Where it started
A shop with forty-one years of history and a website that said almost none of it. One page, a phone number, and nothing a customer could ask a question of. When someone in Dundas asked an assistant where to find a fitting for a particular brand, there was no page anywhere that could answer, so the answer was another shop.
- One indexable page for a business carrying dozens of brands and four departments
- No page for any brand, so a customer searching a brand name and the town found the manufacturer, not the shop
- No guides: none of the questions a woman between forty and eighty actually asks before a visit were answered anywhere
- No structured facts a machine could match to the business listing: hours, address and departments lived only in the shop
What we built
A page for each department, each carried brand, the guides customers ask for, the story and the visit, each with its own address, title, description and structured data, and each written from the shop's own facts rather than a template's.
Customers are forty to eighty. The type is larger, the contrast is higher, the buttons are bigger, and nothing moves that does not need to. The site earns trust the way the storefront does: quietly.
A robots file with a deliberate decision for each AI crawler, a sitemap with honest dates, llms.txt, content signals, a proper 404, and a Markdown twin of every page for the tools that ask. Verified before launch and re-verified for this page.
Hours, address, phone and departments live in a single configuration and reach every page and every structured-data block from there, so the site can never disagree with itself, or with the shop's Google listing, page by page.
Measured results, not vibes
Every number below was measured against the live domain on with the same tooling a client receives, checked against Google's and Bing's published rules. We do not promise rankings or citations; Bing's own guidelines say nobody can. We promise the site gives the machines no reason to look elsewhere.
| Crawler | Feeds | HTTP | Received |
|---|---|---|---|
| Googlebot | Google Search, AI Overviews | 200 | 24,552 b |
| bingbot | Bing, Copilot, ChatGPT search | 200 | 24,552 b |
| OAI-SearchBot | ChatGPT search answers | 200 | 24,552 b |
| ChatGPT-User | A live ChatGPT question | 200 | 24,552 b |
| GPTBot | OpenAI model training | 200 | 24,552 b |
| Claude-SearchBot | Claude search answers | 200 | 24,552 b |
| ClaudeBot | Anthropic model training | 200 | 24,552 b |
| PerplexityBot | Perplexity answers | 200 | 24,552 b |
| meta-externalagent | Meta AI | 200 | 24,552 b |
| Amazonbot | Amazon AI, IndexNow | 200 | 24,552 b |
| Applebot | Siri and Spotlight | 200 | 24,552 b |
| CCBot | The open web corpus | 200 | 24,552 b |
See it live
The proof is public. Visit amarettos.ca, or ask your favourite assistant about women's clothing in Dundas and see who it can actually read.
What this was
This was the full version of the AI Visibility Pass applied to a one-page site: the rebuild, the files, the verification and the written report, delivered as one fixed-quote project. On a managed website the same playbook is simply included from day one.