By Daniel Espejo, Founder & CEO at Omnia. 22 August 2025

How to identify the prompts that decide whether your brand appears in ChatGPT, Gemini, or Perplexity

In this article:

• Why prompts matter more than keywords

• What is a decision prompt and how to recognise it

• AI crawlers and their impact on conversion

• How to identify and audit prompts step by step

• How Omnia simplifies this process

• Conclusion

Why prompts matter more than keywords

Search is changing. Google no longer shows just links, but integrates AI-generated answers. AI Overviews was launched in May 2024 for users in the United States with the promise of ‘doing some of the search work for you,’ and Google said that it would expand to hundreds of millions of users that same year.

This change directly impacts how we measure digital visibility. Organic traffic metrics are losing ground to a growing phenomenon: zero-click searches.

A SparkToro study using data from Similarweb (2024) showed that for every 1,000 Google searches in the United States, only 360 clicks reach the web. Said in other words: most queries end up without visits to the sites.

Pew Research confirmed the same pattern: when an AI summary appears, users click on links in only 8% of visits, compared to 15% when an AI summary does not appear.

In this new context, what matters is not so much where your website appears on Google, but whether your brand appears in the response provided by AI.

What is a decision prompt and how to recognise it

A decision prompt is a question asked to an AI engine that seeks a clear recommendation, not just information. This is the moment when the user is close to making a purchase decision, comparing suppliers, or choosing an alternative.

Unlike an informational query (‘what is HR software’), a decision prompt asks the model for a specific judgement or suggestion. This is what in classic SEO was called a transactional search, but now expressed in the form of a conversation.

Key features of decision prompts:

• Comparative or superlative language: ‘best,’ ‘alternative,’ ‘top,’ ‘recommend.’

• Specific context: sector, country, company size, specific need.

• Action-oriented: the question is not seeking to learn, but to choose.

Practical examples by sector:

• SaaS B2B: “What’s the best payroll software for SMBs in Spain?”
  
  
  

• Turism: “What AI tools are best for planning a trip to Italy?”
  
  
  

• Retail: “What are the best sustainable clothing brands in Europe?”
  
  
  

• Education: “What LMS do you recommend for universities in Latin America?”

Detecting these prompts is key because they are the new battleground for visibility: if your brand does not appear in these types of responses, it simply does not enter into the user's decision-making process.

Furthermore, models tend to give short lists of 3 to 5 options, which are much more limited than the first page of Google with ten blue links. That means that each recommendation counts for much more.

AI crawlers and their impact on conversion

With fewer clicks to websites, AI models need to be fed in other ways. This is where AI crawlers come in: automated systems that crawl millions of pages to extract information and feed the models.

This radically changes the traditional cycle:

• Before: users searched on Google, visited several pages and compared.

• Now: AI does that research for them, consulting multiple sources in seconds and delivering a direct conclusion.

The result is that when a user receives an AI response, they are already at a more advanced stage of decision-making. They don't have to analyse 10 links or compare manually. That work has already been done by the model.

The impact on conversion is significant. According to data from Zeta Global, AI-powered searches can convert up to seven times more than traditional searches, precisely because the user arrives with greater clarity and less friction in the process.

This redefines the funnel: less traffic, but more qualified interaction.

How to identify and audit prompts step by step

Working with key prompts cannot be improvised. It requires a method for discovering, auditing, and prioritising.

1. Initial discovery

The first step is to gather the questions that your customers and prospects actually ask. Here's what you should do:

• Review sales and support conversations.

• Analyse forums, reviews and communities in the sector.

• Observe how questions are formulated in AI engines.

2. Auditing in AI models

The second step is to test these prompts on the main models: ChatGPT, Gemini, Perplexity, Google Overviews, etc. The aim is to see if your brand appears, how it appears, and in what context. For example:

• Is it included in the list of recommendations?

• Is it the first option or does it appear at the end?

• Is it mentioned with positive arguments or in a neutral way?

3. Prioritisation of prompts

The third step is to prioritise. Not all prompts have the same value. Ideally, you should start with 3–5 critical prompts, where query volume, commercial relevance and the gap with competitors converge.

Then you can scale up to dozens of prompts, segmented by product, market or use case.

How Omnia simplifies this process

Doing all this manually is slow and tedious. It requires testing each prompt on several models, saving screenshots, noting results, and repeating the process periodically. This does not mean that it is impossible to do manually, but it takes time and effort. 

At Omnia, we know customers who started doing it manually and, over time, managed to position themselves well in the responses. It should be noted that there were no specific tools for this work before, but now, there are.

This is how Omnia solves this problem:

• Topic Explorer generates dozens of relevant prompts in seconds, also showing their monthly frequency and level of competition.

• The system automatically audits responses in ChatGPT, Gemini, and Perplexity, by country and language.

• It offers a dashboard where you can see if your brand appears, in what position, and alongside which competitors.

This means that instead of spending hours on manual audits, a brand can have a complete map of its visibility in LLMs in a matter of minutes.

Example: an HR SaaS that did not appear in any model responses became the top recommendation in ChatGPT, simply by aligning its content with key prompts identified with Omnia.

Conclusion

The search model is evolving. Clicks are no longer the main reference point, and decision prompts are becoming the new playing field.

Identifying those prompts, auditing how models respond, and measuring your visibility there is the first step to winning in an environment where conversion occurs in the response, not in the visit to your website.

With Omnia, this process is no longer manual and complex. In minutes, you can discover the prompts that matter, measure your presence in them, and make strategic decisions to increase your visibility in the era of generative SEO.