A finance brokerage in Australia asked for a free SEO audit before committing to a monthly package. Its website was custom-built, recently moved to a new domain, and could not say how many enquiries came from Google. My audit scored the site 61 out of 100 on an Ahrefs crawl, found 263 pages that nothing linked to, and ended with a ranked fix list the client’s developer could act on. The client’s name is withheld.
These are audit findings at the start of the work, not results of the fixes. Source: Ahrefs Site Audit, 5 November 2026.
The Challenge
The client wanted more qualified enquiries that become approved loans, not just more traffic. Three things stood in the way. The site had moved to a new domain a few months earlier. Google Analytics could not show how many finished applications came from search, because the key event that should count them recorded nothing for 28 days. And the homepage title described one loan product while the client’s first priority was another.
The site was also custom-built rather than WordPress, so a normal “install a plugin” checklist did not apply.
The Approach
I split the audit into four plain-language areas (on-page, off-page, technical, and GEO, short for Generative Engine Optimization) and wrote it for a business owner, not an SEO. Each area lists what is good, what needs work, and the actions in priority order.
1. Measure before fixing
I read Google Search Console and Google Analytics first, so every recommendation points to a number. That is how the missing enquiry tracking surfaced, and why it became a top action.
2. Check how the site is built before promising fixes
Response headers showed Next.js, with content in a headless CMS behind Cloudflare. That changes the proposal: redirects, templates and tags sit in code, so I agreed to work with the client’s developer and give exact written instructions, instead of editing the live site myself.
3. Test how AI tools see the site
The robots file blocked several AI training bots, a default setting from the site’s protection service. I separated training bots from search bots so the client could decide knowingly. Asked general loan questions, Perplexity named other sites; asked by name, it found the client’s site.
Execution Breakdown
- Sitemap and status check: every sitemap URL returned a normal response, was self-canonical and indexed.
- Crawl with Ahrefs: health score, orphan pages, noindex pages, titles, sharing tags, redirects and response time.
- Redirect check after the domain move: old addresses all went to the homepage instead of the matching page.
- Near-duplicate page check: a large set of location pages differed only by place name, a thin-content risk.
- GEO review: AI-bot access, structured data, authors and dates, plus four AI-search test questions.
- Report and fix packages: a 12-page PDF with evidence, a ranked list of actions, and two fixed-scope fix packages mapped to those actions.
The Findings
- Tracking: the finished-application event recorded nothing in 28 days, and click events inflated the “key events” total.
- Orphan pages: 263 pages had no internal link pointing to them.
- Titles and tags: 144 titles too long, 13 descriptions too short, sharing tags incomplete on all 318 sitemap pages.
- Hidden pages: 20 pages carried a noindex tag, each needing a check.
- AI search: AI Discoverability scored 70 out of 100, with bot access at 74 but machine readability only 35.
- Off-page: almost no independent reviews or mentions.
Want the same review for your website?
Send me your website and your top priorities. I reply with the next step, with no obligation.