Customer Impact is an SEO and GEO agency from Antwerp that helps B2B companies and online stores get found in Google and in the answers of ChatGPT, Gemini and Perplexity.
Since May 2025, Rankshift has been the GEO tool the team uses to measure that AI visibility. Founder Tanguy de Keyzer explains why they chose Rankshift, how they use it for client research such as the iBOOD audit, and what it taught them about their own brand.
Why a GEO agency needs a GEO tool
In Google you have a ranking you can check every day. In ChatGPT you don’t. Ask the same question twice and you get two different answers, sometimes with different brands in them. For an agency that promises clients they will be mentioned more often, that is a problem: without measurement, every result is an impression.
“In the beginning we tested everything by hand. Typing questions into ChatGPT, taking screenshots, putting them in a spreadsheet. After two weeks I knew this doesn’t scale and proves nothing. One answer is coincidence. You need hundreds of answers, week after week, to see a trend you can put in front of a client.”
Tanguy de Keyzer, Customer Impact
That is why Customer Impact looked for a GEO tool that does three things well: track your own questions across all major AI platforms, put competitors in the same measurement, and show which sources a model uses. Rankshift tracks nine platforms, including ChatGPT, Gemini, Perplexity, Claude, Copilot and Google AI Overviews, and does all three in one dashboard.
How Rankshift became a fixed part of the workflow
At Customer Impact, Rankshift is not a standalone report at the end of a project. It comes in at three points in every engagement:
- At the start: the baseline. Before a single page is rewritten, the team measures how often the brand currently appears in AI answers, next to which competitors, and through which sources.
- During the project: the weekly trend. Every week the team puts the numbers next to what happened: new knowledge articles, a mention in a trade publication, community work. That shows what has an effect.
- In reporting: the proof. Clients no longer get screenshots, but a curve over weeks and a list of sources that were added.
“Rankshift has changed how we report. I used to say: we think it’s getting better. Now I show a line going up, per platform and per market. That conversation is a lot easier, also with a CFO.”
Tanguy de Keyzer, Customer Impact
Research at scale: the GEO audit for iBOOD
The biggest project in which Customer Impact used Rankshift was a GEO audit for iBOOD, the European daily deals site active in the Netherlands, Belgium, Germany and Poland. iBOOD’s question was simple: how visible are we when someone asks ChatGPT or Gemini where best to buy something, and what does it cost us when we’re missing there?
The scale of that research:
- 624 real customer questions tested
- 6 AI platforms examined: ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews and Google AI Mode
- 7,259 AI answers analyzed
- 13 months of GA4 data reviewed to measure AI traffic and its value
Why classification makes the difference
With 7,259 answers you have a lot of data, but not yet insight. The insight came from how the team categorized the questions. Every question got labels: the product category (such as electronics, household or garden), the type of question (informational, buying advice or comparison), the market and the language. Every answer was labeled on brand mention, on which competitor was mentioned instead, and on which source the model cited.
That let the team see per category where the brand was strong and where it disappeared completely. A brand can be mentioned often in one product category and be entirely absent in another. You don’t see that in the overall figure, you do see it in the classification.
“The overall figure doesn’t lie, but it doesn’t tell you anything either. Only when we split the 624 questions by category, by question type and by market did we see where iBOOD was winning and where the competitor was walking off with the customer. Rankshift gave us the raw answers and the sources at scale, the classification turned it into a roadmap.”
Tanguy de Keyzer, Customer Impact
The citation sources turned out to be the most useful part. The team mapped the websites AI models cite that mention iBOOD’s competitors but not iBOOD. That list of gap websites became a concrete action list. iBOOD immediately got to work internally on the quick wins from the report.
Their own brand as a testing ground
Customer Impact tests every approach on its own brand first. Tanguy built a Reddit strategy to get mentioned in AI answers and tracked the effect in Rankshift. He described that approach in full in his guide Reddit AI SEO.
The measurement showed the progression in black and white: a mention rate below 2% in the first month, around 8% in the second month and above 20% in the third. In the end he reached the top 1% of ChatGPT mentions and position 1 for his target questions.
“The moment I knew it was working was when Rankshift showed our brand in answers where the model hadn’t even searched the web. Then the model has learned you. Without that measurement I could never have proven that, and I wouldn’t have believed it either.”
Tanguy de Keyzer, Customer Impact
What Customer Impact does differently because of the data
- Measuring per language. Belgian clients are measured in Dutch, French and English. A brand that is mentioned often in Dutch can be completely absent in French.
- Sources over positions. Instead of chasing a single answer, the team works on being present on the sources models cite: trade media, comparison sites, communities and strong knowledge articles on the brand’s own site.
- A stable prompt library. Per client, 20 to 30 category questions without a brand name, 5 to 10 brand questions and 5 to 10 comparison questions. The core stays the same, so the trend remains comparable.
- Being honest about correlation. A GEO tool shows correlation, not causation. That is why the team keeps track of what changed and when.
Conclusion
“AI visibility is the new shortlist. If you’re not on it, you don’t even get asked. A GEO tool like Rankshift tells you whether you’re on it, next to whom, and why. The rest is work.”
Tanguy de Keyzer, Customer Impact
Want to know how visible your brand is in AI answers and what you can do about it? Customer Impact combines measurement with execution as a GEO agency.
Frequently asked questions
What is a GEO tool?
A GEO tool measures how often AI assistants such as ChatGPT, Gemini and Perplexity mention your brand, next to which competitors and based on which sources. It is the counterpart of an SEO tool, but for answers from language models instead of rankings in Google.
How many questions should you track?
For your own brand, 30 to 50 questions are enough to see a reliable trend. For an audit of an online store with many product categories, as with iBOOD, you go to hundreds of questions so you can draw conclusions per category.
How quickly do you see results?
New citation sources often appear within a few weeks. A stable rise in mention rate usually takes two to three months of consistent work.
