SEO audit Melbourne: find what is actually holding the site back.
Most audits are a tool export with a logo on the front. This one runs 16 areas across six phases, joins your crawl to Search Console data, models the click upside as a number, and hands every finding over as an instruction with the affected URLs attached. You get a fix list your developer can work through, not a list of observations.
Signs your site needs an audit.
None of these are proof on their own. Two or three together usually mean something structural is wrong rather than something cosmetic.
Traffic slid and nobody can say why
Clicks are down year on year and the explanations on offer are all guesses. A dated, URL-level click comparison tells you which pages lost and when, which is where the real answer lives.
Pages are indexed but get no impressions
Search Console shows the URL as indexable, and it has never returned a single query. That is a different problem from ranking badly, and it needs a different fix.
You migrated or replatformed
New CMS, new URLs, new theme. Redirect chains, dropped canonicals and reset meta are the usual damage, and they are cheapest to fix in the first few months.
AI search never mentions you
ChatGPT, Perplexity and AI Overviews cite the same competitors every time. That is usually a markup, structure and citability problem, and it is measurable.
Good rankings, poor click-through
Positions look fine and the clicks do not follow. Charting CTR by URL group against a 5% baseline shows you exactly which templates are underselling themselves in the results.
Your last audit went in a drawer
It listed problems without saying who does what, in what order, or what a fixed page looks like. That is the difference between a report and a work plan.
16 areas, six phases, nothing skipped.
This is the actual running order, published in full. Clean checks get reported too, because knowing what is healthy is how you know the audit was thorough rather than selective.
URL segmentation
Can you tell what a page is about from its URL? If not, every category-level insight is unavailable until the groups get built by hand. Most auditors skip this and hand over a flat page list instead of a diagnosis.
What gets checked
- Read the URL pattern across a real sample, not the sitemap index
- Find the category XPath in the template when URLs are flat
- Extract published date the same way, then clean it afterwards
- Build URL groups and join them to Search Console
- Report flat structure as a finding in its own right
Why this comes first
Category-level analysis is where the money is. "Service pages convert impressions at half the rate of blog posts" is a decision a business can act on. "Some pages have low CTR" is not.
What it makes possible
- CTR modelled per template, not averaged across the whole site
- Content decisions applied to a group rather than page by page
- Year-over-year loss traced to a section, not just a URL
- Internal linking judged against the pages that need it
Crawlability and indexability
The unglamorous foundation. A site can have excellent content and stay invisible because the two files search engines look for first do not exist or contradict each other.
What gets checked
- Every sitemap variant requested directly, not assumed
- robots.txt rules, including the AI crawler group
- Canonicals on every tested page, and where they point
- Noindex directives that should not be there
- Trailing slash consistency across templates
Response codes and redirects
Every internal link should land on a 200. Where it does not, the useful half of the finding is which page contains the bad link, not the broken URL by itself.
What gets checked
- Internal 4xx errors, reported with their source pages
- Internal links pointing at redirects instead of the destination
- Redirect chains and loops
- Redirects that terminate in a 404 or a noindex page
- Redirect map hygiene on recently migrated sites
Titles and headings
Title length is measured in pixels, not characters, and the only way to know what Google is doing with a title is to go and look at the live result.
What gets checked
- Titles over 561 pixels, and which element gets trimmed
- Titles under 30 characters that waste the space
- Multiple H1s, and H1s over 70 characters
- H2 duplication and heading sprawl
- Question-format headings sitting on commercial pages
Structured data
Two separate jobs: find what is missing, and throw out the validation errors that do not actually matter. Plenty of flagged warnings are noise.
What gets checked
- Flagged URLs re-run through the Rich Results Test
- Missing schema mapped by page type
- Sitewide Organization and WebSite nodes first
- Graph linked properly with @id rather than floating blocks
- Claimed ratings backed by real AggregateRating data
Internal and external links
Two failure modes: equity leaving the site faster than it circulates inside it, and links carrying no signal because they have no anchor text.
What gets checked
- Pages leaking an unusual number of external outlinks
- Internal links with empty or image-only anchors
- Cross-origin links missing rel protection
- Anchor text utilisation per target page
- Tracking parameters buried inside internal hrefs
- Pillar and spoke wiring across the content clusters
Images
Low severity one by one, meaningful in aggregate. The alt text finding is really an image search opportunity finding wearing an accessibility hat.
What gets checked
- Missing alt text, and alt text over 100 characters
- Images over 100 KB, with the worst offenders listed
- Missing width and height attributes causing layout shift
- Lazy loading applied to the LCP candidate by mistake
- Remote image hosts that should be on an allowlist
Security headers and content type
Rarely a ranking factor, always a trust signal, and the fastest section of the audit to close out. Cheap to fix, awkward to explain when a client finds it first.
What gets checked
- Strict-Transport-Security, Content-Security-Policy
- X-Content-Type-Options, X-Frame-Options
- Referrer-Policy and Permissions-Policy
- Server banner disclosure
- Incorrect content types served on real pages
Performance and rendering
Measured and estimated get separated, and the report says which is which. Lab numbers dressed up as field data help nobody.
What gets checked
- Real field data pulled where the site has enough traffic for it
- Cache-Control on the homepage and key templates
- Total page weight and render-blocking resources
- Font hosting, family and weight sprawl
- Rendered DOM compared against the raw HTML
Search Console reports
Indexing states carry different meanings and get treated as one bucket far too often. "Crawled, currently not indexed" and "Discovered, currently not indexed" are not the same problem.
What gets checked
- Not found (404) and blocked due to other 4xx
- Crawled, currently not indexed
- Discovered, currently not indexed
- Indexable URLs that are simply not indexed
- URLs with no search analytics data at all
- Manual actions, and the page indexing trend over time
Discover and video
Two reports that get misread constantly. One is a content strategy signal, the other is usually not a defect at all, and calling it one wastes everybody's budget.
What gets checked
- Discover-serving URLs bucketed by content type
- The pattern translated into a content recommendation
- Video indexed against not indexed
- Video ownership established before anything is called an issue
- The real consequence stated plainly, including when it is none
CTR and opportunity analysis
The upside gets modelled as a number, because "there is room for improvement" closes no deals and starts no projects. Modelled figures are labelled as estimates, never presented as forecasts.
What gets checked
- CTR charted by URL group, never sitewide only
- 5% used as the baseline expectation
- High impressions with low CTR isolated
- Potential clicks modelled at 1% through 5%
- Weak categories explained before they get condemned
- Every weak URL age-gated, plus a year-over-year click loss list
Content quality and E-E-A-T
Experience, expertise, authoritativeness and trust assessed separately, then every weak page resolved into one of three decisions. An audit that lists weak content without deciding its fate is unfinished.
What gets checked
- First-hand experience signals in the content itself
- Author identity, credentials and linked schema
- Authority signals beyond the site's own claims
- Trust markers: contact, policies, sourcing, dates
- Thin and duplicated content resolved to improve, merge or remove
AI search readiness
The newest section, and the one with the widest gap between what a site claims and what it actually implements. Everything here is verifiable rather than theoretical.
What gets checked
- AI crawler directives in robots.txt, parsed properly by group
- JSON-LD presence and whether content sits in the initial HTML
- Author bylines connected to schema
- Direct-answer formatting and comparison tables
- Dated freshness signals and the brand mention footprint
- llms.txt, with its actual status stated rather than oversold
What you can expect to be told honestly
Some findings are cosmetic. Some categories are weak for a reason that has nothing to do with SEO. Where that is the case, the report says so instead of padding the list.
How that shows up
- Measured data and modelled estimates kept visibly separate
- Clean checks recorded rather than quietly dropped
- Metrics that get misread carry their caveat attached
- No ranking guarantees, because nobody can honestly give one
Backlink and link audit
Run as a sequence, and it gets said out loud that DR and UR do not indicate quality. Every metric here needs its caveat attached or it will be misread.
What gets checked
- Top level domain metrics, in context
- Referring domains against raw backlink count
- Nofollow segmentation and tier 2 review
- Country relevance against the actual target market
- Anchor text profile, then the disavow decision
- Referring domain distribution across top pages
- A monthly acquisition target to close the section
Competitor comparison
Compared against businesses that actually compete for the same customer. Directories, marketplaces, news sites and Wikipedia get excluded rather than padding out a chart.
What gets checked
- Real competitor set agreed before the analysis runs
- Content and keyword gaps against that set
- Referring domain gap, sorted by realistic reachability
- Template and SERP-feature differences on shared terms
Actions and deliverable
The difference between an audit that gets implemented and one that gets filed: every finding ships with the instruction, the sheet of affected URLs, and one worked example.
What you get per finding
- The instruction, written as a task, not as an observation
- The sheet of affected URLs attached
- One worked example showing a finished fix
- Findings ordered by effort against return
- Every weak page carrying a decision
- Clean checks recorded, in client-facing language
Built to hand to a developer
Your developer should be able to work from the sheet without a translation layer. That means URLs, the current value, the target value and the reason, in one row.
Formats you receive
- Report document in plain language for stakeholders
- Spreadsheets per finding with the affected URLs
- Copy-ready values where the fix is a text change
- Schema blocks ready to paste where markup is missing
What a finding looks like.
Same underlying issue, written two ways. Only one of them results in anything changing on the site.
"Some title tags are too long and may be truncated in search results. Best practice is to keep titles under 60 characters. Consider reviewing and optimising your title tags."
Nobody can action this. Which pages? Truncated at what width? What should they say instead? Who writes them, and how do they get in?
"38 titles exceed 561px and are trimmed mid-brand. Rewrite to the values in Sheet 3, column D. The suffix is added by the theme, so shorten the field, not the template."
- Sheet 3 lists all 38 URLs, current title, pixel width, rewritten title
- One worked example, before and after, in the live SERP
- Effort tagged as low, return tagged as medium
- Owner named: content, not development
Priced on site size, because that is what drives the work.
Page count decides how long segmentation, crawling and content decisions take. Everything else is the same method at every tier. Pick the band your indexable page count falls into.
Starter
- All 16 areas, full method
- Technical, on-page, schema and Core Web Vitals
- Search Console indexing and CTR analysis
- 1 competitor compared
- Prioritised fix list with URL sheets
- Written walkthrough by email
Growth
- Everything in Starter
- URL segmentation with category-level CTR modelling
- Backlink and anchor profile audit
- Internal link graph and pillar wiring review
- 3 competitors compared
- 45 minute walkthrough call
Scale
- Everything in Growth
- Custom XPath extraction for flat URL structures
- Discover and video reports analysed
- Improve, merge or remove decision on every weak page
- 5 competitors compared
- 60 minute walkthrough, plus 30 days of follow-up questions
Enterprise
- Everything in Scale
- Faceted navigation and index bloat review
- Hreflang and multi-region checks
- Log file review where logs are available
- Migration risk review
- Findings in dev-ready ticket format, stakeholder presentation
Page count means indexable pages, which is usually fewer than your CMS reports. If you are not sure, send the domain and the band gets confirmed before anything is invoiced. Prices are in AUD and exclude GST. The audit is diagnosis and instructions; implementation is quoted separately, and you are free to hand the sheets to your own developer.
Four steps, no mystery.
Scope and access
Confirm the page band and the competitor set. You grant read-only access to Google Analytics, Search Console and the site. Nothing gets changed on your site at any point.
Crawl and extract
Full crawl with custom extraction where URLs are flat, headers inspected, rendering compared against raw HTML, then everything joined to Search Console data by URL group.
Analyse and model
Indexing states, CTR against the 5% baseline, content decisions, link profile, AI readiness. Every weak page gets a decision and every estimate gets labelled as one.
Deliver and walk through
Report, sheets and copy-ready values, ordered by effort against return. Then a call where your developer can ask questions directly rather than reading between lines.
Three read-only access grants.
That is the whole list. No passwords for anything else, no changes to your site, and access can be revoked the moment the audit is delivered.
Google Analytics
Read-only. Used for organic landing page behaviour and to sanity check what the Search Console data implies. GA4 is fine, older properties are fine too.
Google Search Console
Read-only, and the single most valuable item on this list. Nine indexing reports and the full query data live here and no crawler can see any of it.
The website
Usually nothing more than the live URL. Staging or password-protected sites need a login, and heavy firewalls may need the crawler allowlisted for a few days.
Before you book.
How much does an SEO audit cost?
It starts at $1,500 AUD for sites up to 25 pages and scales with page count: $2,500 up to 100 pages, $4,500 up to 500, and from $7,500 above that. Page count drives the price because it drives the work, particularly segmentation and the per-page content decisions.
Free audits exist and they are template exports. They cost nothing because they took nobody any time.
How long does it take?
Between 3 business days and 3 weeks, depending on size. A 25 page site is usually done inside a week. A 500 page site takes 2 to 3 weeks, because the analysis work scales with the number of URL groups, not just the crawl time. Enterprise scopes are agreed up front.
What access do you need?
Read-only access to Google Analytics, Google Search Console and the website. Nothing on your site gets changed during an audit, and you can revoke access as soon as the report lands.
Will this improve my rankings?
An audit is a diagnosis, not a treatment. It tells you what is wrong, what it will take to fix, and which fixes are worth doing first. Rankings move when the fixes get implemented, and no honest consultant will promise you a position for a fee.
What you do get is a modelled click estimate at 1% through 5% CTR, clearly labelled as a model, so there is a number to weigh the work against.
Do you fix the issues as well?
Implementation is quoted separately, and plenty of clients hand the sheets straight to their own developer instead. That is deliberate. The deliverable is built to be handed over: URLs, current value, target value and the reason, all in one row.
What makes this different from a tool export?
A tool can tell you 38 titles are too long. It cannot tell you the theme appends the suffix, that the trimmed element is your brand name, which 38 URLs matter commercially, or what the replacements should say. Every finding here arrives as an instruction with the sheet attached and one worked example.
Do you audit for AI search as well?
Yes, as one of the 16 areas. AI crawler directives parsed properly by group, JSON-LD, whether content exists in the initial HTML, author bylines linked to schema, direct-answer formatting, comparison tables, freshness signals and brand mention footprint. See AI SEO services for the ongoing version of that work.
What if my site is on Shopify, Wix or Squarespace?
All fine. Platform changes which fixes are available to you, not which problems exist, and the report says plainly when something is a platform limitation rather than a mistake anybody made.
After the audit, the work.
An audit points at the problems. These are the engagements that close them out.
Find out what is wrong, in writing.
Send the domain and the rough page count. You get the band, the price and the turnaround back before anything is committed.
Read-only access. No changes made to your site. Fixed price agreed before work starts.
