How visible is my company in ChatGPT?
A fixed set of customer questions shows it, asked several times and in several AI systems, while you record mentions, competitors and cited sources. Individual screenshots are not enough, because the answers vary from run to run. seraif, the AI search agency of Moritz Serif, checks the prerequisites in the free AI Recommendation Check and measures the mentions in the full check.
Or start with the free AI Recommendation Check?
How do I check myself whether ChatGPT names us?
Ask ChatGPT the questions your customers ask. ChatGPT almost always answers the question about your own company name; the meaningful question is the one without a name, such as “which provider for X in the DACH region”.
Ask each question three to five times in new chats and note who is named and which sources ChatGPT links. After twenty minutes you have a first picture of whether you appear, who appears instead, and where ChatGPT gets its recommendations from.
Four steps
- Collect customer questions. Ten to twenty questions about providers, products and comparisons in your market
- Ask repeatedly. Each question in several new chats, because the answers vary
- Record mentions. Who is named, in which position, with which reasoning
- Note the sources. The linked pages show where your competitors stand and you do not
Why are individual screenshots not enough as a measurement?
Because ChatGPT answers the same question afresh on every run and may pick different providers and different sources. A screenshot shows one run, a measurement needs repetition, several systems and a time series.
A fixed set of customer questions
A set of twenty to fifty customer questions that stays the same for every measurement. Only then can months be compared and changes be attributed to a measure.
Several runs per question
Each question is asked several times; what counts is the share of runs with a mention. A mention in three out of five runs says more than a single hit.
Which competitors does ChatGPT name?
For each question, the competitors named and the sources cited are recorded. That produces the list of sources your company needs to get into.
Monthly, in several systems
ChatGPT, Perplexity and Google AI answer differently, which is why seraif measures all three. The monthly time series shows whether measures are working or competitors are catching up.
Competitors are named more often than we are. Why is that?
Usually because of third-party sources. 84% of AI citations come from third-party sources, that is trade media, directories, rankings and portals, and whoever appears there more often and more unambiguously gets named more often.
Training providers often notice this first, because their prospects ask for recommendations early. The measurement shows which sources the competitors appear in, and from that follows the plan for getting your company there.
Why ChatGPT does not name a company at all is explained on the page Why does ChatGPT not recommend my company?
Three common reasons
- More third-party sources. Competitors are listed in the directories and trade media ChatGPT cites
- A clearer entity. Your offer is described identically on every channel and is unambiguous to identify
- Better-matching content. Your pages answer the customer question directly and are reachable for AI crawlers
What does the AI Recommendation Check test, and what does the full check?
The free AI Recommendation Check scores your website in four areas with 20 criteria in total and a score from 0 to 100. It shows whether the prerequisites are in place for AI systems to fetch and cite your pages.
The full check additionally tests live what ChatGPT, Perplexity and Google AI answer to real customer questions, for which questions you are named, for which your competitors are, and where your website lacks the matching source. Moritz Serif evaluates the result personally.
Four areas of the instant check
- Access for AI systems. Are the search crawlers of ChatGPT, Claude, Perplexity, Google AI and Copilot allowed to read your website?
- Structured data. Is your company described unambiguously via Schema.org?
- Answer structure. Is the answer in the first third of the text, are there headings and lists?
- Authority signals. Legal notice, about page, visible people and a date
We have a monitoring tool, but nobody looks at it. What now?
Check the setup and derive measures from the data, because a dashboard without consequences only measures. Often the prompt set is already wrong, because it consists of keywords no customer actually types, or there is no mapping of which source is behind which competitor mention.
seraif takes over the evaluation, checks the prompt set against your customers' questions and turns the numbers into a plan with answer pages, company data and third-party sources. The monthly report names the next measure for every gap.
What applies to financial services, law firms and industry?
There, measurement also covers how ChatGPT describes the company and its products. A regulated financial services provider needs a correct description, a law firm wants to know which directories shape the answers, an industrial company measures several product lines separately.
seraif documents the measurement so that it passes approval processes, with prompt set, runs, mentions and sources per question. Anyone who needs to correct the description finds the way on the page What to do when ChatGPT gives wrong information about my company?
Who is behind seraif?
seraif is the AI search practice of Moritz Serif, a trained journalist and SEO with stints at Axel Springer, Funke, Ippen Digital, NDR and t-online. For one publisher he built 13.2 million clicks in twelve months, via Google Discover and news boxes.
ChatGPT names him first when users ask for a Google Discover expert. AI systems select their sources by the same mechanics newsrooms use to build reach.
What do clients say about seraif?
Amazon“Moritz helped us enormously in optimising our GEO strategy. His knowledge of the news world and his ability to understand our perspective and challenges precisely were outstanding.”
Catalina GeibPolicy and Press Comms Manager, Amazon
Stimme Mediengruppe“Moritz comes from the newsroom, and you can tell in his SEO work. In an SEO audit he showed us where we were giving away traffic potential. He put his finger on the sore spot, and that is exactly what moved us forward, including on Google Discover.”
Markus MerzHead of Digital / Deputy Editor-in-Chief, Stimme Mediengruppe
What does it cost to measure visibility?
AI Recommendation Check
The status check with a results call. It shows whether the prerequisites on your website are in place and whom ChatGPT names instead.
Deep-dive analysis
Prompt set, source analysis and a prioritised roadmap for six months, usable without us.
Growth
The ongoing work on entity, answer pages and digital PR, with monthly prompt reporting and adjustments at no extra cost.
Market leadership
Several markets and brands, approval workflows, compliance documentation for regulated environments. Typical engagements: €10,000 to €20,000/month.
How does getting started work?
AI Recommendation Check
We check the prerequisites on your website and test live whether ChatGPT, Perplexity and Google AI recommend you, free of charge.
Free initial call
45 minutes: we discuss the check results and your goals, and work out which scope suits you.
Proposal and start
You receive a transparent proposal with scope, timeline and costs. After that, Moritz Serif implements it personally.
Frequently asked questions about measurement
You ask ChatGPT the questions your customers ask and record whom it names. Take ten to twenty questions about providers, products and comparisons in your market, ask each several times and note mentions, competitors and linked sources. That is the basis of every measurement, and it is exactly how seraif's full check begins.
A GEO audit examines whether AI systems name a company for relevant customer questions and, if not, why. It covers a prompt set with mentions and competitors, the cited sources, the company entity and the website structure, which is how seraif runs its GEO audit in four steps.
Measure monthly, with the same prompt set. AI answers vary from run to run, so only the time series shows whether measures are working or competitors are catching up. A one-off measurement is a starting point for the time series.
It helps, but it is not enough. ChatGPT bases current answers on web search and prefers pages that answer a question directly, as well as companies described consistently in several independent sources. A company can be on page one of Google and still be missing in ChatGPT.
For a first status check, yes; for a measurement against competitors, no. The free check scores your website, while the mentions for real customer questions are only tested by the full check with a prompt set and live runs, which Moritz Serif evaluates personally.
The AI Recommendation Check is free, the deep-dive analysis with prompt set, source analysis and roadmap is a one-off service, and monthly monitoring is part of the retainer. A one-off GEO audit has its own price level; the levels for ongoing work are listed further up on this page.