Making a stealth-mode AI company the source AI assistants cite

The situation

In April 2026, axonis.ai had a Domain Rating of 1.6, eight referring domains and a site health score of 11. The blog had no canonical tags and no article schema. There was no structured Google Ads program, and the LinkedIn awareness campaign was running at a 0.07% click-through rate with most impressions served outside the United States. Nobody had measured how AI assistants answered when a buyer asked about Axonis by name.

The goal was to make Axonis findable, citable and credible wherever enterprise and defense buyers research AI infrastructure: in search, in AI assistants, and in their LinkedIn feed.

The method

Authority is built in a particular order. Engines and AI models have to be able to read the site correctly before links to it count for much, and links have to exist before paid traffic lands on pages anyone trusts. So the work ran foundation first, then authority, then demand, with measurement running the whole way through.

• Rebuild the technical foundation. One self-referencing canonical on each of the 9 page templates and 32 blog URLs, with the sitemap to match. BlogPosting schema on all 32 posts, with author, publish date, modified date, image and publisher. A restructured FAQ page, a corrected homepage JSON-LD, and fixes for duplicate titles and missing alt text. The sitemap was rebuilt in Webflow and submitted to Bing, the index behind ChatGPT search. Every fix shipped through two rounds of step-by-step developer instructions and a developer Q&A.
• Earn authority from the sources AI models already cite. A backlink and AI-visibility strategy aimed at those sources, a long-form Medium article and G2 listing copy that put the Axonis story where buyers and models look, and a Wikidata entry and Google Knowledge Panel plan to establish Axonis as a recognized entity.
• Build paid media around the real buyer. Google Brand Protection, Search and Remarketing campaigns launched in late April, Display and Performance Max followed in May, and targeting was tightened to the US in June. LinkedIn was rebuilt around US and North American operator, vertical and conference-attendee audiences, with 400 negative keywords across 8 shared lists protecting the Google budget.
• Measure every week. 40 buyer questions tracked across six AI platforms and benchmarked against 17 competitors, plus a weekly dashboard covering SEO, AI visibility and AI citations. Google Ads changes ran as a weekly diagnosis and change plan, with a person approving every change.

The result

Search and AI visibility

Site health moved from 11 to 92. When a buyer asks ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews or Google AI Mode about Axonis by name, Axonis averages rank 1.02 in those answers, so it is listed first almost every time. Every platform now mentions Axonis in at least 93% of these answers, and Google AI Mode and Claude reach 100%. In head-to-head comparison questions, Axonis's presence rose from 89% to 97%.

Citation rates rose on ChatGPT (20.4% to 27.9%, flagged by Scrunch as a high-confidence movement), Gemini (19.0% to 30.1%) and Google AI Mode (23.8% to 27.5%). Unique axonis.ai URLs cited grew from 7 to 24 in a single month. The most-cited pages are /platform, /faq and /decision-intelligence; after the FAQ rewrite, that page's weekly citations rose from 2 to 17.

Paid media

On Google Ads, click-through rate rose from 3.25% to 3.87% between the first full week and mid-June. The June vertical and audience pivot delivered 6.5x the reach on flat spend, with impressions growing from 20.6K to 134.2K, and new users on the site rose 237%. Over the 30 days from July 25 to August 23 the account produced 107 contact requests, 79 of them from Performance Max remarketing.

On LinkedIn, the rebuilt campaigns ran at engagement rates between 4.1% and 4.6% across the Defense and Intel, General Brand, Government and Smart Cities, and Ai4 conference audiences. Defense ads drew clicks from the US Army, US Air Force, US Navy and Canadian Armed Forces; the Ai4 campaign drew clicks at a 3.98% click-through rate from Deloitte, IBM, Amazon and the University of Washington.

What the numbers do not say

Referring domains fell from 381 to 374 between September and October, which is ordinary month-to-month churn, not a trend. The AI-visibility figures cover questions that already name Axonis; they show how AI assistants answer when a buyer asks about Axonis, not how often Axonis appears in unbranded category questions, which is the next phase of the work. Presence was already high in April, at 92.2%, so the larger movement is in citations of axonis.ai as the source. The 30-day lead count is one window in a program that was still being tuned, and lead quality is Axonis's to judge. The share of site traffic from Google and LinkedIn is the client's figure, from their analytics, and is reported here as they stated it.

What the client said