How to choose which prompts to analyze?

BotRank.ai is designed to help you analyze the visibility of an entity (company, brand, product, person, etc.) in the answers of AI engines. A crucial starting point is knowing which prompts to analyze.

A prompt is the question or query you submit to the AI. Choosing the right prompts is essential to evaluate and improve your entity's visibility in the generated responses.

This article presents a methodology with five technical approaches to identify the most strategic prompts to track on BotRank.ai, allowing you to build a relevant analysis strategy for your AI visibility.

Aligning GEO strategy with SEO strategy

Your visibility strategy on traditional search engines (SEO) is a goldmine for your strategy on AI engines (GEO). The keywords you are already trying to rank for on Google are powerful indicators of your audience's search intent. It is therefore logical to use them as a foundation for defining your prompts.

How to do it? Transform your SEO keywords into conversational questions. If one of your strategic keywords is "SME accounting software", you can break it down into a series of prompts for BotRank.ai:

•"What is the best accounting software for an SME?"

•"How to choose an accounting software for my small business?"

•"What are the essential features of an accounting software?"

•"Are there free accounting software for SMEs?"

This approach ensures that you monitor queries directly related to the topics for which you have already invested resources and developed expertise.

Getting inspiration from “People Also Ask” (PAA)

The "People Also Ask" (PAA) section of Google is an invaluable source for understanding the questions that users are actually asking around your strategic keywords. These suggestions are formulated in a natural and conversational way, which makes them very close to the prompts used on generative AIs.

How to do it? Manual search: type your main keywords into Google and analyze the PAA boxes that appear. Explore the questions and note the ones that are most relevant to your business.

Or with the use of tools: tools like AlsoAsked or AnswerThePublic automate this process. They allow you to enter a keyword and visualize a tree of associated PAA questions, giving you an overview of your audience's questions.

By integrating these questions directly as prompts in BotRank.ai, you ensure that you cover the search intentions validated by Google itself.

Using Google Search Console (GSC)

Google Search Console is your most reliable data source, as it shows you the exact queries that users have typed to find your site. By using a regular expression (RegEx), you can easily filter and identify queries formulated as questions that are already generating traffic to your site.

How to do it? Identify the questions with a filter:

1.Go to your performance report in Google Search Console.

2.Click on + NEW to add a query filter.

3.Select Custom query (regular expression)

4.Enter the following regular expression to filter the questions:

```regex^(how|what|which|why|when|where|who|is it that|can I)```

‍

This filter will display all queries that start with an interrogative word and that have generated impressions or clicks for your site. These queries are high-value prompts because they represent the actual formulations already used by your audience to find you.

Analyzing data from Semrush, Ubersuggest, etc.

If you use third-party SEO tools like Semrush, Ahrefs, or Ubersuggest, you have powerful features for searching for keywords in the form of questions.

Semrush's "Keyword Magic Tool", for example, has a "Questions" filter that instantly isolates all interrogative variations of a basic keyword. These tools provide you not only with the questions but also with valuable metrics such as search volume and keyword difficulty, helping you to prioritize the most important prompts to track.

As a reminder of the key features per tool:

•Semrush: Keyword Magic Tool ("Questions" filter) - Provides volume, trend, and difficulty for each question.

•Ahrefs: Keywords Explorer ("Questions" report) - Identifies questions asked on more than 10 search engines.

•Ubersuggest: Keyword Ideas ("Questions" tab) - Offers a long list of questions with volume data.

Practicing “Query Fan-Out”

The "Query Fan-Out" is a technique used by the AI engines themselves. They break down a complex query into several sub-questions to compile a complete answer. You can adopt a similar approach to enrich your list of prompts: list all possible variations of the same question.

How to do it? Use an AI engine like ChatGPT to do this work for you. Give it one of your strategic questions and ask it to decline it.

Example of a prompt for ChatGPT:

I am looking to analyze my visibility on AI answer engines. Here is a central question for my business: *'What is the best CRM for a real estate agency?'*. Act as an expert in SEO and GEO, and generate 20 variations of this question. Include synonyms ('software', 'platform'), different angles (for beginners, for large agencies), varied intentions (comparison, price, features), and different levels of specificity.

This method allows you to quickly generate an exhaustive list of prompts, covering a wide range of search intentions and ensuring that you don't miss any visibility opportunities.

In conclusion, defining the right prompts to analyze is the first fundamental step to succeeding in your Generative Engine Optimization strategy. By combining these five approaches, you will create a complete and relevant list of prompts, aligned with your business strategy and the real needs of your audience.

‍