Be the air purifier
AI recommends.

1. Executive summary

• Your growth came from conversion, not traffic. Organic revenue is up 143% year on year, on only 11.9% more sessions. The store got dramatically better at converting the visitors it already had. That lever is now largely pulled, and traffic is the constraint. GA4
• You rank, then lose the click. Of the queries Search Console names for you, 215,661 impressions a quarter sit in positions four to ten and return 810 clicks. A 0.38% click rate from page one is not a ranking problem, it is an answer problem: AI Overviews sit on 61% of your category's searches and they are resolving the question before anyone reaches you. Search Console Ahrefs
• AI already cites you, by accident. On "hepa filter", "hepa filtration", "air purifier for mould" and a dozen more, your best position is inside the AI Overview itself. Nobody planned that. It came from two genuinely good pages. Imagine it done deliberately across the catalogue. Ahrefs
• The authority that decides AI answers is not yours yet. When someone asks an assistant for the best air purifier in Australia, it assembles the answer from Choice, RTINGS, HouseFresh and the National Asthma Council. You are approved by Sensitive Choice and installed at Monash Health, the University of Sydney and CSIRO. That proof is not where the machines are reading. Ahrefs

So the whole opportunity is one sentence: you hold 6.9% of the non-branded organic clicks in your category, and every extra 1% is worth about $53,400 a year. Your branded traffic is not part of this and does not change.

Built on your own numbers: a new-customer order value of $1,100, and a first-purchase rate on non-branded traffic of 2.32%, which makes each non-branded click worth $25.55. Every column below is non-branded, and your $651,000 of branded organic revenue sits on top of all of it, unchanged. Modelled, plus or minus 40%

The last column counts what those customers are worth over the life of the machine, not just the first order. See What a customer is actually worth below. AI referrals sit alongside all of this, not inside it: ChatGPT, Gemini and Claude already send you $66,115 a year at a $1,482 average order, and that grows with the same work. GA4

A non-branded searcher is someone who does not know you yet, so when they buy, they buy a machine. Across the last twelve months you sold 2,721 INOVA units for $3,116,762, an average machine price of $1,145, which lines up with the $1,100 you quoted us. Then the filters follow.

Derived from your GA4 item data: $989,930 of filter, pre-filter and carbon revenue against $3,116,762 of machine revenue over the same twelve months. Because your installed base is still growing, there are fewer old machines buying filters than a settled business would have, so this figure is conservative and should rise as the base matures. A cohort export from Shopify would firm it up. GA4 item data Derived

Annualised from the 90 days to 6 August 2026. GA4

2. How AI-ready you are today

We score every brand out of 100 against our AEO Maturity Model: four pillars, Technical, Content, Authority and Measurement, each out of 25. AI-Ready is 80 out of 100, which means Professional on all four. That is the standard we build clients to, and getting you there is what the six months is for. You sit at 55 today.

3. The market you are missing

We triangulated every volume in this document across three sources, Google Keyword Planner, Ahrefs and SEMrush, and used the median. Single-tool numbers are unreliable in this category: Ahrefs alone would have understated the basket by 10%, and Keyword Planner alone would have overstated it by 29%.

Named non-branded queries only, Search Console, 7 May to 4 August 2026. Benchmark click rates are the industry ranges for a standard results page. Search Console

Three of your ten biggest non-branded terms return zero clicks despite page-one positions. Search Console

Ahrefs tracks where a site's best position on a keyword actually sits, and for a growing list of your terms that position is inside the AI Overview, not in the blue links below it. You are already being quoted as the answer.

4. The opportunity, in dollars

The table in the executive summary is built from four numbers, all of them yours, and all of them non-branded. Here is the chain in full.

Roughly one search in five survives ads, shopping carousels and AI answers as a winnable organic click, and you currently take about one in fourteen of those. With AI Overviews on 61% of your category, that one-in-five is the number most at risk of tightening, which is the argument for moving now. Modelled

Each non-branded click is worth $25.55 to you, because roughly one in 43 of these visitors buys, and when they do they buy a machine at about $1,100. That is the figure every case in the table multiplies. Counted over the life of the machine rather than the first order, the same click is worth $33.66.

That is the whole model: win more of the 209,040 clicks.

The mid case arriving, on non-branded revenue: certified by month three, the system handed to you at month six, all of it live by month twelve. Hover any point for the maths. Modelled, directional

Two levers, both measured. Click-through on the impressions you already get, as you move from position 4 to 20 into the top three and start being the cited answer rather than the link below it. And more impressions, as the territories you have no page worth ranking for start to rank. The conservative case assumes only the first lever really lands; the high case assumes both do.

Every figure here is non-branded. Your $651,000 of branded organic revenue is unchanged and sits on top. Replacement filters are recurring revenue and would add further upside again, but we have not modelled them, because we do not yet have your repeat-purchase data. Modelled

5. The six-month transformation

Three phases. We move through them as your site is ready, not on a stopwatch. The goal across the six months is AI-Ready at 80 out of 100, and we get there as early as the content allows.

Fix what is intercepting the click. FAQ and Q&A markup across every collection and product page so your answers are liftable. Rewrite the titles and descriptions on the pages that rank but do not win, starting with the ten head terms above. Repair /pages/compare. Resolve the second domain, inovapurifiers.com, which currently competes with you on ten of your own keywords. Set up the entity layer, the markup that tells an assistant what INOVA is, where it manufactures, and which institutions use it. Measure Core Web Vitals properly and fix what needs fixing.

Twelve content clusters, mapped to 94 keywords and 124 AI prompts. Two are outright gaps today: objections and evidence, where the sceptical buyer decides, and provenance and trust, where your Australian manufacturing should be doing far more work than it does. Full coverage of the category is 114 pieces. The core six clusters, which carry most of the demand, are 72. We size the actual build to your retainer and start with the highest-value territories, HEPA and filtration first because you already win there.

Get INOVA into the places AI reads when it answers your category: Choice, HouseFresh, RTINGS, asthma and allergy resources, health and building publications. Turn Monash Health, the University of Sydney, CSIRO and UNSW into published, citable case studies rather than logos on a page. Build the review and comparison surface so that when an assistant assembles "best air purifier in Australia", your evidence is in the pile it reads from.

What is included, how the engagement runs and the guarantee are all on the offer tab.

Appendix A: where you sit today

GA4, 9 May to 6 August 2026, annualised. The AI Assistant row is GA4's own channel grouping. Counting every ChatGPT, Gemini and Claude referral across all mediums, the true AI figure is $66,115 a year. Organic Social carries 32,206 sessions at a 0.13% conversion rate, which is worth a separate conversation. GA4

Your site average order value is $658. AI-referred buyers spend 2.25 times that. They arrive having already been told what to buy, which is exactly why being the brand named in that answer is worth more than another click. GA4

Appendix B: the demand detail

Low search volume on the commercial cluster is not the same as low value. A single aged-care or hospital deployment is worth many times a home sale, and those buyers ask assistants rather than search engines, which is why the prompt library covers that territory in depth. GKP + Ahrefs + SEMrush

The full 94-keyword basket, the 124-prompt AI library, the cluster map and the scorecard workings are in the companion workbook on the data tab. Tier 1 means all three tools agreed within a narrow band. Tier 2 means two sources, or a wide spread worth treating with more caution. GKP + Ahrefs + SEMrush

Appendix C: where to play

Twelve content territories, built from the keyword basket and the prompt library together, scored against what already exists on your site.

Appendix D: who you are up against

Two different fights. Retailers and global brands compete for the transaction. Review sites and health bodies compete for the answer, and they are the ones AI reads.

Ahrefs, August 2026. Against the Australian specialists you are competitive. Against the sites that shape the AI answer you are 35 to 50 Domain Rating points behind. Ahrefs

Appendix E: where the numbers come from

• Core Web Vitals. The PageSpeed API hit its daily quota while we were building this. It is a week-one task and could move the Technical score either way.
• Repeat and lifetime value. Replacement filters are recurring revenue and we have deliberately left them out of every projection. Once we can see repeat purchase behaviour, the return on this work goes up, not down.
• Offline and institutional revenue. Everything here is online store revenue. Showroom, phone, dealer and tender business sits outside GA4, and some of the commercial content territory will influence it in ways this model does not count.

Every figure marked your own data is measured. Every figure marked modelled is a projection built on those measurements, with the assumptions stated beside it. We do not mix the two.