- Mention rate across AI engines went from 13% to 73% over five months. Citation rate went from 5% to 33%.
- Average Google position moved 3.2, then 2.2, then 1.2.
- Revenue was $240,000 in 2025, the year before we started. They are projecting $550,000 now.
- They were never invisible on Google. They were invisible inside AI answers, which is a different problem with different fixes.
- Being named and being cited are not the same thing. At the end of the engagement there was still a 40 point gap between the two, and that gap is the next phase of work.
The problem was not Google
Sound Elite Theatres designs and installs high-end home theatre and audio systems in Vancouver. It is a considered purchase, expensive, and researched hard before anyone picks up a phone.
They were not invisible. On Google they were already ranking, averaging around position 3.2 on their target queries, some of them sitting at 2. By the standards of local search, that is a business doing fine.
The problem was somewhere else. When a customer skipped the search bar and asked an AI engine directly, which is increasingly what happens with a purchase this size, Sound Elite was named in roughly 13% of answers. The engines understood the category perfectly well. They were naming other people in it.
That gap does not appear in analytics. Branded traffic looks healthy. The missed demand never reaches the site, so no number goes down. You simply do not get the call, and nothing tells you why.
What we measured first
Before changing anything, we established where they actually stood.
We built the real set of buying intents for the business and tested each across multiple phrasings rather than single keywords, because nobody asks the same question the same way twice. Each was run against four engines: ChatGPT, Claude, Gemini and Perplexity.
For every response we recorded three things:
- whether the brand was mentioned at all
- whether it was cited, meaning the domain actually appeared as a source
- how strongly it was positioned, graded from a passing mention up to a primary recommendation
We ran the same measurement for their named competitors, broken out per engine. Knowing you are absent matters less than knowing who is there instead.
The intents split into two kinds, and they behave very differently. The first is product visibility: someone knows what they want and is looking for who supplies it. The second is content authority: someone has a problem, has not settled on a solution, and the engine is deciding whether to recommend hiring a professional at all. The second is harder, and it is worth more.
What we changed
A deterministic audit across the whole site, page by page. It starts with whether the AI crawlers are even allowed in, and checks seven by name: GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended and CCBot. All seven were verified with access on their site. Then sitemaps, schema markup, meta titles and descriptions, canonicals, HTTPS, broken links, image alt text and page speed.
Findings become action cards carrying a risk rating and exact step-by-step instructions for their platform. Nothing applies itself. They approve it, we apply it, then we re-run the audit to prove it is fixed.
That loop is the part the client noticed.
“[Victory Velocity] are the first people who offered to actually do it, instead of handing me another list.”
An audit is not a deliverable. A list of problems is not progress. The work is the fixing, and then proving the fix landed.
What moved
Work ran from March to July 2026.
Mention rate: 13% to 73%
Close to three in four answers now name them, where it was closer to one in eight before. That is a 60 point move, and it is the number that most directly represents being in the running at all.
Citation rate: 5% to 33%
Citations grew faster in proportional terms than mentions did, 6.6 times against 5.6. One in three answers now points at their domain as a source, where it was one in twenty.
Average Google position: 3.2 to 1.2
Measured across three points in the engagement: 3.2, then 2.2, then 1.2. The technical work that makes a site legible to AI crawlers is largely the same work that makes it legible to Google, so this moved alongside the rest rather than as a separate project.
What it was worth
Sound Elite did $240,000 in revenue in 2025, the year before we started. They are currently projecting $550,000.
The gap that is still open
Both numbers improved sharply, but they did not converge. At the end of the engagement Sound Elite was named in 73% of answers and cited in 33%. A 40 point spread.
Being named is not the same as being linked. They had strong local authority already, which is what gets a business named, but not enough published material for engines to cite as a source. That is a content problem rather than a technical one, and it is the next phase of the work.
We have since run the same measurement for other clients, and it is what our later batches in Toronto and Singapore are built around.
Why this generalises
Nothing here was a trick. Engines cite sources they can read, parse and corroborate. A site that is technically legible and backed by consistent signals gets named. One that is not, does not.
It feels like a trick only because almost nobody is measuring it yet. The most useful thing about this engagement is not any single number, it is that the numbers existed at all. You cannot fix a gap you have never measured, and most businesses in this position do not know the gap is there.
A specialist local business that starts measuring now is competing against a field that still believes Google is the only scoreboard.