How the AEO Flywheel Compounds: One B2B SaaS, One Gap, Six Cited Posts, a Real Share-of-Voice Shift

The AEO flywheel compounds when a single unanswered buyer question becomes the seed for a linked cluster of posts, each new citation exposes the next question the engines cannot answer well, and attribution proves which post pulled the visit. The result is not one lucky citation. It is a rising share of the answers in your category that name you, tracked week over week. Here is exactly how that plays out across a 90 day cycle for a funded early-growth B2B SaaS.

The company in this walkthrough is composite. The mechanics come straight from how Clark runs its four jobs, and the same loop has already been documented end to end in a funded B2B SaaS that closed its first ChatGPT-attributed sale in under 30 days.

The starting gap: one prompt, four engines, zero mentions

Day one, Prompt Monitoring asks the buyer's real question across ChatGPT, Claude, Gemini, and Perplexity: "best contract lifecycle management tool for mid-market healthcare." The daily brief comes back the same way for six days: three competitors named, our SaaS invisible in all four engines. That prompt goes amber. It becomes a content gap topic in Clark's queue, not because a keyword tool said so, but because measured demand shows the assistants have nothing good to quote.

Share-of-voice at week one: 4% across the tracked prompt set. That is the baseline the CEO gets on slide one.

Post 1: the answer-first primer that got quoted first

Clark writes the seed post: "What actually counts as a mid-market healthcare CLM, and how do buyers evaluate one?" The opening paragraph answers the question completely in three sentences, using question-shaped H2s that mirror how a compliance director would phrase things to an assistant.

Publish Monday. By Friday, Perplexity cites the post in two variants of the seed prompt. ChatGPT picks it up the following Tuesday. That is the first-order win. Most teams stop celebrating here and wait.

Posts 2 and 3: the second-order gaps the citations exposed

Here is where the loop gets interesting. Once the seed post is cited, Prompt Monitoring starts logging the follow-up questions users ask assistants after they see our name. Two new gaps surface within ten days:

• "How does mid-market CLM pricing compare across the top three tools?" Claude and Gemini answer this one confidently, but they still do not name us.
• "What integrations does a healthcare CLM need for HIPAA workflows?" All four engines hedge; no one owns the answer.

Clark writes both. The comparison post is deliberate and even-handed; the integrations post is a deep-dive grounded in the customer's own product documentation. Both publish within the same two-week window. Both get cited within seven days, this time on ChatGPT first.

Share-of-voice at week four: 11%. The leaderboard shift is small but the trend line is up and to the right, which is what a Head of Growth actually needs to show a board.

Posts 4, 5, and 6: the third-order compounding

By week six, the pattern is visible. Each citation on an engine spawns a shortlist of prompts that engine now handles better with our name in it, and a longer list of adjacent prompts where it still hedges. Clark writes to the hedges:

• A case-study style post on a specific redlining workflow, cited by Perplexity within four days.
• A mistakes post on the CLM procurement errors that cost healthcare buyers a quarter, cited by Gemini and ChatGPT.
• A best-of style shortlist post that positions the category honestly, cited by all four engines within two weeks.

Six posts. All grounded in real gaps. All aimed at the specific questions the buyers actually ask. Every one of them written from evidence, not from a keyword database. This is what Clark's question finder is built to do.

Share-of-voice at week twelve: 27%. That is not a vanity number. That is over one in four AI answers across the tracked prompt set now naming the company by default.

The measurable proof: attribution closes the loop

The flywheel only compounds because attribution is running underneath it. Every AI-referred session that lands on one of the six posts gets stamped with the source engine, tied to the conversion event it fires, and rolled into a per-post, per-engine snapshot. By the end of the cycle, the growth lead can point to specific ChatGPT-cited posts that produced named leads and dollars, then feed those winning topics back into the question finder for the next quarter's roadmap. Everyone shows you rank. We show you revenue.

What to take from this

Do not plan for a two week win. Plan for an eight to twelve week cycle where the first citation is the beginning, not the point. Track share-of-voice, not just appearance. And close the loop between which engine cited you, which prompt it answered, and which conversion followed, because without that, you are guessing at what to write next.

If you want to see your own starting gaps before you commit to anything, the free AI visibility check will show them in a few minutes.