All Collections
June AI
June AI explained
June AI explained

Accurately and immediately answer questions with June AI, June’s AI-powered SQL editor

Updated over a week ago

June AI uses GPT-4 to translate questions about your product into SQL queries that can be executed on our database.

Here's how you should think about it ⬇️

Think of June AI as your product analytics genie.

It does all the complicated stuff, but the cool part is you just ask your questions in plain English. No need to wrestle with setting up complex queries or mess around with infinite configurations.

It’s all about keeping things simple and easy, just the way we like it at June.

What you need to know

  • June AI lives in the context of a report: it's basically a super power of it. Just hit Ask AI from the report dock when you don't find the answer you needed in a report

  • To generate the query, June AI looks into your event names as well as user traits and group traits. So if you can, try to be specific and refer to them as much as you can. For more details, here's our event schema

How to ask questions to get 90% hit rate 🚀

  • Don't ask vague questions or use jargon.
    No questions like:
    - "why are my users churning?"
    - "who are my power users?"

  • Be as specific as you can: explicitly reference you event names and trait names
    Example: "What's the ratio between 'event_name1' and 'event_name2' for the last 30 days". For this reason, keep another tab of June opened with the events section in handy as well as the people section where you can see traits.

  • If you have a complex question break it down and iterate on the result.
    Example: instead of asking "What's the unique count of users and total count grouped by week of 'event_name' with the property 'property name = property value' for the last two months"
    Go for something like: "give me the unique count of users that triggered event_name in the last two months" and then "only look at the event 'event_name' where the property named 'property_name' has the value 'property_value' " and then "add a second column with the total count" and then "Group those by week for the last 2 months"

Other tips

  • DOs:

    • You can refer to your users as users, members, or seats

    • You can refer to companies as companies, accounts, organizations, or groups

    • You can ask for the visualisation you want in the prompt. For a: List > "List the people…", Number > "The total", "count", Line chart > "The evolution of..."

  • Once the result is satisfying, click on Pin to this report to save the insight at the bottom of the report!

  • If you're a SQL master you can tweak the SQL yourself but hitting the <> symbol

  • Make sure to rename the insight in a meaningful way so that everyone in your workspace can understand it 😉

  • ❗️AI generated insights do not take into account the audience selected in the report unless you specify it in the prompt ❗️

Did this answer your question?