+3,573.8%
Account sign-ups
2026* vs 2024
+3,612.5%
Subscriptions
2026* vs 2024
+1,396.1%
Mentions in AI Overviews
Nov 2026 vs May 2025
+274.1%
Mentions in ChatGPT
Nov 2026 vs Nov 2025
*2026 data runs through September, so final growth will be higher.
Objective
Holded wanted to attract more new users (account sign-ups) and then convert them into monthly paying subscribers.
To achieve this, the work focused on two fronts:
AEO
Answer Engine Optimization
Make Holded's content appear as the direct answer in search engines, for example in Google AI Overviews, featured snippets and Bing answers.
GEO
Generative Engine Optimization
Make language models (ChatGPT, Gemini, Perplexity, Copilot…) cite and mention Holded when a user asks about invoicing, accounting or business management.
The goal was to gain visibility on every channel where users look for answers, combining three levers:
Organic
Better organic rankings in Google and Bing.
Citations
More citations (links to holded.com) in AI answers.
Mentions
More brand mentions in LLMs, even without a link.
Our approach: three starting premises
Before acting, we set three ideas that guided the whole strategy.
Every LLM and every index plays by its own rules
ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews do not choose their sources the same way. Each uses its own index, weighs each type of source differently and presents answers in its own way. So we did not apply a single recipe: we analyzed each platform separately and adapted the actions to each one.
Brand mentions matter more than citations
A citation is a link to holded.com inside the answer. A mention is the model naming Holded as a solution, with or without a link. What users read in the answer is the brand, not the source. If the LLM recommends Holded when someone asks “which invoicing software should I use?”, that mention influences the decision more than a link at the bottom of the answer. That is why mentions were our main KPI.
Citation
A link to holded.com inside the answer.
Mention Main KPI
The model names Holded as a solution, with or without a link.
Improving organic rankings in Google and Bing improves visibility in LLMs
The main AI assistants query search engine indexes to build their answers: Gemini and AI Overviews use Google, while ChatGPT and Copilot rely on Bing. If Holded ranks better in those indexes, it is more likely to appear in AI answers. Traditional SEO thus became the foundation of the AEO and GEO work.
What did we do?
Working on visibility in LLMs is not like working on traditional SEO.
In Google we have real data: search volumes, keywords and positions that come from a database.
LLMs
In LLMs there is no public database of what people ask, so research starts from hypotheses about what each user would ask.
LLM answers are also probabilistic: the model generates each answer on the spot, and the same question can return different answers depending on when, how and who asks. That is why we do not look for Holded’s “position” in one specific answer, but for how often it appears across many answers to similar questions.
- 1
We defined buyer personas to understand what kind of queries they make
We built different Holded user profiles (for example, freelancers, SMEs and accounting firms) with their needs, doubts and way of phrasing things. From them we deduced which prompts each profile would write in an AI assistant. That way we worked with real questions, not isolated keywords.
- 2
We identified which prompts trigger brand mentions
We ran those prompts on the main LLMs and analyzed the answers. We detected which types of questions made Holded get named and in which ones competitors appeared instead. This gave us a map of where we were already present and where there was opportunity.
- 3
We designed a strategy to gain presence in those answers
With that map we prioritized the actions: which content to create, which to optimize and how to present the information so models use it and recommend Holded as a solution.
- 4
We used Google's and Bing's tools to see which URLs have the most visibility
We relied on the data the search engines themselves provide (Google Search Console and Bing Webmaster Tools) to find out which Holded URLs have the most visibility in their indexes. Since those indexes feed LLMs, these URLs were the basis for prioritizing the work.
Results
Sign-ups and subscriptions
*2026 data runs through September, so final growth will be higher.
AI Overviews (Google)
ChatGPT (Ahrefs)
The Holded case shows that visibility in AI answers translates into business: more brand mentions, more account sign-ups and more paying subscribers.
Want to rank in search and AI ahead of your competition?
Book your free SEO + GEO strategy



