How to create a set of relevant questions to ask AI Search Engines

1. Click on AI Search…

2. Select a project from the dropdown

3. Click on '+' button to create a new Flow

You can have multiple flows per project. They are used for organizing different groups of questions for different types of analysis. For example, you might have:

• A group of questions that are unbranded (which is the default setting)

• A set of questions which are branded

• Questions about different categories or sub-categories

It is entirely up to you.

4. Enter a name for your flow, e.g. "Tesla" and complete the form.

5. (Optional) Enter a description and click 'Create Flow' to save.

The LLM will definitely know the entity 'Tesla', but for lesser-known brands, a description is helpful as it gives our AI more context to work with. e.g. I have typed "Electric car manufacturer and retailer".

6. Well done you have created your first 'Flow' and you should see the primary brand node on the canvas.

7. (Optional) To give you more room to work you can close the menus.

Click this icon to close the flow menu.

8. (Optional) Click the chevron to close the Authoritas main navigation on the left-hand side of the page

9. (Optional) You can also click and drag to widen or narrow the width of the sidebar

10. Now click on the brand node to focus on it

The right hand side panel will update to show you which nodes you can connect to it

11. Click on 'Product Category'

N/B: We need to give our LLM enough context to create a set of relevant questions. So, each flow must have a brand node, at least one product category, and each product category must have key factors for us to generate questions. 
​
Product Categories are like topics - they are a way of separating the analysis into relevant different segments. e.g. Menswear, Womenswear.

12. Give the Product Category a name

e.g. "Family Cars"

13. Give the Product Category a description

This helps the LLM understand exactly which niche you are researchinge.g. "Consumers looking to buy an electric car for their family"

14. (Optional) Add another Product Category

The dialog box stays open so you can add multiple product categories in one go. If you don't want to add any more product categories, you can click the X or cancel button. In this example, I've added a second e.g. "Performance Cars".
​
When you have added all your product categories, click 'Save'.

15. (Optional) You can now click and drag the highlighted elements around the canvas to organise the flow.

16. (Optional) You can use the zoom controls to zoom in and out or re-centre your flow.

17. Now all you need to do is add 'Key Factors' to each product category

So, click on your first Product Category Node. e.g. Family Cars.

18. The sidebar on the right-hand side will update to show you which types of nodes you can connect to a Product Category node

Click on 'Key Factor' - this will open a new dialogue box.
​
(Optional) You could at this stage add further subcategories to each product category and then revisit each subcategory and add key factors to them instead of product categories.

19. Click on 'GENERATE' to use AI to generate these 'Key Factors' for you

Key Factors are criteria which consumers find very important when considering product purchases or making decisions in this category.

Our LLM will generate up to five key factors for you for each product category or subcategory. 
​

Alternatively, you can manually add your own key factors instead of or in addition to the AI generated ones.

20. You need to generate key factors for every product category

So, repeat this process for all your other product categories. e.g. Click on 'Performance Cars'.

21. Click on 'Key Fatcors' again

22. Click on 'GENERATE'

23. Well done - you now have a complete 'Flow' and can now generate questions and then create a run

Once your flow has a brand node, product categories, and key factors, this is enough context for our LLM to generate questions. Optionally, you can customise this even more by adding buyer personas and/or buyer journey nodes, or completely custom prompts which help you modify the questions that are generated. 
​
Custom modifiers are useful for creating questions in different languages or specifying a specific location. 
​
Read more about these advanced node types here: https://help.authoritas.com/en/articles/11651143-how-to-get-started-with-brand-monitoring-in-ai-search-engines-authoritas-ai-search-platform
​
If you just want to generate questions now then skip the 'optional' steps below and move straight to the Question tab.

24. (Optional) If you want to customise this flow you can click on a brand or product node and then select a 'Custom Multiplier' from the right-hand side panel

Custom modifiers are like custom AI prompts — they create variations of each question for every Key Factor. Use them to explore different perspectives like personas, intents, or journey stages - e.g. region, language, or intent. 
​

25. You can customise your flow and the type of questions generated by adding one or more items here.

e.g. I want to get questions and answers that are relevant to UK consumers, so I've added 'United Kingdom' as a location. You can add more than one item at a time, so I could add mutliple different locations, e.g. Different countries or Towns or cities in the United States.
​
If you add more than one item, then it will create a set of questions based on your product categories and key factors for every location. So effectively, it will multiply the number of questions you generate by the number of locations. 
​
Another good use case here would be to use languages; you could generate a set of representative questions in multiple languages based on the same product category and key factors.

26. Add one or more items for this custom modifier.

e.g. "United Kingdom".

27. Add a description for the item.

(Recommended) - this is not essential, but you will get better results if you put in a description. 
​
e.g. "Consumers living and working in the United Kingdom. Questions MUST include "UK" in the text.
​
​Tip: Note the use of 'MUST' in caps - this is a good way of forcing an LLM to follow your instruction.

28. Click on SAVE

29. (Optional) Add Buyer Personas or Buyer Journeys

At this stage, you could can still click straight to the questions tab and generate questions. But you can also have the option of adding more custom nodes, or add a buyer personas or buyer journey node.
​
Click on a brand node or product category node. Normally, you would add this at brand node level, but we have catered for the fact that different categories might have different buyer journeys or different personas.

30. Click on Buyer Personas

Clicking on Buyer Personas let you generate question variations for different customer types, e.g. Students, First-Time Buyers, Value Seekers, Tech Enthusiasts, Parents, Professionals, etc.

31. Click 'GENERATE' to use AI to create relevant groups of buyers

This will multiply the number of key factors you have by the number of personas to give you a new expanded total number of questions. 
​
You can, of course, edit each buyer persona, delete them, or add your own.

32. (Optional) Add Buyer Journey node

Similary, you can generate Questions at different stages of the buyer journey funnel by adding a buyer journey node. Buyer Journey stages help you explore questions for each step in the customer journey, e.g. Awareness, Consideration, Decision.

33. Generate Questions: Click on Questions tab.

You are now ready to generate questions and it is as simple as clicking on the questions tab and clicking the generate button.

34. Click on 'GENERATE'

35. You will see a 'Generating Questions' message

You will see a progress bar as questions are generated. The number of questions generated is based on the number of key factors and multipliers you have added to your flow. It should take anywhere from 1 minute to 10 minutes if you've added a lot of nodes.

36. Review the generated questions

Once the questions have been generated, you can click the download button to download them to Excel. You could then edit and add to them in Excel and re-upload them if necessary.
​
You can also filter by tags to review the questions by product category or key factor. 
​
You can delete individual questions or delete all questions and start again.

37. Click on Run Analysis to send queries to AI Search engines

Once you are happy with your questions, simply click on the 'Run Analysis' tab to send your questions to the AI engines of your choice.

38. Click on 'CREATE RUN'

This is the best step to get started, but you can also schedule a run to run on a regular schedule, e.g. Hourly, daily, weekly, or monthly.

39. Search for the AI Search Engines/Models you want to analyse

40. Click into the search box and scroll down to select your desired model

41. For example, to select Google AI Overview tracking, just click on 'Google AIO'

42. (Some models only) Select any additional options

For some AI search engines additional options will appear. For Google and Bing you can do region selection, language selection, and even change the default safe search settings. We recommend leaving it with the default 'Safe Search' set to on.

43. Continue selecting the AI Search Engines you are interested

N/B: You must select a model that supports citations if you want to see the websites and pages that large language models are using to formulate their answers.

44. Click 'Create Run' to send your questions to your desired engines

You will see an estimation of how many Prompt Credits this run will consume. You will also see a progress bar showing you what percentage of your Prompt Credit allowance is remaining.
​
Please email support@authoritas.com if you are unsure as to how many prompt credits you have.

45. The run will now start and the questions will be sent to each search engine

You'll see a progress bar on the screen. This can take anywhere from 5 minutes to several hours depending on the number of questions generated. 
​
When the run has finished you can click on the 'Responses' and 'Sources' tabs to see your results.