Validate ASK BOSCO® AI Analyst results
Verify with follow up questions
How to frame your questions
Basic Questions
One of the approaches for asking questions that will help you get to the expected answer faster is to think of the question in parts.
- What computation do you want ASK BOSCO® AI Analyst to perform? Try to write it in a way that anyone will understand.
- What groupings do you want to see in the final results?
- What filters do you want to apply while generating the answer?
A good question clearly defines all of these aspects in an unambiguous manner. Note that since ASK BOSCO® AI Analyst is a conversational experience, you can ask your questions incrementally.
Complex questions
For deeper analytical insights, you can leverage AI Analyst's advanced computational capabilities. This allows you to move beyond simple aggregations and ask questions that require more complex logic. For example:
-
Calculate growth (percentage change) of measures over time, such as year-over-year or month-over-month.
-
Perform arithmetic operations to derive ratios and percentages, like the percentage of total.
-
Analyze product sales to measure performance relative to the total product inventory.
-
Embed a query within another query, known as a subquery, to filter based on dynamically computed values (for example, "top 5 product categories").
-
Apply conditional logic (if-then-else) to create custom categories or perform what-if analysis.
| Capability | Question | Description |
|---|---|---|
|
Growth |
Show the growth of sales year over year. |
Tracks annual sales growth rates over consecutive years, providing clear visibility into long-term business trajectory. |
|
Growth |
What is the growth of sales weekly for PPC for the last 8 weeks? |
Provides detailed week-over-week growth analysis for PPC across a 2-month window. |
|
Ratios/Percentages |
What is the percentage share of sales for each product? |
Calculates each product’s contribution to overall sales, providing a proportional view of your portfolio performance. |
|
Market share |
What is the market share of sales for trousers by region. |
Calculates the percentage of total sales that comes from trousers for each region. |
|
Subqueries |
What is the total sales of jackets only from the top 5 cities by sales? |
Performs targeted analysis by first identifying your highest-performing cities, then analyzing jacket sales within those markets. |
|
Subqueries |
Show campaigns that did not have any sales in the last 3 months. |
Identifies potentially under-performing campaigns by highlighting those with no recent sales activity. |
|
Conditional logic |
Show stores that had sales greater than $1000 yesterday and more than $500 today. |
Identifies consistently high-performing retail locations by applying multiple time-based sales thresholds. |
|
Conditional logic |
Show total sales by channel for last month. Highlight channels as top performing… |
Calculates total sales per channel, and then categorizes each channel based on its sales using a conditional formula. |
Why Questions
ASK BOSCO® AI Analyst can now answer "Why" questions.
How to work with "Why" questions Is explained in this article. Read on for more insights.
Mathematical computations
ASK BOSCO® AI Analyst can:
- Count things in your dataset (for example, how many campaigns are there?)
- Aggregate metrics (for example, what is the average sales?)
- Perform complex mathematical computations (for example, How does the ctr for the channel source shopping compare to performance max?)
Grouping
ASK BOSCO® AI Analyst can:
- Help you perform the computations for specific groups (for example, what is the total sales by channel? )
- Create grouping on date columns on the fly (for example, what are the yearly sales? or what are the monthly sales?)
Filtering
ASK BOSCO® AI Analyst can:
- Perform computations and grouping operations on a filtered set of data instead of the full table (for example, what are the total sales for August? or what are the sales for this month? or what are the sales for SEO in July 2024?)
Example questions
Below is a quick overview of the kinds of questions you can ask by a combination of above skills. Note that this is not an exhaustive list.
Simple Computations
|
Sub-category |
Sample questions |
|
Counting |
|
|
Aggregation |
|
|
Conditional counting / aggregations |
|
Common Date Intents
|
Sub-category |
Sample questions |
|
Group by date |
|
|
Filter by date |
|
Advanced Computations
|
Sub-category |
Sample questions |
|
Ratios and percentage |
|
|
Comparison |
|
|
Advanced calculations (change, nested aggregations, conditional logic, etc.) |
|
Further types of questions in ASK BOSCO® AI Analyst?
ASK BOSCO® AI Analyst will support a few additional types of questions. Note this is not an exhaustive list.
Descriptive questions about selected data source
|
Sub-category |
Sample questions |
|
Descriptions |
|
|
Explanations |
|
Make recommendations for leveraging selected data source
|
Sub-category |
Sample questions |
|
Follow-up questions |
|
|
High-level str |
Validate ASK BOSCO® AI Analyst results
You can verify ASK BOSCO® AI Analyst's results by checking the query and by comparing to the previous step in the conversation.
Let's say you ask ASK BOSCO® AI Analyst, What are the monthly sales for channel/source Shopping in Google last year?
The answer is shown:
ASK BOSCO® AI Analyst has performed the following actions :
- Mathematical computation: The mathematical computations are highlighted in green in the query syntax. In the example above, we can see that the revenue column is marked in green, which shows that we have computed the total revenue.
- Groupings: The groupings are highlighted in blue in the search query. In the above example, we can see that monthly grouping is one of the groupings being used. The column label also shows that the Monthly date column is used for this monthly grouping.
- Filters: The filters are represented in grey. The query also shows that the system has applied a filter on source=’google’, date=’last year’, and Channel/source=’shopping’.
We recommend that you always check the search data query, as it provides an easy way to validate all the important choices of the column values picked by ASK BOSCO® and how it is using them to build the answer.
Verify the changes after follow-up questions
ASK BOSCO® AI Analyst considers each new question in the conversation as a follow-up to the previous question. As your answer becomes more and more complex in the conversation or when you start the ASK BOSCO® AI Analyst experience from a dashboard, you don’t have to continuously validate the query for each answer generated by the system, but you can easily verify.
In the case of follow-up questions, we provide a quick summary before the answer which showcases what has changed from the previous step. For example, if we tell ASK BOSCO® AI Analyst to “include Performance Max for last 18 months” for the question asked in the previous step. ASK BOSCO® AI Analyst shows that it added Channel/Source = 'Performance Max' and changed Date = 'last 18 months'.
