AIV Index · share of mention, tracked monthly
We measure how often ChatGPT, Claude, Perplexity, and Gemini name you when buyers ask — every month, against your competitors.
A buyer asks ChatGPT
“Who are the best estate agents on Cape Town’s Atlantic Seaboard?”
Buyers are pointed to Pam Golding and Seeff. Whether your brand is in that answer is what AIV measures.
We check all four engines, every month
ChatGPT · Claude · Gemini · Perplexity
Share of mention · example
Being named is only the first rung
The question that decides a buyer’s next purchase is now answered by an engine that names two or three brands and moves on. Whether one of them is you is invisible from where you sit.
More buyers ask AI first
More buyers now ask ChatGPT, Claude, Perplexity and Gemini directly, and get one answer instead of a page of links.
No answer dashboard
You cannot rank in an AI answer the way you rank in ordinary search results. Google now reports how often its AI features showed a link to your site, but no engine shows you what its answers say about you or your competitors.
Visibility can be moved
In a lab test with fixed sources, adding citations, quotes and statistics lifted a page’s visibility by up to 40% (Aggarwal et al., KDD 2024). In a full search pipeline, rewriting body text alone lowered it, while better titles, headings and metadata raised it (SAGEO Arena, KDD 2026). Visibility moves with what the engines can read and quote, which is why it is worth measuring before and after.
569M
GPTBot requests measured in a single month, on one hosting network alone
Vercel / MERJ, Dec 2024
28%
of Googlebot's crawl volume: what the four main AI crawlers, combined, ran at
Vercel / MERJ, Dec 2024
+40%
the most that adding citations, quotes and statistics lifted a page's visibility, in a lab test with fixed sources
Aggarwal et al., KDD 2024
0
engine dashboards that show what AI answers say about your brand
Google’s AI report (31 Aug 2026) counts link impressions only
Per engine
You might own Claude and be invisible on ChatGPT. We break your visibility down engine by engine, so you know where to fight.
Share of mention by engine · example
Here is one answer of the kind the engines give, built end to end — then whether you were in it.
A buyer asks · ChatGPT · live web search
“Which estate agency should I sell my Cape Town home with?”
For selling a home in Cape Town, the agencies that keep coming up are Pam Golding, Seeff, RE/MAX, Rawson, Chas Everitt — drawn from Property24, PrivateProperty and review sites, not from the agencies’ own websites…
Share of mention · this question, 8 asks · example
Pam Golding6/8
Seeff5/8
RE/MAX4/8
Rawson3/8
Your brand
0/8
A brand left out of every answer is never told. That’s the blind spot AIV closes.
See if you’re on the listPublished research
Edition 2 — October 2026 — every prompt panel and the full method published. The same measurement, run for your brand, is what you get here.
3
industries
1 114
grounded answers
108
public prompts
The methodology is published. Each engine’s raw responses and the sources it consulted are stored against your tenant for audit — every answer readable in full on the dashboard. You can buy a refresh between sweeps, and you can take the data to a consultant who is not us.
Twenty to a hundred priority prompts your buyers actually ask (we provide an industry default set) plus three to five competitor domains. If your brand has a Wikidata entry, we use it so every mention is matched to the right company.
Every month we issue each prompt to ChatGPT, Claude, Perplexity, and Gemini — with live web search on, the way the consumer apps actually answer, not from memory. Each prompt runs three times per engine, five on Scale and on your highest-priority prompts, to smooth out the randomness; we report the aggregate, not a single answer.
Share-of-mention and the sources behind every answer — on every tier — including which sources get cited in your category instead of you. Pro adds conviction (does an engine recommend you, or merely name you?), auto-flags competitors surfacing in your answers you're not tracking yet, and adds Slack and Telegram to the email drift alerts every tier gets when your share moves materially or a rival overtakes you in an engine.
Cancel any time from your PayFast account. No VAT is added: we are not a registered VAT vendor.
One brand, a monthly sweep across all four engines. For founders and small SA businesses who want to know where they stand before spending on it.
A deeper prompt set, Slack and Telegram alerts, and a written summary of what moved and why from your second run on. For brands actively investing in AI search visibility.
Multi-brand measurement for agencies and enterprise teams — 3 brands included, add more at R1,200 a month each.
Need a reading between monthly sweeps? On-demand refreshes run at R100 / R250 / R500 per run by tier. Start tracking my brand.
Open what matters. Anything not here: hello@autoalphaadvisory.co.za.
When a buyer asks ChatGPT, Claude, Perplexity, or Gemini a high-intent question (“best audit firm in Cape Town”, “SA fintech compliance tool”), the engine answers with a small list of recommendations. AI visibility is how often your brand appears in that list, in what position, and which sources sit behind the answer. We measure all three.
Traditional SEO ranks you in Google’s ten blue links. AI engines answer in prose and cite a handful of sources; they name two or three brands and move on. You can rank #1 organically and still be absent from the answer a buyer gets from ChatGPT, Claude, Perplexity or Gemini. The Vercel/MERJ Dec 2024 study found that most AI crawlers, GPTBot among them, do not execute JavaScript, so a site that renders fine for Googlebot can be a blank shell to GPTBot. SEO is necessary; AI visibility is the other half.
Yes — raw single-shot answers are noisy. That is why each prompt runs three times per engine, five on Scale and on your highest-priority prompts, with live web search, and we report the aggregate, not a single response. Drift alerts only fire when share-of-mention moves by 10 percentage points or more between runs — random flicker stays below the threshold. The methodology is published in full.
Both, and the gap between them is where the value is. A plain mention count can’t tell “named in passing” apart from “actively recommended” — but buyers only act on the second. On Pro and Scale we score conviction per engine: of the answers that name you, how many actually recommend you. (In our SA property Index, Edition 2, October 2026, the category leader was recommended in 97% of the answers that named it; the two property portals were recommended in 36% and 50% of theirs.) Pro and Scale also auto-surface competitors turning up in your answers that you don’t yet track. And on every tier you see which sources the engines cite in your category — so you know who’s being quoted instead of you.
Four differences. (1) ZAR pricing without the international markup; Starter at R650. (2) The product is bundled with AAA’s audit engine — every Pro and Scale tenant gets a site-readiness check from the same engine: can AI crawlers actually read your pages, and are you listed on the third-party sites AI engines rely on for your category — supporting context, not just mention counts. (3) Local methodology grounding: prompts are tuned for SA buyer-intent language; competitor sets default to SA-region brands. (4) Nothing is hidden: the methodology is published, and every raw answer is readable on your dashboard. Side-by-side comparisons, with dated sources, are on the Compare page.
Yes — AIV Index is live. Sign up from any ‘Start tracking my brand’ button; no waitlist, no invitation. Pricing starts at R650/month (Starter).
Yes. Existing AAA retainer clients get the first month at the Starter tier price for any tier, by arrangement. Email hello@autoalphaadvisory.co.za to add AIV Index to your existing engagement.
Coverage: ChatGPT, Claude, Perplexity, and Gemini — each queried with live web search enabled, the way the consumer apps answer. We re-evaluate quarterly and add engines that hit measurable buyer-side share — Mistral Le Chat, DeepSeek, Grok, and others are on the watch list. Local AI products (e.g. South African finance bots) are considered for Scale tenants on request.
One domain. Monthly sweep across ChatGPT, Claude, Perplexity, and Gemini. The first number is the hard one to hear: it’s where you actually stand.
The AI Visibility Brief — Vol. 01 — By Auto Alpha Advisory