DIY SEO audit checklist: find what is holding your website back

Start an SEO audit with a small set of important pages. Check whether Google can access them and whether a real customer can enquire. Record what you verified, what failed and what's still unknown, then fix confirmed blockers first. With access ready, 30 minutes gives you a first triage, not a full audit.

What does an SEO audit check?

An SEO audit is an evidence check, not a score generator:

This checklist has 25 checks in five areas. Each check scores up to 2 points:

It's an owner friendly screen for search and enquiry problems. The checklist can't certify a whole site, legal compliance or security. Nor can it promise rankings or revenue.

This page covers search. Our DIY website audit checklist covers the wider design and conversion review. For the bigger picture, see our guide to SEO for small business.

Before you start: scope, access and free tools

Sort out access and your sample before the clock starts. You'll want business controlled access to the right Search Console property, the website and any Business Profile. GA4 helps where it's already set up. Never put passwords in the worksheet.

Access gaps are common. Our audit sample is 100 website audits from May 2023 to September 2026. In about 1 in 4 audits, the business couldn't get owner access to its own property.

If access is missing, record who owns it and the recovery task. Keep going with the public checks. Mark the rest U. Fixing access can take longer than the audit, so start it first. Our setup guide shows how to set up Search Console properly.

Then choose a small sample that represents the site:

• the homepage

• an important service or product page

• a genuine location page, if you have one

• a useful article or category page

• a page you suspect is weak

Record the date, the business model, the main conversion and the device. Note current rankings, traffic and enquiries too. That gives you a baseline before anything changes. Four or five pages can't prove every URL is healthy. Larger shops, JavaScript apps and multi location sites need wider checks.

Free tools cover the first pass. Each one answers a narrow question:

A crawler can widen the check. Free versions usually cap the number of URLs per crawl. Large or repeated crawls sit outside a 30 minute pass.

A free audit tool flags what it can observe. It can't judge whether your service details are accurate or which fix matters commercially. It can't see every rendered state or the quality of your enquiries. So ask what was scanned, when, with which access and how each recommendation was checked.

Don't get taken for a ride by a free scan that ends in a rebuild quote. A sales report isn't independent proof that you need one.

How the scoring works

The score is an Elev8d editorial aid for prioritising work. It isn't a validated ranking model. Use these states:

• 2: the sampled evidence meets the check.

• 1: a confirmed partial or noncritical gap.

• 0: a confirmed failure.

• U: unknown or not checked yet. No points, and no guessing.

• N/A: the check doesn't apply. Write down why.

Give each check a short evidence note. If one check covers several conditions, record the finding that matters most. Don't average away a blocker.

An online only business may not be eligible for a Business Profile. Mark those local checks N/A with the reason. Don't build a fake profile or score five local fails to keep the total out of 50.

That's an editorial workflow, not a Google scoring system. Report four things from it:

Say five checks are N/A and two of the remaining 20 are U. Report coverage as 18/20 and the points out of 36. Don't scale it to 50.

A complete, fully applicable audit can use these exclusive bands:

A band describes recorded gaps in your sample. It doesn't prove strong rankings, commercial viability or urgency.

Some findings override any total:

• the site is down or compromised

• a serious manual action

• an accidental indexing block on a core landing page

• a failed primary enquiry path

A high total can't clear them. An unknown critical check also blocks an all clear until you verify it.

Contain urgent problems first, then investigate. A low score without a critical issue doesn't justify a paid engagement on its own.

Technical SEO checks 1 to 5

Technical checks ask whether Google can reach, read and trust your key pages. This simplified model shows where each check sits:

Each stage is a diagnostic question. Passing one doesn't guarantee the next.

Check 1: are important pages indexed as intended?

Open the Page indexing report, then inspect each sampled URL with URL Inspection. Record the indexed status and the date. Note the canonical too, the main version you want Google to use.

• 2: the intended pages are indexed as expected.

• 1: a confirmed, limited gap.

• 0: a core landing page is excluded by mistake.

• U: you couldn't verify the status.

Not every utility or duplicate URL should be indexed. Don't panic at a long list of excluded URLs. Check the pages that matter. A live test only confirms Google can access a page. It can't show the page is already indexed.

A site: search can turn up examples, including pages exposed by accident. Google says its results aren't always exhaustive. The count isn't an index inventory. A missing result doesn't prove exclusion either.

A common mistake is comparing a site: count with your number of real pages. Treat zero impressions as a reason to investigate, not proof of failure. Our Search Console guide shows how to use Search Console to investigate a URL.

Check 2: can Google crawl and render the important content?

Inspect the current response first. Where it matters, check the rendered main content too. Record the intended robots rules, working links and any redirect chain or loop. Compare the declared and Google selected canonicals for suspicious duplicates.

Score 2 when the sample shows no real defect. Score 1 for a confirmed, limited defect. Score 0 when key content can't be fetched or rendered. A missing self referencing canonical isn't a failure on its own. Google says none of the canonical methods are required.

Three distinctions matter here:

• Canonical: a strong signal, not a command. Google can choose another URL.

• robots.txt: it controls crawling. Google says it's not a way to keep a page out of Google. Use a noindex tag or a password for that. Noindex tells Google to leave a page out of results.

• JavaScript: content a person sees may be missing from the rendered HTML. Google can't index what isn't rendered.

The live test and the indexed view answer different questions. The live test can't predict the canonical Google will choose. Stop at a documented issue and hand the fix to the technical owner.

Review the sitemap here too. It helps Google discover pages. Submitting one doesn't guarantee indexing, though. A missing sitemap isn't a disaster on its own. Leave navigation and internal links for check 9, so one broken route isn't scored five times.

Check 3: is there a material performance problem?

Run PageSpeed Insights on the homepage and a key landing page. Record the mobile results, plus desktop where it's relevant. Keep the field assessment and the lab run apart:

Field data covers the previous 28 days. It describes the exact URL, or the whole origin (your whole site) when the page lacks data. The lab run is a current simulation. They can disagree.

Google judges field data at the 75th percentile. At least three in four visits need to meet each threshold:

Anything in between is amber. That means needs improvement. Our guide helps you understand Core Web Vitals in more depth. Google's Lighthouse bands put 0 to 49 as poor and 50 to 89 as needs improvement. Scores of 90 to 100 count as good.

The field Core Web Vitals are what Google's ranking systems use. No lab score is a ranking pass mark.

Score the evidence you have and state its scope. A repeatable serious delay or interaction failure on a key page is a confirmed problem. One low lab score needs diagnosis, not a 0. Missing field data isn't a failed test either. Follow the tool's own assessment and don't make up a missing value.

After a fix, a better lab run doesn't replace 28 days of field history straight away. Don't chase a lab score of 100 while the enquiry form is broken. Our speed guide shows how to diagnose a slow website.

Check 4: can a customer use the site on a real phone?

Pick up a real phone and complete the main customer task. Compare it on a desktop where that's useful. Test:

• navigation and readable content

• tap targets and forms

• booking or purchase steps

• whether important content and links are there

Score 2 for a successful task and 1 for a confirmed friction point. A blocked task scores 0. Record the actual failure, not a generic mobile score. A call to action doesn't need to sit above every possible fold to pass. Our guide to mobile SEO covers deeper diagnosis.

Check 5: is there a security problem or manual action?

Check HTTPS on key pages and look for browser warnings or obvious compromise. Then open the Security issues report and the Manual actions report.

The security report covers hacked content, malware and social engineering. Its sample URLs aren't a complete list. A manual action means a human reviewer at Google found pages breaking its spam policies. Affected pages can rank lower or drop out of results.

Score confirmed defects by impact. Use U if you can't open the reports. Don't assume all's well. A clean report and HTTPS are useful evidence. They aren't a security audit.

Active compromise or a manual action on a primary page is an escalation, whatever the total. Don't remove security software, edit server rules or bulk delete URLs to make a warning go away. Record the message, the affected pages, the time and the business impact. Then get the right technical help.

Retest technical fixes properly

Any change that could break something needs a simple record. Take an accidental noindex on a public service page:

• Change: the authorised developer removes it and records the old setting.

• How you'll know it worked: someone retests the live response and the indexing permission.

• Monitoring: indexed status gets checked separately over the following weeks.

• Recovery: the old setting stays on file in case the change breaks something.

Don't mark rankings restored just because a setting changed. The fix is done. The outcome is still unknown.

Page SEO checks 6 to 10

Page checks look at what people and Google see first on each sampled page.

Check 6: does each important page have a clear, accurate title?

Inspect the actual title element, not the shortened browser tab. It should describe the page. Watch for confusing duplicates or repetition. Google generates title links automatically from several sources. So a different search result isn't proof the HTML is broken.

Score a demonstrated clarity problem, not a character count. Google sets no length limit. It truncates long titles to fit the screen.

That example is made up. Only name a service and area the business genuinely covers. Don't add same day service, accreditation or results you can't back up. Our meta title and description checker lets you edit pasted titles with an example preview. Google can still rewrite what shows.

Check 7: does the main heading identify the page?

Use one clear main heading, the H1, as an editorial convention. Then check the visible heading and the page structure. Google's SEO starter guide sets no ideal number of headings. It doesn't set an H1 rule either. So don't score extra H1s as a failure on their own.

Read the main heading aloud. Could a stranger tell what the page is about? If not, rewrite it, even when a tool passes the markup.

Check 8: do headings help people navigate?

Read the headings without the paragraphs. Together they should explain the service, the process, the limits and the next step. Keep a sensible hierarchy for readers and assistive technology.

Familiar labels like "Our services" aren't wrong everywhere. Score whether this page can be understood and navigated. Then rewrite the specific vague or missing heading.

Check 9: do useful links reach the important pages?

Follow the navigation and the links inside the copy across your sample. Look for broken destinations and important pages missing from normal navigation. A crawler can extend this review. But a small sample can't prove there are no orphan pages, the ones nothing links to.

The most important page doesn't need the biggest raw link count. Add useful links without quotas. Our guide to site architecture and internal linking covers the structure.

Check 10: does the main enquiry path actually work?

With the business's permission, send a clearly labelled test enquiry. Confirm it lands in the intended inbox or customer system. Then check someone responsible can reply. Test phone routing or booking too, where they matter.

Score 0 for a confirmed failure, even when a success message appears. A browser message or an analytics event alone doesn't prove delivery. Check the recorded measurement event separately. Then exclude the test from your reporting.

Across 100 website audits from May 2023 to September 2026, we sent labelled test enquiries. About 1 in 5 sites with an enquiry form failed the test. The request didn't reach the person who answers it within one business hour.

Our conversion guide shows how to investigate the enquiry path. Our website audit checklist goes deeper on forms and tracking.

Content quality checks 11 to 15

Content checks ask whether each page helps a real customer decide.

Check 11: do service pages answer buying questions?

Check each service page covers:

• what's offered and who it suits

• genuine coverage and cost drivers

• the process and its limits

• a clear next step

A custom service can explain pricing factors instead of a fixed price. Score gaps that stop a reasonable customer deciding.

Try this. Ask someone outside your industry to read the page for a minute. Then ask what you do, where and roughly what it costs. It's a small usability exercise, not a conversion study. Our guide shows how to improve a service page for search.

Check 12: does every sampled page have a useful purpose?

Look for redundant variants and obsolete pages. Watch for location pages that imply services or premises that don't exist.

Don't fail a useful short page for having fewer than 200 or 300 words. The same starter guide sets no magic word count. A common mistake is deleting a page because it had no recent organic traffic. Check its customers, links and conversions first. Seasonal, legal or support roles matter too.

If a removed page has a genuinely relevant replacement, a permanent redirect can help. Where nothing replaces it, a 404 (not found) or 410 (gone) status can be right. Google treats both codes the same way. It drops the URL from its index if it was indexed.

Don't redirect every old URL to the homepage or an unrelated service. Update your internal links too.

Check 13: does the page match the search need?

Compare the query, your actual offer and the current result types. Then check the page delivers what that search needs:

Mixed intent happens. Don't automatically fail an article that targets a buying question. A service page can answer an informational question too. Longer content doesn't guarantee rankings either. Our guide to content strategy helps you plan the gaps.

Check 14: are important claims supported?

Look for verifiable qualifications, accurate process details and approved work examples. Not every business needs testimonials, named clients or before and after photos. Regulated services may face advertising and confidentiality limits. A missing proof format isn't a failure if the claim is backed another way.

Never fake evidence to lift the score. Our guide to E-E-A-T for small business covers evidence people can check.

Check 15: is the information still accurate?

Review key pages for anything that changes over time:

• services, fees and locations

• staff and contact details

• legal references and screenshots

Give each important update an owner.

A fresh Updated date without a real review doesn't fix stale content. Score known errors by how much they matter. Note which pages you didn't check.

Quick content fixes to look for:

• Thin service page: add the buying details customers ask about.

• Outdated fees or staff: correct them and name an owner.

• Near identical suburb pages: consolidate them, or add real local information.

Local SEO checks 16 to 20

Local checks only apply to businesses that can have a real local presence. Google ranks local results on relevance, distance and prominence. Completing a profile doesn't move your real location or guarantee a rank. An incomplete field doesn't make you invisible either.

Read the full eligibility rules first. Questions about your website's service area content can still apply either way.

Check 16: is the Business Profile eligible, controlled and accurate?

Start with the business model, not completeness. Google requires in person contact with customers during stated hours. Online only businesses aren't eligible. Mark the profile checks N/A with that reason:

For an eligible profile, confirm an appropriate owner, a truthful name, hours and a working contact route. Photos need a useful purpose, not a quota. Our profile guide helps you check profile eligibility and setup.

Check 17: do categories and services describe the real business?

Check the primary category and listed services against what you genuinely offer. Google's representation guidelines say to use the fewest categories that describe the core business. Never use them as keywords.

A common mistake is copying a competitor's categories. Theirs can prompt a look. They don't justify a category you don't meet. No category choice comes with a guaranteed improvement.

Check 18: is the review process genuine?

Start with the request. It should be neutral, optional and sent to every eligible customer. Check responses are appropriate and recurring problems get fixed. Record review count and recency as context only.

There's no minimum count, weekly quota or need for customers to name a service. A burst of reviews alone doesn't prove fraud. Google's review policy bans incentives and selectively asking happy customers. Our reviews guide shows how to request and manage Google reviews responsibly.

Check 19: are customer facing business details accurate?

Check the website, profile and important directories for wrong or old phone numbers, locations and names. Search your old number, address and business name to find outdated listings. Harmless formatting differences aren't a failure.

Service area businesses should hide their address on the profile, under the same guidelines. Don't expose a home address across directories just to make the details match. Our guide to local citations covers listings in more detail.

The local SEO readiness checker is a self assessment. It records your answers rather than scanning anything.

Check 20: are location and coverage pages useful and truthful?

Review the pages you have and the real need for new ones. You don't need one page per suburb. A missing location page isn't a failure if existing content explains where you work.

Check real serviceability, useful local information and links. Watch for any page that implies a branch you don't have. Doorway pages are built to catch similar searches and funnel people to a less useful page. The risk depends on purpose and behaviour, not a shared template alone. Our local guide helps you plan genuine local coverage.

External trust checks 21 to 25

This area covers outside evidence and customer facing credibility. It isn't a pure backlink score.

Check 21: what genuine external links can you verify?

Open the Search Console Links report and check relevant examples. Google says it isn't a comprehensive list of every link. Record how useful and authentic the observed links are. Note real opportunities too.

Don't fail a new business because few links appear. A government or education domain isn't automatic proof of a useful endorsement.

Check 22: is there evidence of manipulative link activity?

Review known purchased links and contracted placements in context. Judge patterns, not single odd links. Google's spam policies treat links bought for ranking as link spam. Paid links are fine when they're properly qualified.

A foreign language, a low tool score or an unfamiliar domain doesn't prove a harmful scheme. Mark U when you don't know where a link came from. A disavow file asks Google to ignore specific links:

Google's disavow guidance limits it to many spammy, artificial or low quality links. Those links must have caused a manual action, or be likely to. Used wrongly, a disavow can harm performance. A common mistake is uploading one because a tool labelled links toxic.

Check 23: is accurate independent information available?

Search your brand name and read relevant mentions, association records or coverage. Mentions help customers verify you. That's different from claiming every unlinked mention is a ranking signal or an AI citation guarantee.

A new or specialist business may have a small, legitimate footprint. Score a documented opportunity or accuracy problem in context, not a platform count.

Check 24: are the profiles you actually use maintained?

Check the directories, social profiles and map services your customers actually use. Look at ownership, accurate identity and working links. This is a management question, separate from check 19's core contact details.

You don't need Instagram, LinkedIn and every directory. Mark an unused, irrelevant channel N/A rather than creating work for points.

Check 25: can customers verify the business and its claims?

Review your About and contact information, relevant credentials and ownership details. Check that policies are disclosed properly too. Then compare any structured data with the visible facts.

Google says markup must be a true representation of the page. Optional markup isn't needed for basic indexing. Passing the Rich Results Test doesn't guarantee a rich result either. Our guide helps you choose and check schema markup.

Count a missing credential once, even if several checks touch it.

A worked score of 31 out of 50

This example is a service business, with all 25 checks applicable and assessed. The numbers are teaching inputs, not an Elev8d client result or a benchmark. Two problems stand out. A core service page carries an accidental noindex. And the contact form shows a success message without delivering the enquiry.

The total sits in the 30 to 39 band. Checks 1 and 10 still override it. Both are critical failures.

Check 2 scores 2 while check 1 scores 0. Google can fetch and render the page. The accidental noindex only stops it being indexed as intended.

Check 5's single point is a documented maintenance gap, with no active compromise. Check 22's point is one sponsored listing with an unqualified paid link and no manual action. It needs correcting and a scope review, not a disavow.

The fixes, in order:

1. Restore enquiry delivery. Run a successful labelled test, confirm receipt and assign a responder.

2. Have the authorised owner remove the accidental noindex. Verify the live change, then monitor indexed status.

3. Improve the main service page's buying information, using the gaps from checks 11 and 15. Then confirm it still works on a phone.

Fixes 1 and 2 can run in parallel where the owners allow. Both override the reassuring 31. Keep the performance and link findings in the issue record for later follow up. Don't add articles or buy links just to raise the total.

If only checks 1 and 10 later move from 0 to 2, the total becomes 35/50. That illustrates the arithmetic. It doesn't prove recovered rankings or extra customers. Don't report an after score until its evidence is verified.

Your 30 minute first pass

Thirty minutes is an example triage, once access and the sample are ready:

Not everyone can verify 25 checks in that time. Score first and fix later. Mark anything unfinished U and come back to it.

If you find active compromise or a broken enquiry route, stop the clock. Escalate it before you finish a neat score.

Printable SEO audit worksheet

Print this page or copy the tables below into a document or spreadsheet. A printed copy won't add up totals for you. Use your scope record from the start as the header. Leave the worked example's scores out of your copy.

State key: 2 is verified in the sample, 1 a confirmed partial gap and 0 a confirmed failure. U means not verified. N/A means not applicable, with a reason. Keep notes factual. Don't store customer names, passwords or confidential submissions in a shared worksheet.

Then fill in the summary:

Only a complete, fully applicable audit of all 25 checks gets a band out of 50. Keep three things apart: the score, how sure you are of the evidence and the business impact. The worksheet is a work record, not a certificate.

Give any finding that needs more room its own record:

For a repeat audit, keep the previous date, sample and scoring rules. Broaden the sample on purpose and say why. A higher score from checking easier pages isn't an improvement. Report completed fixes separately from business outcomes.

Common mistakes and when to get help

The worst audit mistakes come from treating every finding as equally urgent:

Confirmed critical technical defects deserve prompt attention. Technical issues don't always come first, though. A harmless metadata warning shouldn't outrank a misleading offer or a failed form. Weigh each task on:

• its impact on customers and enquiries

• how many important pages it affects

• the strength of the evidence

• its dependencies and how easily it can be reversed

Don't stop all worthwhile work because one unrelated check is incomplete.

If you pay for an audit, ask for:

• the scope and the pages sampled

• the evidence behind each finding

• actions in priority order

• who implements each fix

• how each fix gets retested

A free scan isn't a diagnosis. Our guide to SEO red flags covers the sales tactics to watch for. Our packages guide explains what an SEO package should include.

Get help for active security problems, serious manual actions or migration risk. Repeated indexing failures count too. So does a broken customer path. If a redesign is coming, our website migration SEO checklist covers the risks. A narrowly scoped developer fix may be enough.

The score helps organise findings. It isn't the trigger to buy. A paid SEO audit from us is a full review with a prioritised roadmap you keep.

What we recommend at Elev8d

Frequently asked questions

Next steps

Written by Ajay K.Co-founder, Elev8d

Ajay is the co-founder of Elev8d and writes most of what you will read here. Psychology grad turned marketer, which is a fancy way of saying he cares about why people click, not just where things rank. Every guide is written in plain English for Australian businesses, tested on real campaigns before it earns a spot on this site.