TL;DR. Share of voice in AI search is the percent of tracked buyer prompts whose answers name your business. It is one number, trending, and it is how you manage AI visibility like revenue instead of hoping. By the end of this project you can run the full measurement cadence: a weekly share-of-voice check, a monthly deep read, and a quarterly refresh, in about 30 minutes a week.
- Watch us re-measure our own share of voice (the receipts)
- What is share of voice in AI search?
- What does measuring do in the real world?
- Why is Step 7 forever?
- How do you set the measurement cadence, step by step?
- How do you make this your own?
- How do you train someone else on this?
- Where do you share your result?
- Frequently asked questions
Watch us re-measure our own share of voice (the receipts)
This is Project 7 of 7 in the AI SEO playbook. It is the shortest habit and the one that makes every other project compound: what gets measured compounds.
What is share of voice in AI search?
Take your 50 tracked prompts from Project 2. Ask them. Count the answers that name your business. That percent is your share of voice. Named first counts double in spirit: it is the difference between being the recommendation and being the alternate.
The reason it has to be a tracked list, not vibes: AI answers vary. One answer changing means nothing. A pattern across fifty means everything. Share of voice turns a noisy, invisible battlefield into one honest number that only has one job: move up.
What does measuring do in the real world?
Every client on our books has this number on a live dashboard, next to their bookings and revenue, because visibility that does not connect to money is a hobby. The sauna brand from Project 1 started at 17% named; the point of the scan was not the grade, it was the trend line that every following month gets judged against. Advanced Recovery Cryo's "first answer in its market" status (the case study) is not a trophy, it is a weekly number being defended.
And the sneaky-powerful metric is not on any dashboard: it is the "How did you hear about us? Google / ChatGPT or AI / referral" line you added to your intake form in Project 2. The monthly count of "heard from AI" leads is the number that pays the bills.
Why is Step 7 forever?
Because the incumbent's lead compounds monthly. Whoever is the answer in your market today is collecting the reviews and citations that make them the answer next year too. Measurement is how you catch a slip in week two instead of quarter three, how you prove which project's work moved the number, and how you defend the position once you take it. Steps 1 through 6 win the ground; Step 7 holds it.
How do you set the measurement cadence, step by step?
The drill. Set three recurring calendar blocks today:
- Weekly (30 minutes): re-run a slice of your 50 prompts. Ten a week keeps the whole list covered monthly with a weekly pulse. Log the one number: share of voice, trending.
- Monthly (30 minutes): the deep read. GSC clicks and position for each cluster from Project 6, plus the count of "heard about us from AI" on your intake form. Two numbers next to each other: visibility and money.
- Quarterly (one evening): genuinely refresh your key pages. New data, new questions, new photos. Engines discount cosmetic date-bumps; they reward real updates.
- Grade yourself on the 0 to 5 AI-awareness score each quarter: 50-prompt list running (+1), fundamentals solid (+1), top 5 pages answer-shaped (+1), on at least one page the AI cites (+1), tracking share of voice weekly and AI leads monthly (+1). Most local businesses score 0. A 3 beats nearly every competitor you have. A 5 makes you the incumbent the machines defend.
We shipped all seven projects on ourselves in one day, then re-ran the audit 10 across both engines the same evening. The number did not move: ChatGPT 2 of 10, Perplexity 1 of 10, identical to the morning baseline. We published the flat number on purpose, because that is the whole lesson of this page: engines move on weeks of reviews, citations, and crawls, not on one heroic day.
The cadence is now on the calendar: ten prompts every Monday (the weekly pulse), the monthly deep read (GSC per cluster plus "heard from AI" leads via our score form's source tags), and the quarterly refresh. Our one number is share of voice on the audit 10. On the 0 to 5 AI-awareness score we now grade ourselves a 4: the list runs, the fundamentals shipped, the pages are answer shaped, we are tracking weekly; the missing point is being on a page the AI cites, which is Project 5's pitch list in flight. Watch the scoreboard climb, or catch us confessing if it does not.
The receipts live in the Project 7 finale post in our build-in-public thread on X.
How do you make this your own?
Your upgrade: pick YOUR one number. Some owners run on share of voice, some on "heard from AI" leads, some on first-position answers for the three money prompts. Choose the single number that connects most directly to your revenue, put it somewhere you see every Monday, and let the rest be supporting detail. One number gets managed; five get ignored.
Software company? Split the number in two, mentions and citations, because most AI citations never say your name: the B2B SaaS upgrade explains the ghost-citation gap. Product brand? Add carousel inclusion and star-rating health to the weekly number: the eCommerce upgrade has the scorecard.
How do you train someone else on this?
Trade numbers monthly with one accountability partner: another owner running this playbook. Fifteen minutes: each shares the trend, the one thing that moved it, and the next month's single bet. Explaining WHY your number moved is where the diagnosis skill gets built, and a witness makes the cadence unbreakable.
Where do you share your result?
Post the trend, not the snapshot: "90 days ago AI named us in 2 of 10 buyer questions. Today it is 6 of 10, and 9 customers this month told us an AI sent them." Numbers with dates are the most credible marketing there is, and the post itself becomes another source machines read. Link back here and tag us; the best trend lines get featured on the playbook.
Score yourself: tracking share of voice weekly and "heard from AI" leads monthly is the fifth and final +1 on the playbook's 0 to 5 AI-awareness score. That is the full curriculum. If you want the machine to run the measurement for you, that is literally what our dashboards do all day.
Frequently asked questions
What is share of voice?
Share of voice in AI search is the percent of your tracked buyer prompts whose answers name your business. Tracked across a fixed list of 50 prompts and re-run on a schedule, it is the single trending number that tells you whether your AI visibility is growing or slipping.
What should SEO reporting include for a local business?
Three things, monthly: share of voice across your tracked prompts, Google Search Console clicks and average position per topic cluster, and the count of leads who said they heard about you from AI or Google on your intake form. Visibility, rankings, and money, one page.
How often should I re-run my prompts?
A slice weekly (ten of fifty keeps a weekly pulse), the full list monthly. Weekly catches slips early; monthly gives you the honest trend. A single snapshot is noise.
When do results start showing?
Answer-shaped pages and schema can influence AI answers within weeks. Citations and clusters compound over months. The measurement habit is what tells you which is happening, and either way the number should only move one direction.
Two ways to run this playbook. Get your free AI Visibility Score in about 30 seconds: 50 prompts, 3 AI engines, scored live, your baseline measured for you, plus a free 20-minute walkthrough with Matthew. Or get The Playbook and keep doing it yourself.
