A dossier, not a brief
One topic, researched once. Facts only you can say, plus cited web claims — not a prompt stuffed with keywords. Both go into the dossier before a word is drafted.
Home
/
- ChatGPT
- Gemini
- Google AI Overviews
- Google AI Mode
- Grok
- Perplexity
- Claude
AI Visibility
Visibility
Position, Share of Voice and Sentiment, Over Time
Every run scores the answers that come back. You get your average position and mention count per category, plotted across runs, with the latest run's distributions underneath and a ranked table of which brands the engines named most.
- Your average position and total mentions per category, per run
- Ranked table of every brand the models named — who, and how often
- Sentiment, share of voice and mention rank distributions for the latest run
- Filter by engine, category, prompt type and date range
- Versioned methodology — when we change a probed model, the timeline flags the date
Six engines. We set the cadence.
ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Grok, Perplexity and Claude are included. LLM analysis is weekly. Google AI surfaces are checked daily. Daily LLM analysis is Enterprise. You edit the questions. We set when they run. Prompts are grouped by intent, so unaided awareness and reputation are measured separately rather than averaged into one number.
The researched cluster
Two topical clusters a month, done properly
Monitoring is the instrument. The researched topical cluster is the intervention. Search Console and site traffic are how you know it worked.
A pillar and a few supports
From that dossier: one pillar page and three to five supporting articles. Enough to cover the topic. Not enough to teach the models you are noise.
Thin ideas do not ship
Ideation rejects thin pages, duplicates, and “Complete Guide to X.” That gate is real. We will not help you publish thirty uncited posts a month.
Write
Your numbers in the dossier
Measurements, usage and observations only you can say go into the dossier with the cited web claims, before a word is drafted.
Every piece cited
The pillar and supporting pieces are written from that dossier. Sources sit on the article.
You review before live
You review each piece before it goes live. That is the default.
Keyword Intelligence
Mention Rate and Google Rank on the Same Card
For each keyword you track, BrandKarma shows how often you and each competitor you track are mentioned in AI answers, which domains the engines cite as sources, and where you currently sit in classic Google results — with difficulty, cost per click and search volume alongside.
- Mention rate for you and each tracked competitor, with change since the previous run
- Cited domains, with citation counts — see who the engines treat as the source
- Keyword difficulty, cost per click and monthly volume
- Google position and clicks for the same keyword, from Search Console
The Join
Measurement stack
Four Sources, One Digest
An AI visibility score on its own is a number without a consequence. The dashboard puts AI mentions, your actual site traffic and classified web mentions on one screen — and the digest reads all three against your Search Console history.
- Google Search Console — link a property and we pull sixteen months of daily site history in the background, plus three months of per-keyword rows
- First-party web analytics: sessions, visitors, pageviews and traffic sources
- Cookieless by design — the beacon writes no cookies, no localStorage and no sessionStorage, so it adds no consent banner of its own
- Paid attribution without ad cookies: click IDs captured on landing and carried to your app
- Classified web mentions: tone, page kind, claims, a key quote and a suggested action — not a raw dump
- A scheduled digest written across AI answers, search, traffic and mentions
Integration
MCP server
Point Claude or any MCP client at BrandKarma and work in conversation: read analytics, manage monitored prompts, edit brand configuration, research keywords.
Run the loop from an agent →
Headless content API
Fetch published content by brand, language, category, tag or slug and render it yourself. This site’s blog runs on it. API documentation.
Tracking SDK
@brandkarma/tracker on npm, with React
and Next.js App Router bindings — or a one-line image beacon if you would rather not install
anything.
The Loop
- 01Measure
Six engines, Search Console and site traffic. We set the cadence.
- 02Research
One topical cluster: a dossier of cited claims, not a keyword list.
- 03Write
A pillar and three to five pieces you would actually publish.
Then we measure again — same instrument, not a fourth step. Published pieces are pulled from one API. We will not help you flood.
What BrandKarma is NOT
Not a content mill
We will not help you flood. Thin, duplicate and “Complete Guide to X” ideas are rejected.
Not a monitor-only SKU
Measurement without a way to change the number is a vanity metric. The cluster is the intervention.
Not a PIM
We do not manage your product data. That is a different product.
A headless CMS you do not have to use
Published content is delivered over a read API. Render it in your stack, or ignore that half of the product entirely.
Not a black box
You can open any score and read the exact prompt, the engine and the raw answer behind it, and every run is stamped with the model set that produced it.
Not a rank tracker with AI stickers on it
The point is the join: what the engines say, next to what Search Console and your own traffic recorded.
See what AI says about you
Request a free AI Visibility Report. We run your top categories across ChatGPT, Perplexity and Gemini and send it within 48 hours.
See what AI says about you