with approved, logged workflows.

Next article

Your AI pilot worked. Why did it break at scale?

Read next

More from the blog

Guides
Guides9 min read

How to find duplicate files in Google Drive

Google Drive has no one-click duplicate finder. Four manual methods, where they break down, and what it takes to stop copies coming back.

Guides
Guides6 min read

Content audit: a framework for knowledge estates

An inventory tells you what you have. An audit tells you what to do with it. A framework for scoring documents and deciding their fate.

Guides
Guides6 min read

How to clean up Google Drive before Gemini

Gemini reads what your people can open. Six steps to sort out Google Drive before it does, in the order that removes the most wrong answers first.