Roof Racks On The Run / Search & AI plan
Three months of location pages have worked. Rankings are climbing and the Google listing is busier. But Google and the AI assistants have learned only half the business: they know you fit roof racks, and they do not know you sell them. This is what the numbers show, what is causing it, and what we do next.
Search Console data starts on 5 July, which is when the site was connected, so July is the baseline. Everything since is movement off that line.
| Month | Times shown | Visits | Average position |
|---|---|---|---|
| July (from the 5th) | 11,879 | 176 | 18.3 |
| August | 14,061 | 146 | 18.0 |
| September | 14,825 | 183 | 14.4 |
| October (1 to 4) | 1,787 | 25 | 11.8 |
Average position is where you sit in Google's results across every search you appear in. Lower is better.
The headline is the last column. You have moved from an average of 18.3 to 11.8, which is the difference between page two and the bottom of page one across the board. Times shown are up about 25 per cent from July to September.
The interesting part is that September was the strongest month, and nothing new went live in September. The sixteen location pages built over winter kept maturing on their own. That is what compounding looks like, and it is the argument for keeping the page build going.
Listing activity grew about 80 per cent from May to August, then eased back in September. Some of that is seasonal and some of it is the quiet month.
You now rank near the top for the suburb searches: mobile roof rack installation robina at 2.2, roof boxes palm beach at 2.0, ladder racks palm beach at 2.9. They are also pulling product searches tied to a suburb, which is the six product tiles on each page earning their keep: roof boxes in Burleigh Heads, kayak carriers in Nerang, bike racks in Currumbin, ladder racks in Varsity Lakes.
Southport, Broadbeach and Mudgeeraba are already ranking between 2.1 and 3.8 without a page existing. Those three are the easiest wins left on the board.
I put three real questions to Google's AI mode, the one that now answers above the normal results. The split is clean, and it is the whole problem in one screen.
Question 1
Who can install roof racks on my car on the Gold Coast?
You are given your own heading, Mobile Installation (We Come to You), above every shopfront on the Coast. The answer even says you "supply and fit major brands including Rhino-Rack, Thule, Yakima, Rola and Prorack".
Question 2
Where can I buy roof racks on the Gold Coast?
Six businesses named. Roof Racks Galore, GCB Bars & Racks, TJM, Bushwakka, Autobarn, Supercheap. You are not one of them, despite having more reviews than five of the six and a better rating than all six.
Question 3
Who sells Rhino-Rack and Thule roof racks near Burleigh Heads?
Three businesses named, all with shopfronts. You are absent, even though there is a live Burleigh Heads page, a Thule page and a Rhino-Rack page on the site.
There is one more detail worth knowing. When the AI described the competitors it quoted their websites. When it described you, every source it listed was your Google listing. Your listing is carrying the business on its own, and the website is not earning a mention on a buying question.
It is a real search and you should keep winning it. The trouble is what it implies. A machine reading "installation" everywhere concludes you fit the racks a customer already owns, and so it files you under services rather than under shops. When someone asks where to buy, you are not in the set it draws from.
This is what makes the fix worth doing. You are not missing from the buying searches. You are already in them, and already ranking well, on pages that were never built to sell anything.
| Type of search | Searches | Times shown |
|---|---|---|
| Your business name | 1 | 197 |
| Fitting and installation | 122 | 2,782 |
| Brand and product | 167 | 4,068 |
| Buying, price and stockist | 23 | 224 |
| Everything else | 687 | 11,505 |
Your top 1,000 searches for the period, sorted by what the person was trying to do.
Brand and product searches are more than a fifth of everything you are shown for. People are typing Prorack, X-Country, Thule and Tradesman by name, and Google is picking your site to answer them.
Those are top of page one positions on searches with a buyer behind them, and you have earned them already. What is missing is everything the page would need to say, and show, to turn that visibility into an enquiry. That part is all fixable.
Google now reports separately on where a site is used in its AI answers. In three months the site was used 5,190 times across 111 pages. The single most used page is not the homepage and not a suburb page. It is a product page, the ProRack Aero Flush Bar, used 822 times. The roof racks product category is third at 479, and a second product page is eighth.
The AI tools already prefer your product pages. Those pages are the thinnest part of the site and roughly half of them are not even in Google's index. Fix that and you are feeding the engine the thing it is already reaching for.
149 pages are indexed and 203 are not. Of those, 137 have been found by Google and left out, and another 16 were crawled and left out. The product list contains 170 items and only about 95 are in the index, so roughly 75 products Google has seen and decided not to keep.
The shop page itself has no main heading on it at all, which is unusual enough that it is probably part of why the catalogue is treated as low value.
Behind each product page is a block of data Google reads to understand what is for sale. On every one of the 170 products the price in that block is 0.00 and the stock status is "in stock". A shop selling 170 items for nothing is not a shop Google will show in a shopping result, and it is not a shop an AI will recommend.
The proof is in the results themselves. Across the whole site the only enhanced result Google produces is a review stars snippet. There is not a single product result. Only 22 of the 170 products register as valid product items.
Google also reports that none of the products carry a brand. The business stocks Rhino-Rack, Thule, Yakima, Rola, Prorack, X-Country and Tradesman and tells Google the brand of none of them.
And the homepage carries a section headed "Why No Prices or Shipping?". A person reads that as honesty. A machine reads it as confirmation that this is not somewhere you buy things.
There is a block of identity data that sits across every page and tells Google and the AI tools who the business is. Right now it contains a name, a logo and two social links. No address, no phone number, no opening hours, no service area, no business category, nothing about products. The sixteen suburb pages we built are the only pages on the site that properly identify the business at all, and they describe a fitting service.
Nothing anywhere declares this as a store, as an auto parts retailer, or as a stockist of those seven brands. The site is also set to American English.
The Thule, Rhino-Rack, Yakima, Rola, Prorack, Tradesman and X-Country pages, the six vehicle pages and the six trade pages all run off the same template. Measured against each other they are 78 to 80 per cent identical. Each is about 1,000 words of which maybe 200 are unique to that page, and nine of them have no description written for the search result.
These are exactly the pages that should be saying "we stock this, here is what we carry, here is what it suits". They are the weakest pages on the site.
Every location page leads with "Roof Rack Installation" in its address, its page title and its heading. The Gold Coast page is titled "Roof Rack Installation Gold Coast". There are two competing Gold Coast installation pages splitting the signal between them. The main Gold Coast page and all sixteen suburb pages have no description written, so Google writes its own, and it writes an installation one.
Separately, the homepage still says "over 30 years" where the agreed line is 35 and over, and still says "Worksite" in two places where we settled on jobsite.
The principle is simple. Do not drop installation, because it is the thing winning right now. Add the selling half beside it everywhere, especially in the places machines read rather than people.
The zero dollar price has to come off all 170 products either way. There are two honest ways to do that and they are both legitimate. It is your call which one.
We take the price out entirely and mark every product as quote on request, which is exactly how you work today. Nothing changes in how you quote, and it stops Google seeing a shop that sells 170 items for nothing. It will not get you into Google's shopping results, because those need a figure to work with.
Each product family carries a band, the way you would answer it on the phone. "Roof racks from $X fitted." Not a price list, and not a commitment on any particular vehicle. This is the stronger option, because it answers the first question every buyer and every AI assistant asks, and it is what makes the product data complete enough to show up in shopping results.
Everything else in this plan works the same under either choice. This one just decides how far the catalogue can go.
Three things repeat in every single month of the retainer. They are what keeps the ground moving while the one off build work gets done.
The three questions at the top get asked again every month. That is the real scoreboard, and it is the one you can check yourself.