Fill the room,
lift the number.

1. Executive summary

• Sunfire has the room, it just is not being found. Your dinner peak is slammed from 7:00 to 8:30 and quiet either side, breakfast is lumpy and weather-dependent, and lunch and the afternoon have a lot of open capacity. Every one of those gaps is a booking that a "where to eat in Queenstown" search could fill, and today almost none of that search reaches you. You
• Your dayparts are the least contested terms in town. Breakfast, best breakfast, lunch and brunch in Queenstown sit at keyword difficulty 0 to 2, real demand at unclaimed positions rather than hard ones. Almost no local venue does proper search, content or AI work, so the layer that decides bookings is sitting open. Triangulated
• The AI answers are being set right now. Google already fires AI answers on "things to do in Queenstown", and travellers increasingly ask ChatGPT and Perplexity where to eat. There is no clear AI-preferred all-day or breakfast spot in Queenstown yet. Whoever builds the reviews, readable content and citations over the next six months becomes the default answer, and that is hard to unseat once it sets. Ahrefs

Triangulated NZ + AU search demand, July 2026. The full keyword, prompt and cluster research behind every number here lives in The data tab. GKP + Ahrefs + SEMrush

2. How AI-ready you are today: the AEO scorecard

We score every venue out of 100 against our AEO Maturity Model: four pillars (Technical, Content, Authority, Measurement), each out of 25. Certified is 80 out of 100, Professional on all four, the standard we build clients to. Sunfire sits at 40 today, an Emerging score with one real strength and three clear, buildable gaps.

Flame sits at 50 / 100 on the same model and is in good shape, full at dinner every night, so this document is about Sunfire. Flame gets a light touch as part of the work, described in section 5. Live crawl + Lighthouse + Ahrefs

The detailed findings, how the scoring works, and where every number comes from are in the appendix.

3. The market you are missing: search and AI

In a tourist town, where to eat is a researched decision made on a phone before arrival, in Google Maps, a "best restaurants Queenstown" search, and increasingly an AI answer. The demand is real and it skews straight into Sunfire's dayparts: breakfast, brunch, lunch and the all-day discovery searches. The pattern is the whole story: strong monthly demand, unusually low difficulty, and Sunfire almost nowhere.

These are the real monthly search volumes for the terms that map onto Sunfire's trade, triangulated across three tools for New Zealand and Australia. Steakhouse and dinner-only terms are Flame's territory and are deliberately left out of this table.

Search volume triangulated across Google Keyword Planner, Ahrefs and SEMrush (median of three), New Zealand and Australia. Ahrefs alone read these local terms low, "breakfast queenstown" sat far below Keyword Planner and SEMrush, which is exactly why we triangulate. Rank is Sunfire's Ahrefs position (sunfire.co.nz is not in our Search Console), so "Not ranking" means outside the top 20. The pattern is the whole story: real demand, difficulty often at 0 to 2, and Sunfire absent or bottom of page one on the two terms it ranks for at all, because the non-branded engine has never been switched on. The full triangulation with all three sources is in The data tab. GKP + Ahrefs + SEMrush Ahrefs rank

Across New Zealand and Australia the Queenstown dining basket runs to roughly 27,000 high-intent searches a month once triangulated, before international visitors and Maps-native searches, which means the real number is higher. About 90% of the core dining searches show a Google Maps local pack, where the choice is actually made. On top of that sit the AI questions: we mined a 31-prompt library from real sources, and the daypart, discovery and practical clusters are Sunfire's. The full library, the cluster maths and the market breakdown live in the appendix and The data tab.

Demand is keyword-anchored with a conservative AI fan-out uplift, directional plus or minus 50%, so treat it as a floor. The two biggest lanes, breakfast-and-daypart and discovery, are exactly where Sunfire has room to fill. Modelled, AI-overview verified in Ahrefs

4. The opportunity, in dollars

Restaurant search converts as incremental bookings, valued at your real spend per head. Sunfire's room to fill is specific and you named it: a second dinner seating either side of the 7:00 to 8:30 peak, plus weekday breakfast and lunch and the cocktail and afternoon trade. Here is what each of those is worth per booking, at a party of around three.

Spend per head is yours; value per booking is spend per head times a party of about three. Flame already runs three seatings a night; Sunfire runs about one, so a second seating is where most of the upside sits. Your figures

One curve, the mid case: the new bookings arriving as an annualised run-rate, drawn as an S-curve. Almost all of Sunfire's demand today is unmet, so the baseline is near zero and the curve is the new revenue coming on. The shape is deliberate: slow to month 3 while foundations and content go in, steep months 3 to 9 as authority and momentum turn into rankings, reviews and AI citations, then tapering to month 18 as it saturates. The six-month engagement ends at month 6, where you own the system, and it keeps compounding to the sale. By month 18 the line is at roughly a $200K annualised run-rate, the upward trajectory an incoming buyer wants to see. Hover any point to see the maths. Modelled, directional

Chain: blended new bookings per week times 52 times a blended value of about $170 per booking (weighted across the four dayparts above). The mid case is roughly +23 bookings a week across the dayparts, a handful a day. Return is that annualised run-rate against the $30k six-month build, a 5 to 10x range over the 12-to-18-month sale horizon. All figures are NZD, directional at plus or minus 40%, and they annualise the booking run-rate reached by month 18, not a full-year total. Modelled, plus or minus 40%

5. The six-month push

This is a focused, hard-and-fast six months, not a long-tail SEO project. The offer is the same four-pillar engine we build for every client, sequenced to create momentum you can show, and it is done for you: you approve, we build and run it.

The full breakdown, deliverables and pricing live in The offer tab.

Search volumes triangulated across Google Keyword Planner, SEMrush and Ahrefs (NZ + AU), July 2026; they exclude international visitors and Maps-native searches, so they understate true demand. Footprints from Ahrefs Site Explorer. AEO score from our AEO Maturity Model against a live crawl, Lighthouse and Ahrefs. Booking values use your own spend per head and party size. Revenue figures are directional, modelled from those inputs. Pricing is set out separately.

Appendix A: Sunfire's current position

Sunfire is a two-year-old all-day dining restaurant in Queenstown doing roughly $6.2M a year. The reputation is real, the food is good, and the trade is lumpy: a slammed dinner peak from 7:00 to 8:30 and quiet either side, breakfast that swings with the weather, and lunch and the afternoon with a lot of open capacity. This is growth on a good business, not a rescue. The one thing it is not doing is being found.

Flame's 19 years of reviews and awards show up in search; Sunfire has barely any footprint, which lines up with the lumpy trade. The consistency Flame enjoys is partly a search and reputation effect, and it is buildable for Sunfire. Ahrefs Site Explorer, NZ

Illustrative of the trade you described, not a measured chart: the dinner peak is full, and breakfast, lunch and the afternoon have room to fill. Every empty seat in those windows is a booking a "where to eat in Queenstown" search could bring, and today almost none of that search reaches you. The second dinner seating, either side of the peak, is the single biggest unlock. You

Appendix B: the scorecard in detail

Here is what sits behind the 40, and who fixes each piece. Your one real strength is authority, the reviews and reputation most venues never build; the work is a technical fix, content an AI can read, and measurement.

Appendix C: the demand detail

We mined a library of 31 real questions visitors ask AI engines and search about Queenstown dining across both venues, from Google's People Also Ask, Reddit, review language and forums. The daypart, discovery, views, practical and local clusters are Sunfire's, and they are what we carry into the plan. Here is a sample.

A sample of the Sunfire-relevant library, all of it in The data tab with per-question cluster, intent and journey stage. Volumes are keyword-anchored directional floors with a conservative AI fan-out uplift, plus or minus 50%. Directional

And here is who the answers name today, cluster by cluster, versus where Sunfire shows up.

Appendix D: where to play, your daypart territories

The same demand, grouped into the territories a plan executes against, each scored on how much of it Sunfire can realistically own. Priority weighs demand, whether AI is live, and whether you can win it uniquely as an all-day restaurant.

Priority is a 1 to 5 score, where 5 means lead with it now; it is not a count of pieces. The full cluster maths is in The data tab. Cluster model

Appendix E: the field, and the Australia and visitor angle

The striking part of Queenstown dining search is how open it is. Almost no local venue does proper search, content or AI work, difficulty across the core dining terms sits at 0 to 7, and there is no clear AI-preferred all-day or breakfast spot yet. This is unclaimed ground, not a hard climb.

Four in five diners come from out of town and decide where to eat on a phone before arrival, and about 90% of the core dining searches show a Google Maps local pack. That is why the Google Business Profile and reviews work is the biggest single lever, and why being readable and reviewed decides bookings. Ahrefs You

Appendix F: where the numbers come from

Measurement runs in our own platform, so the dashboard, the data and the method are yours to keep, not rented from a third-party tool. The full research behind every table here is in The data tab. Method