Rebuilt to #1 for 'car insurance' in Australia's most contested vertical
A dual recovery across traditional search and AI search. We took one of Australia's largest insurance comparison platforms through the ranking decline, the content rebuild, and the AI-search transition. Sustained, end-to-end work across two fronts rather than a one-off win.
What we were working with
"Car insurance" is one of the highest-value organic search terms in Australia, estimated at $1.9m in monthly traffic value, and one of the most contested. Our client, one of Australia's largest insurance comparison platforms, had held strong top-2 rankings for years.
Beginning with the July 2021 Google broad core update, the term became structurally harder to own. Google began surfacing direct insurers (AAMI, Budget Direct, NRMA) ahead of comparison platforms. The platform dropped from position 2 to positions 6-8 almost overnight.
This is the hardest type of core update impact to respond to: you can't become a different type of site. What you can do is control everything else and wait for Google's intent model to re-evaluate.
Rankings partially recovered through the September 2021 update, back into the top 2. But the volatility continued. August 2023 brought another drop. November 2023 pushed rankings to positions 8-10. By March 2024, the platform dropped off page 1 entirely for the first time, at its worst page 3.
It was clear this wasn't just intent shift volatility. A cumulative site-quality problem had developed alongside the intent shifts, and Google's bar for what constituted a worthy result had risen. Quote starts had dropped to 70% of plan targets. The business case for a full remediation programme was clear.
At the same time, a second front was opening. Google's AI Overviews were resolving informational queries inside the SERP without a click. Pages like "cost of septoplasty", where users research a procedure cost before considering health insurance, were losing clicks while rankings and impressions held.
If AI Overviews pulled from competitors instead of the platform, the brand was cut out of the consideration moment entirely. Not because it wasn't ranking, but because its content wasn't structured for AI extraction.
What the analysis revealed
Rather than reacting immediately, we ran a structured delta analysis, comparing the platform's pages against top-ranking competitors to determine whether the problem was a relevancy adjustment, intent shift, or site-quality issue. This distinction matters: an intent shift requires patience and positioning; a quality problem requires deep, sustained work across the site.
The analysis pointed to a combination of all three, with site quality as the controllable lever. A core principle we applied: avoid the knee-jerk reaction. A drop during a major algorithm update is rarely fixed by a single change, and premature changes without diagnosis risk making things worse. Sites impacted by a core update rarely recover until the next core update, meaning the programme needed to be built for the long term.
Content gaps
Missing subtopics across the car insurance vertical that competitors were covering in depth. The platform's head page was ranking on authority, not topical completeness.
Thin content and generic authorship
Pages lacked the E-E-A-T signals Google's quality raters were increasingly weighting. No named experts, no demonstrated first-hand experience or credentials behind the advice.
Internal linking inefficiency
PageRank was not flowing efficiently to the highest-priority commercial pages. Authority earned through informational content and digital PR wasn't being directed where it was needed most.
Cumulative site-wide quality signals
UX, layout, ad density, and content presentation were all contributing to a lower overall site-quality assessment. Not any single smoking gun, but a battery of issues compounding each other.
How we approached it
Kitchen-sink remediation
The core principle was to treat this as a site-wide quality problem requiring sustained, comprehensive improvement, not a checklist of quick fixes. Google rewards significant improvement in quality over the long term. We addressed everything simultaneously: content, authority, technical health, UX and trust signals. No cherry-picking, because isolated fixes rarely move the needle during core updates.
Content rebuild and topical completeness
Identified and closed topical gaps across the car insurance vertical, every subtopic a user or Google might expect a definitive comparison platform to cover. Rebuilt the head page with significantly deeper, data-led content. Removed thin, generic sections and strengthened heading specificity and content structure to improve relevance signals. The revamped page went live 16 October 2024 and was back on page 1 within 6 days.
E-E-A-T expert programme
Built a named-expert programme bringing in David Koch as the platform's financial expert, with his name and credentials attached to relevant content. This directly addressed the authorship gap Google's quality raters flag on YMYL content. Named credentials are now table stakes in financial content, not a differentiator, but we were early, and it compounded with the authority and content work.
Digital PR and authority building
Built a dedicated digital PR function from scratch. Since inception, the programme has earned 3,603 links. In the most recent financial year alone, it secured 800+ links, growing domain rating from 58 to 73. Critically, we redirected PageRank from high-authority digital PR feature pages to target commercial pages, funnelling earned authority directly to the pages that needed the ranking boost most.
Meta title innovation
Introduced a named savings claim into the car insurance meta title, a single change that pushed the platform from the bottom of page 1 to the top. This triggered a competitive arms race, with major competitors now adopting the same tactic. The platform moved first and currently owns the most prominent claim in the SERP. Sometimes the highest-leverage change is a 60-character title tag.
AEO content programme
Developed an AI Engine Optimisation style guide from first principles, applied it to a proof-of-concept page, built the measurement framework on Profound, and began a scaled rollout across the content library. On the proof-of-concept page, AI visibility rose from 25% to 100% and citations in AI responses doubled from 4 to 8. The programme now has 4 pages live with consistent results, 4 more launching this week, scaling to 10 per week.
The AEO style guide
Rather than making ad hoc changes to individual pages, we developed a repeatable content framework that restructures pages for AI extractability without sacrificing human readability or brand voice. We applied the guide to a controlled proof of concept, tracking results before and after across AI visibility, citation counts and click data, then systematised the rollout.
Question-first headings
Every H2 and H3 is written as an explicit question or precisely defined topic. Before: 'Car insurance costs.' After: 'Does comprehensive car insurance cost more than third party?' The effect is a page that becomes modular, scannable and AI-readable. Each section a discrete, self-contained unit of information that AI systems can extract independently.
Answer-first structure
Every heading is followed immediately by a direct answer. No build-up, no preamble. Direct answer first sentence, then supporting data, then optional clarification. Users get the answer at once. AI systems extract it cleanly. The scroll-to-find-the-answer behaviour that increases bounce rate and reduces dwell time is eliminated.
Fact density upfront
Concrete data points, definitions and named entities appear in the answer block, not buried in supporting copy. This directly improves AI parsing, as models weight specific, verifiable information more heavily than general commentary.
Two-layer tone
Answer blocks use clear, factual, minimal language. No analogies, no personality, no fluff. Supporting content after the answer block is where brand voice, storytelling and persuasion live. This separation serves both AI extractability and human engagement.
Sections as standalone blocks
Each section is written to be comprehensible in isolation, as if a user or AI system might encounter it without reading the rest of the page. Internal links are placed after the answer block, so the answer reads as self-contained knowledge rather than navigation.
Table discipline
Tables are kept small and tightly scoped. Key insights from tables are always repeated in the surrounding text. Data cannot live only in a table. This improves both AI parsing and accessibility for skim readers.
The numbers that matter
Rebuilt from page 3 to an effective #1 across the five most important car insurance keywords, representing 212,000 average monthly searches. Positions solidified from 5-10 to 1-3 across three consecutive core updates.
Organic sessions on the car insurance landing page grew from 93,000 to 324,000. Estimated monthly traffic value: $1.9m.
Car insurance quote starts recovered from 70% to a consistent 100-120% of plan. The direct commercial outcome of ranking recovery.
3,603 links earned since the digital PR programme launched. 800+ links in the most recent financial year alone.
AI visibility across four tracked prompts on Profound rose from 25% to 100% on the proof-of-concept page. Programme now scaling to 10 pages per week.
Citations in AI responses doubled. The page's question-format headings are now cited in AI Overview answers. Post-launch clicks increased 160% week-on-week.
Organic ranking position, "car insurance"
Ranking position over time. Lower is better. The page launched 16 October 2024 and recovered through three consecutive core updates.
What we'd do differently
The E-E-A-T programme was the right call, but we'd move faster on it. Named expert credentials on YMYL content are table stakes now. We'd go further with video content from experts, first-person experience content, and proprietary research no competitor can replicate.
We'd build GEO measurement into the programme from day one. The car insurance queries the platform targets are increasingly resolved in AI Overviews and ChatGPT. Traditional rankings are strong, but AI citation requires its own content architecture and measurement.
On the AEO programme, we'd add cross-platform citation tracking earlier. ChatGPT and Perplexity are increasingly where high-intent health and financial queries land. We'd also push harder on digital PR in parallel. The sources AI models cite are the same publications that drive traditional SEO authority, and running both programmes together would compound the citation effect faster.
See more of our work
Want results like these?
Talk to our team about what we can do for your business.
Get in touch →A digital marketing agency in Auckland that makes you money - and proves it.