SEO & GEO
How do you do an SEO audit? The complete step-by-step plan
Copy for AI
You do an SEO audit in seven layers, from technical to strategic.
Each layer answers one question:
- Can Google crawl and index your site?
- How fast and stable is it (Core Web Vitals)?
- Do on-page and content aim at the right keywords?
- Do internal links feed the right pages?
- How strong are your backlinks and authority?
- Where does the competition beat you?
- Which findings come first?
The verdict: a website SEO audit is only as good as its last step. Rank every finding by impact, or you’ve built a to-do list nobody works through.
With Google Search Console, Screaming Frog and PageSpeed Insights you get a long way for free.
Below, each step covers what you look at, an example, and the mistake most people make.
The seven layers build on one another, from purely technical at the bottom to strategic at the top:
What do you need before you begin an SEO audit?
Three free tools and half an hour of preparation.
Connect Search Console first
Make sure Google Search Console is connected to your domain. Do this first.
Verification via DNS can take up to two or three days with some providers. And a new property needs a few days before data appears.
Our guide to Google Search Console walks through the setup.
Download a crawler
Download the free version of Screaming Frog SEO Spider.
It crawls up to 500 URLs per crawl, enough for most B2B sites.
Here’s the product page, with the free limit stated:
Screaming Frog’s product page, October 2026. Free up to 500 URLs; a licence lifts the limit.
Price: the paid licence is €245 a year (Screaming Frog pricing, checked 20 October 2026).
Keep the rest at hand
Three more things complete the kit:
- PageSpeed Insights for the speed measurement
- optional: a free Bing Webmaster Tools account
- optional: one of the SEO tools with a free backlink check
You don’t need more for a first proper audit.
Start with a goal, not a tool
Write down which pages bring in revenue or leads. Those pages weigh more heavily in every step below.
A mistake on your most important service page counts more than the same problem on a forgotten blog article.
Our experience: we always start with a baseline. We connect Google Analytics, Search Console and Bing Webmaster Tools before judging anything.
Step 1: Crawl and indexation: can Google find your pages?
If Google can’t crawl or index a page, the rest of your optimisation doesn’t matter.
Where to look
In Search Console, open the Page indexing report (“Pages” under Indexing).
Look at indexed versus not indexed pages, with the reason per group.
Our walkthrough of the Search Console indexing report explains each reason.
Also crawl your site with Screaming Frog. Filter on the Status Code, Indexability and Canonical columns.
What you look for: four things:
- pages accidentally set to
noindex - important URLs blocked in robots.txt
- 404s that are still linked internally
- canonical tags pointing to the wrong page
Also check that you’ve submitted a valid XML sitemap.
Check how Googlebot actually crawls
The Crawl Stats report shows what Googlebot really hit. Look at the breakdown by response:
Crawl Stats in our own Search Console for customerimpact.be, October 2026. The breakdown by response shows how many requests hit a 301 or a 404.
A rising share of 404 or 5xx responses is a crawl problem worth fixing first.
Our experience: when traffic drops, we first look at crawl stats and Search Console for a clear break date. Then we match that date to site changes.
Example and common mistake
For example: Search Console shows 1,200 not-indexed pages with the reason “URL marked ‘noindex’”.
On inspection, a webshop template sets every filter combination to noindex.
That’s fine, but three important category pages are caught up in it too.
Common mistake: panicking over a high number of excluded pages. Google itself says not indexed “is not necessarily bad”.
Many exclusions are correct: pagination, filters, thank-you pages.
Don’t look at the total; look at whether your revenue pages are indexed.
Step 2: Technical and Core Web Vitals: how fast and stable is the site?
Now that the site is findable, you measure how smoothly it loads.
Measure per page type
Run your main page types through PageSpeed Insights: homepage, service page, blog article.
Note the field scores for LCP, INP and CLS.
Here’s what that field data looks like for our own site:
PageSpeed Insights field data for customerimpact.be on mobile, October 2026. CLS at 0.13 fails the assessment.
Use the Core Web Vitals report in Search Console for the picture across all pages:
The Core Web Vitals report in our own Search Console, October 2026: 349 mobile URLs need improvement because of layout shift.
So even our own site has a finding here. That’s normal; the report tells you which template to fix.
The thresholds
What you look for: three Core Web Vitals thresholds:
- LCP: 2.5 seconds or less
- INP: 200 milliseconds or less
- CLS: 0.1 or less
PageSpeed Insights takes those field scores from the Chrome UX Report over the previous 28 days, at the 75th percentile.
So a fix only shows up there weeks later.
Also check mobile-friendliness, a valid HTTPS certificate, and whether images load in a modern format at the right size.
Tip: our free website speed test gives a quick first read before you dig into PageSpeed Insights.
Example and common mistake
For example: your LCP on mobile is 4.1 seconds. The cause turns out to be an uncompressed 1.8 MB header image.
Compressing it to WebP brings it back to 2.2 seconds, an improvement that touches every visitor.
Common mistake: chasing a perfect score of 100. The biggest gain sits between slow and fast, not between fast and perfect.
Go from red to green, then spend your time on content.
Our guide to improving Core Web Vitals lists the usual fixes.
Step 3: On-page and content: are your pages aiming at the right keywords?
Here the audit shifts from machine to human.
Export and sort
Export all title tags, meta descriptions and H1s from Screaming Frog. Sort on missing, duplicate and too-long titles.
Then manually assess your ten most important pages.
Does the content match the search intent of the keyword you want to rank for?
What you look for: four warning signs:
- missing or duplicate title tags
- pages that aim at no clear keyword
- thin content, such as an 80-word service page
- keyword cannibalization, where two pages fight over one keyword
Example and common mistake
For example: three blog articles all have “seo audit” as their main keyword. Google doesn’t know which to show and ranks all three mediocrely.
You merge them into one strong article and redirect the other two.
Common mistake: rewriting titles on gut feeling, without looking at the actual queries.
Start from the queries a page already gets
In Search Console, open the Performance report, filter per page, and see which search terms that page already gets impressions on:
Queries for one page in our own Search Console, October 2026. Related queries on one URL is the healthy pattern.
That’s your starting point, not guesswork.
Our guide to the title tag and meta description shows how to rewrite from there.
Step 4: Internal links: do you distribute your authority to the right pages?
Internal links tell Google which pages matter, and help visitors move on.
Find the under-linked pages
In Screaming Frog, the Inlinks column under “Internal” shows the internal links to each page.
Sort ascending and you see which important pages get too few.
What you look for: four things:
- service pages with only one or two internal links
- orphan pages with zero incoming links, reachable only via the sitemap
- broken internal links
- vague anchor texts such as “read more”
Example and common mistake
For example: your most important service page has two internal links, while an old blog article has forty.
You add contextual links to that service page from relevant articles.
Google says changes can take anywhere from a few hours to several months to show.
So judge the effect after a few weeks.
Common mistake: linking to everything in the footer and calling that “internal link building”.
A link in the running text, with a relevant anchor, weighs more than the fiftieth identical footer link.
Google’s starter guide notes that anchor text tells users and Google what the linked page is about.
Our experience: we don’t start with backlinks. We usually restructure the site first: internal linking, titles and descriptions.
Step 5: Backlinks and authority: who refers to you and how credible are you?
Authority largely comes from outside: other sites that link to you.
Check your referring domains
Use a free backlink check; most SEO tools offer a limited free version.
Look at your referring domains for quality and relevance, not merely numbers.
Search Console gives you a free first view in its Links report:
The Links report in our own Search Console for customerimpact.be, October 2026. Scroll down to Top linking sites to see every referring domain.
What you look for: four questions:
- how many unique domains link to you
- whether those are relevant, trustworthy sites
- whether you lost valuable links (broken backlinks)
- whether a suspicious network of spammy links points at you
Compare your number of referring domains broadly with your competitors’.
Tool labels aren’t Google’s verdict
An external tool shows more detail, including its own spam labels:
Referring domains for customerimpact.be in our own Ahrefs account, October 2026. The red SPAM tags are Ahrefs’ label, not a Google penalty.
Two odd domains out of 401 is normal noise. Our guide to toxic backlinks explains when it isn’t.
Example and common mistake
For example: your competitor has 60 referring domains from your sector; you have 12. That partly explains why they rank higher.
Your finding: a targeted digital PR and guest article approach to close that gap.
Common mistake: fixating on a single domain authority number. That’s a tool’s estimate, not a ranking factor from Google.
Ten relevant links from your field are worth more than a hundred irrelevant ones.
More on that in our explainer on backlinks.
SEO link audit: internal links or backlinks?
An SEO link audit covers both: the links inside your site and the links pointing at it.
They answer different questions, so check them separately:
| Internal link audit | Backlink audit | |
|---|---|---|
| Question | Do my links feed the right pages? | Who links to me, and is it healthy? |
| Main tool | Screaming Frog (Inlinks, broken links) | Search Console Links report, Ahrefs |
| Typical finding | Orphan page, broken link, “read more” anchor | Lost links, odd spam domains |
| Usual fix | Add contextual links, fix 404s | Reclaim lost links, rarely disavow |
| Under your control? | Fully | Only partly |
Rule of thumb: fix the internal side first. It’s fully in your hands and works on every page at once.
Long redirect chains belong in the same pass: every internal link should point straight at the final URL.
Step 6: Competition: where do they rank higher and why?
An audit in isolation misses context.
Search like a buyer
Choose your two or three most important competitors. Search your core keywords in an incognito window.
Look at which pages sit above you, and why. Is their content deeper, their technical side faster, their authority stronger?
What you look for: three gaps:
- keywords where they rank and you don’t (content gaps)
- page types they have and you lack, such as comparison pages or tools
- the format Google rewards for that search term
Our step-by-step SEO competitor analysis goes deeper on each gap.
Example and common mistake
For example: on “outsource seo audit” a competitor tops the results with a page that names the costs. You only have a general service page.
Your finding: create a specific page that answers that question.
Common mistake: copying the competitor blindly. Look for the why behind their position, and do it better, not the same.
Step 7: Prioritise by impact: from findings list to action plan
This is the step that makes an audit valuable.
Score every finding
You now have dozens of findings. Without prioritisation, that’s a to-do list no one works through.
Score each finding on three axes:
- Impact: how much traffic or how many leads this delivers
- Reach: how many important pages it touches
- Effort: how much work it is
Formula: impact times reach, divided by effort.
What you look for: the handful of interventions with high impact and low effort. In most audits a small share of the fixes delivers most of the result.
Those go to the top.
Common mistake: working in the tool’s order, from top to bottom. A tool sorts on the severity of the “problem”, not on business impact.
You prioritise on revenue.
Worked example: how do you rank SEO audit findings?
Say your audit delivers four findings. Here’s how you rank them:
- Three revenue pages are on noindex (impact high, effort low). One line of code, direct effect on your most important pages. Priority 1.
- Most important service page has only 2 internal links (impact high, effort low). A few links from existing articles. Priority 2.
- LCP on mobile is 4.1 seconds (impact medium, effort medium). Compress images, touches all visitors. Priority 3.
- 200 old tag pages have thin content (impact low, effort high). Touches no revenue pages. Priority 4, or simply skip it.
Here are the same four findings on an impact and effort grid:
The tool will probably put finding 4 at the top as “200 problems”.
You put it at the bottom, because it delivers nothing.
That’s how an audit that delivers results differs from one that only shows a score.
First clear the blockers, then the wins with the most leverage.
How do you audit a single URL for SEO?
Start with the URL Inspection tool in Search Console, then check the page itself.
A single-URL audit is the quick version of steps 1 to 4, for one page that underperforms.
First, ask Google what it sees
Paste the URL into the search bar at the top of Search Console.
The URL Inspection tool tells you whether the URL is on Google. It also shows which canonical Google picked and whether robots.txt allows the crawl.
Watch out: the default view is the last indexed version. Click “Test live URL” to see the page as it is now.
Then run five page checks
Work through five checks on the page:
- Indexable: no
noindex, a 200 status, the right canonical - Speed: the URL’s field data in PageSpeed Insights
- Title and H1: both carry the query the page should win
- Intent: the content matches what the SERP rewards
- Links in: at least a few contextual internal links
The Performance report then shows which queries that URL already gets (as in step 3).
Tip: Google sets a daily limit on inspections and indexing requests per property. For dozens of URLs, submit a sitemap instead.
Use a single-URL audit before you rewrite a page, not after.
What are the common mistakes in an SEO audit?
Four mistakes undo most audits:
- Optimising for vanity numbers. Indexed pages or a score of 92 say nothing about revenue.
- Seeing the audit as a one-off. A cleaned-up site slowly decays. Repeat the core checks every quarter.
- Focusing only on technical. Tools miss whether you aim at the right keywords; that needs a human eye.
- No owners. Without an owner and a deadline per finding, your action plan stays put.
For a deeper technical pass, use our technical SEO checklist.
Checklist: SEO audit step by step
Tick these off in order:
- Search Console connected, indexation report reviewed
- Site crawled with Screaming Frog, status codes and indexability checked
- Core Web Vitals measured with PageSpeed Insights (LCP, INP, CLS)
- Mobile-friendliness and HTTPS checked
- Title tags, meta descriptions and H1s assessed on your top 10 pages
- Search intent and cannibalisation checked
- Internal links to revenue pages inventoried
- Backlink profile and referring domains reviewed
- Two or three competitors compared
- Multilingual sites: hreflang, URL structure and redirects checked
- All findings scored on impact, reach and effort
- Top 3 actions with owner and deadline recorded
How do you do an international SEO audit?
An international SEO audit checks whether every language or country version reaches the right searcher.
It adds three layers to the seven steps: URL structure, hreflang and redirects.
Check the URL structure
First note how the versions are split. Is it a country domain, a subdomain or a subdirectory?
Google’s guide to multi-regional and multilingual sites weighs the options like this:
| Structure | Example | Strength | Weakness |
|---|---|---|---|
| Country domain (ccTLD) | example.de | Clear geotargeting | Costly, one country per domain |
| Subdomain | de.example.com | Easy setup, flexible servers | ”de” can mean language or country |
| Subdirectory | example.com/de/ | Easy setup, low maintenance | Geotargeting less clear from the URL |
| URL parameter | example.com?loc=de | None worth having | Google doesn’t recommend it |
Our setup: customerimpact.be runs Dutch, French, English and German in subdirectories on one domain.
The audit doesn’t ask which structure is best. It asks whether yours is applied consistently.
Check hreflang on every version
Then crawl each language version. Export its hreflang tags.
Google’s hreflang documentation sets the rules you check against:
- Return links: if two pages don’t point to each other, Google ignores the tags
- Self-reference: each version lists itself and all other versions
- Valid codes: ISO 639-1 language, optional ISO 3166-1 region (so
en-gb, neveren-uk) - Fallback: Google suggests an
x-defaultfor languages no version matches
Here’s how the checks stack up, from blocker to nice-to-have:
Our guides to hreflang and fixing hreflang errors show the fixes per error.
Check redirects and language detection
Finally, open each version from a different country setting.
Google advises against automatically redirecting users from one language version to another. Those redirects can keep users and search engines from seeing all versions.
So a language banner beats a forced redirect.
Watch out: Googlebot mostly crawls from the US. Content that changes by IP can hide your other versions from Google.
For the strategy behind it, read our guide to international SEO and our multilingual website strategy.
How do you also become the source AI cites?
The same foundations make you citable in AI answers.
A site that’s technically clean, clearly structured and answers the right questions isn’t only found better in Google.
It’s also what generative search engines need to cite you.
They need pages that are crawlable, unambiguous and give a clear answer per page.
That’s the core of generative engine optimization, and it starts with the same foundations you lay in this audit.
Frequently asked questions
How long does an SEO audit take?
It depends on the size of the site.
Preparation takes about half an hour, but Search Console verification via DNS can take two or three days, and Google says changes can take months to show.
Which tools do you need for an SEO audit?
Google Search Console, Screaming Frog (free up to 500 URLs) and PageSpeed Insights cover most of it.
A backlink tool and Bing Webmaster Tools are optional.
How often should you do a website SEO audit?
Repeat the core checks every quarter. A cleaned-up site slowly decays again.
What is an international SEO audit?
An SEO audit for a site with several languages or countries. On top of the usual checks, it tests URL structure, hreflang return links and language redirects.
What is the most important step in an SEO audit?
Prioritisation. Rank every finding on impact times reach, divided by effort, so the fixes that bring in customers come first.
Want your audit turned into a plan?
Don’t want to do the audit yourself, or want your findings turned into a prioritised plan with expected impact?
Take a look at our SEO service. Get in touch and we’ll go through your site together.
Free website scan
Enter your website and get an automatic scan within minutes, with concrete technical and SEO improvements. No sales pitch.
We only use your details for your scan. No spam, unsubscribe anytime.