The challenge
Buyers stopped clicking and started asking. Discovery moved into AI answer engines (ChatGPT, Gemini, Google AI Overviews), where being cited matters more than ranking and the click often never happens. Traditional SEO was measuring the wrong thing.
The surface area was large: 37 products, 19+ web properties, and no playbook for getting a cybersecurity brand named inside an AI answer. The team was lean. A campaign wouldn't hold at that scale. It needed a system.
The insight
Optimize for inclusion and citation in AI answers, on top of a healthy technical foundation: AEO first, SEO second.
Every asset serves two readers at once: the buyer evaluating it and the model extracting from it. Get both right and they reinforce each other instead of competing.
What I built
I designed the program as an operating system with five layers. Each answers one question, so every new piece of work has an obvious home.
Operating manual: how it must behave
AEO first, SEO second. One question, one home: a single naming and messaging standard, so the whole company feeds the engines one story. A routing rule for every task. And task discipline: new work needs a named owner and deadline, or it has to move a measurable visibility signal.
Context: what's true
The product messaging and naming standard, the brand kit, entity and knowledge-graph health, a weekly SEO/AEO digest, and a monthly benchmark database. Program state carries forward day to day, so nothing starts from zero.
Infrastructure: what it can reach
AirOps, Profound, Semrush, Google Search Console, GA4, Amplitude first-touch data, Jira, Notion, and Claude.
Engine: what does the work
One rule decides where each task goes:
Scale never goes to the human. Judgment never goes to the machine.
- If the work is deterministic, high-volume, and schedulable, it becomes an always-on workflow. In AirOps, that includes AEO blog briefs and outlines, AEO YouTube video optimization, AI product perception reports, competitive analysis, deep market research, and content strategy analysis. Scheduled Claude workflows handle monitoring, reporting, and program updates.
- If it needs judgment, cross-source synthesis, or a narrative, it becomes a purpose-built Claude skill: SEO/AEO strategy and redlines, AEO blog writing, schema markup, technical SEO diagnosis, product-claim accuracy checks, data and tracking analysis, program launches and executive decks, and weekly priorities.
Processes: when it runs
The program runs on a fixed rhythm instead of on memory:
- Daily: an AI bot access monitor, with red alerts sent straight to my inbox
- Monday: the digest, the SEO/AEO report, a dashboard rebuild, and the weekly plan
- Tuesday to Friday: a daily update
- Friday: a weekly close and a stakeholder snapshot
- Monthly: Profound, Semrush, and knowledge-graph benchmarks
- On event: core-update monitoring
Swipe sideways to inspect the full diagram →
The results
Most-cited brand in our category's AI answers, ahead of CrowdStrike and Palo Alto Networks.
First-touch pipeline sourced by organic and AI search in H1 2026, including the company's first AI-search-attributed closed-won deal.
AI share of voice, at an average cited position of 1.65. When we appear, we're at or near the top of the answer.
AI search conversion versus the site average, making it the top-converting channel.
Search impressions year over year while clicks stayed flat. That's the zero-click shift, measured and turned into a pipeline strategy rather than treated as a traffic problem. Measured in Google Search Console.
What carries forward
The part that transfers isn't the tool stack. It's the architecture: rules, context, data access, skills, and schedules, each defined once and owned by one person or process. I've since turned it into a business-agnostic operating system that any growth team can run on.