Best Practice Guide

Passing Dates

Always pass a date when possible. Leaving out a date will cause the AI to guess. For accuracy, it is best to give it a date.

Poor Question: What were my sales in September?
Good Question: What were my sales in September 2024?
Why: Computers interpret time differently, which can cause the incorrect year to be used when answering questions.

Poor Question: How many first time customers did I have?
Good Question: How many first time customers did I have in September 2024.
Why: By leaving out the date, it might try to calculate based on lifetime of the account, last year or year to date. Resulting in different answers based on your conversation. To ensure you get the correct information, make sure to pass the date

More Examples:

Poor Question: What was the total sales for customers tagged “VIP”?
Good Question: What was the total sales for customers tagged “VIP” in November 2024?

Tell the AI if it should search based on the Title, SKU, or Tag.

Poor Question: What were my sales for shirts?
Good Question: What were my sales for products tagged “shirts”?
Why: The AI might try to search based on Title, SKU or Tag, giving back different results each time.

Poor Question: What were my sales for blue shirts?
Good Question: What were my sales for products with “Blue Shirts” in the title, for the last 90 days.
Why: Leaving out the date will get you different results each time you ask since the AI will choose what date range to use.

*IMPORTANT*

Searches are not case sensitive.

Searching for “Vip”, “viP”, or “VIP” will yield the same results.

Searches are based on similarity and not exact match.

Searching for products tagged “blue” will return products tagged “blue-shirts”, “blue shirts”, and “blue_shirts”.

Sales forecasting now supported.

You can ask to forecast your sales for the next “X” amount of days.

Only supports days from today going forward, e.g. Next 7 days, 30 days, 90 days.

Does not support future months, e.g. What will be my sales for February 2025?