Guide
An SEO audit template that gets acted on
Most audit reports are a list of everything wrong with a site, sorted by category, delivered as a spreadsheet nobody opens. This is a structure for the other kind - the one that produces work.
Why most audit reports get filed instead of actioned
There are two audiences for an SEO audit and they want opposite things. The person who has to implement it wants a complete list. The person who has to approve it wants to know what it costs, what it earns, and what to do on Monday. A report that serves only the first reader gets filed; a report that serves only the second is not an audit.
The structure below serves both by separating them: a short section that could be read alone by someone who will never touch the site, and a long appendix for whoever will.
The audit that gets filed
Sorted by category. Every issue given equal weight. Findings phrased as observations - "1,240 pages lack meta descriptions". No sizes, no owners, no arithmetic. Arrives as a 60-tab spreadsheet.
The audit that produces work
Sorted by what it pays. Three to five findings stated as consequences, each with the value at stake and the arithmetic shown. Every action has an owner type and a rough size. The first page survives being read alone.
The six-section structure that serves both readers
Six sections, in this order, and the order is the argument: findings before fixes, impact before detail. Somebody who reads only the first two sections should be able to approve the work; somebody who reads only the last three should be able to do it.
- 1
1. What we found, in one page
Three to five findings, each stated as a consequence rather than an observation. Not "1,240 pages lack meta descriptions" but "the pages that earn the most traffic are being shown to searchers with text Google wrote itself, and we can write it instead." If a finding cannot be stated as something the business loses, it belongs in the appendix.
- 2
2. What it is worth
For each headline finding, the traffic or revenue at stake, with the arithmetic shown. Ranges are honest and more persuasive than false precision. This is the section that decides whether anything happens, and it is the section most audits skip.
- 3
3. What to do first
A ranked list of actions, each with an owner type (developer, writer, someone with CMS access) and a rough size. Ranking is by impact over effort, not by category - a title-tag rewrite and a server-response fix belong next to each other if that is the order that pays.
- 4
4. Technical findings
Crawlability and indexation first, because a page that cannot be indexed cannot be improved. Then response codes and redirects, canonicals and duplication, site speed on real devices, structured data, and mobile parity. Each with the affected URL count and how it was measured.
- 5
5. Content and intent
Which pages target which searches, where two pages compete for one query, where the intent of the ranking page does not match the intent of the query, and which topics the site has no page for at all. This section produces the most work and the most revenue.
- 6
6. Authority and competition
Referring domains against the sites currently outranking them, what those sites have that this one does not, and which link gaps are realistically closable. Be honest when the gap is not closable - it changes the strategy rather than the tactics.
What to check, in the order that leaves you something usable
This is the working list, and its order is doing a job. Audits get interrupted. Ordered this way, an audit you abandon halfway through is still a deliverable, because the checks that gate everything else are already done and the ones left over are refinements.
Indexation gates content, content gates authority. Work down, not across.
Indexation - do this first
- Which pages are indexed against which should be, and the gap in both directions.
- robots.txt and meta robots: nothing important blocked, nothing private open.
- XML sitemap present, current, submitted, and free of redirects and 404s.
- Canonical tags: self-referencing by default, pointing somewhere sensible when not.
- One canonical host, with everything else redirecting to it in one hop.
Crawl health
- Internal links to 404s and to redirect chains longer than one hop.
- Orphan pages that exist but nothing links to.
- Crawl depth: how many clicks from the home page to the money pages.
- Parameter URLs and faceted navigation generating near-infinite crawl paths.
Content
- Two or more pages targeting the same query, competing with each other.
- Title tags and H1s written for a person rather than a keyword list.
- Thin pages: under a couple of hundred words with nothing unique on them.
- Pages that rank on page two - usually the cheapest wins in the whole audit.
- Topics the competition covers and this site does not.
Speed and experience
- Core Web Vitals on the templates that matter, measured on mobile.
- Images sized for their container and served in a modern format.
- Render-blocking resources on first paint.
- Layout shift from banners, cookie notices, and late-loading fonts.
Put a number on every headline finding
This is the section most audits skip and the one that decides whether anything happens. You do not need precision. You need arithmetic somebody can follow and disagree with, which is a much lower bar than being right.
Take a finding and walk it to money in public. A page ranking fourth for a term with a known monthly search volume, at a click-through rate you state rather than assume, at the site's own measured conversion rate, at their own average order value. Show every number and where it came from. Give a range rather than a point, because a range is honest and a point invites an argument about the wrong thing.
Two rules keep this from becoming fiction. Use the client's own numbers wherever they exist - their conversion rate, their order value, their close rate - because a figure they recognize is one they will not spend the meeting challenging. And when a finding cannot be walked to money, say so and put it in the appendix rather than inventing a value for it. An audit with four costed findings and forty uncosted ones is more credible than one where everything has a suspiciously round number attached.
The uncomfortable version of this rule: if you cannot state what a finding is worth, you may not actually know whether it matters. That is worth discovering while writing the report rather than in the meeting.
Rank the fixes by what they pay, not by category
The instinct is to group findings by type, because that is how the tools export them. It reads well and it produces the wrong order of work.
Grouping by category tells the client to do all the technical work, then all the content work, then think about links. That ordering has no relationship to what earns anything. A title rewrite that takes twenty minutes and a server-response fix that takes two days belong next to each other in the list if that is the order in which they pay.
Rank by impact over effort and publish the ranking as a single list, with the category as a label rather than a heading. Put the page-two rankings near the top: pages already sitting between eleventh and twentieth are usually the cheapest wins in the entire audit, because the site has already done the hard part and the remaining distance is small.
Then draw a line after the first five and say plainly that the things below it can wait. An audit that ranks eighty items and recommends all of them has not ranked anything.
Write recommendations somebody can pick up and start
The gap between a finding and a recommendation is small on the page and enormous in what happens next. It comes down to whether the sentence names a person and a size.
The difference between a recommendation and a finding is that a recommendation names who does what. "Improve internal linking" is a finding wearing a verb. "Add links to the six service pages from the three highest-traffic blog posts, listed in appendix C - about an hour in the CMS" is a recommendation.
Size everything, even roughly. An audit with no effort estimates forces the client to guess, and the guess is always "more than we can afford right now". Sizes also protect you: an hour-long task that has been quietly assumed to be a week is how audit recommendations get declined without anybody saying no.
Name the kind of person rather than the person. "Developer", "someone with CMS access", "a writer who knows the product" - because you rarely know their org chart, and a task assigned to nobody is a task nobody starts.
Deliver it so it survives the meeting you are not in
Your audit will be forwarded, skimmed on a phone, and discussed by people who have not read it while you are not in the room. Almost nothing about how audits are usually delivered accounts for that.
Four delivery habits that survive being read without you.
- Make the first page complete on its own. Assume it is the only page that gets read, and that whoever reads it is deciding.
- Deliver it as a page, not an attachment. A spreadsheet requires a decision to open; a link opens itself, and you can tell whether anybody looked.
- Put the date and the method on every finding. Six weeks later, "is this still true?" is the first question, and an audit that cannot answer it gets re-commissioned rather than actioned.
- Write one sentence per finding that a non-specialist could repeat out loud. That sentence is what actually travels through the organization; everything else is support for it.
The free audit, and why you can only do a few
Most agencies give away a version of this to win work, and they are right to: a specific read on someone's own site is the most persuasive sales material that exists. The trouble is arithmetic.
A real audit takes hours. Give one away and you have spent a morning on a prospect who may not answer the phone. So the free audit gets rationed to the inquiries that already look promising, which means the prospects who most needed convincing never receive the thing that would have convinced them. Or it gets thinned into a tool-generated score that says nothing specific, and the prospect can tell.
The way out is to separate collecting from judging. A short set of questions collects what changes your read - the platform, the spend, what they have already tried, a link to the site. Your judgment is already written down: your process, your pricing, the five things you always find, the things you refuse to do. Put those together and the first read can be prepared for every inquiry rather than the promising few, while they are still paying attention.
The full audit stays the paid deliverable. What changes is that the conversation about buying it starts after they have read your thinking rather than before.
Common questions
- How long should an SEO audit take?
- A useful first pass on a small site is a few hours with the right tools. A full audit of a large e-commerce site is days. Sell the first pass, quote the rest once you know what you are dealing with.
- What should an SEO audit report include?
- A one-page summary stated as consequences, the value at stake, a ranked action list with owners and sizes, then technical, content and authority sections in full. The summary should survive being read on its own.
- How often should a site be audited?
- A full audit annually, or after a migration, redesign, or a ranking drop nobody can explain. Between those, a monthly check of indexation and the pages that matter catches most problems while they are small.
- Should I use an audit template or start fresh each time?
- Use a template for the checks and write the summary fresh every time. Templated findings are obvious to read and are the reason most audit reports feel generic.
- How do I price an SEO audit?
- Price the first pass on the time it takes you and the decision it unlocks, not on the page count. A common shape is a fixed fee for the first pass that is credited against the first month of work if they proceed, which removes the objection that they are paying twice for the same thinking.
- Should the audit include competitor sites?
- Include enough of them to make the authority section meaningful and no more. Two or three sites currently outranking them on the queries that matter is usually enough. A full competitive teardown is a separate piece of work and pricing it as part of the audit is how audits become unprofitable.
- What if the audit finds that the site is basically fine?
- Say so, early and plainly. It is the most trust-building sentence available to you, and it is nearly always followed by a real conversation about what is actually limiting them - which is usually the offer, the market, or the sales process rather than the site.
See it built
Complete example businesses - the knowledge base, the questions, and a finished page you can read.
See what this would look like for your business
Answer a few questions and it pitches the consultation lead magnet it would build for you - the concept, the questions, and what each person would receive. No account, no card.