Natural Language General Guidelines

Clarity in Querying

To ensure accurate results, queries should be structured with explicit detail. Instead of using general terms like "players from Spain" specify "players from the country Spain".

Similarly:

Ideal: "total NGR for the year 2023"

Not ideal: "total NGR for 2023"

This avoids misinterpretations and ensures more accurate results. While the interpretation will probably work anyway, it is always best to be explicit and not leave room for any assumptions.

Data Granularity

BetIntel.ai does not provide transactional-level data by default. The minimum level of data granularity available for the default datasets is daily aggregates. Queries attempting to retrieve data on a per-transaction basis will not return results or return daily aggregated results.

Keyword Interpretation

When interacting with the system, hover over the highlighted orange keywords to verify that they are correctly interpreted. This helps confirm that the intended data points are recognised as expected.

Hovering over orange-highlighted keywords (or tapping on them on mobile or tablet devices) will display a tooltip with the keyword's interpretation.

Show the top 10 performing btags by conversion rate in the past 7 days. Show conversion rate as a percentage to 2 decimal places. Include count of signups and count of ftds for the same period

Searching for String Patterns

For string searches, use like [text to search] to find text that closely matches the enclosed phrase, or [text to search] for an exact match. Text enclosed within [ ] is not interpreted as a keyword, ensuring it does not affect query processing. This helps maintain precise pattern matching within datasets.

The sample query below searches for the top 10 games with a category equal to "Slots", based on the Return-To-Player (RTP) value.

List the top 10 games with category [Slots] by RTP for the current year

The below sample, finds any games that closely matches the term "Dracula". The results would include all games with the following names:

• Dracula forever
• Dark dracula
• Dracula

"Show me all my brands and total wagers for casino game like [Dracula] for last year"

Structuring Date-Based Queries

To ensure proper query interpretation, date periods must always be placed after the subject matter.

• Correct: "Show active customers this month"
• Incorrect: "This month show active customers"
• Correct: "Show total bonuses for the year 2022"
• Incorrect: "For the year 2022, show total bonuses"

Dynamic Insights

When generating insights that update automatically over time, use relative date terms such as current month, current year, or last year instead of fixed dates. This ensures that saved insights with an automatic refresh rate remain dynamic, adjusting seamlessly as time progresses. As a result, both the insight itself and any workspaces referencing it will always reflect the latest data for the relevant period.

Examples:

• "Show total NGR and total wagers for last month" (Dynamic)
• "Show total NGR and total wagers for January 2025" (Static, requires manual updates)
• "Show active players this month that are inactive this month" (Dynamic)

Querying Across Vertical Data

For general queries across all verticals, simply omit the vertical name.

• Example: "Show total NGR for the past week group by day." (This returns NGR across all products.)

To retrieve specific vertical-based metrics, explicitly mention the vertical in the query.

• Example: "Show total casino NGR, total casino GGR for the past week group by day."

Using Keywords for Data Grouping

Group By

The group by keyword is used to categorise results by a specific attribute.

• Example: "List conversion rate for last month group by acquisition source."
• Example: "Show the average RTP for my game category [Slots] for the past month, group by week."
  • Grouping by week ensures the dataset averages RTP per week.

Distinct

The distinct keyword is used to retrieve unique attributes.

• Example: "Show me a list of distinct casino games."
  • This will return a unique list of games without duplicates.
• Example: "Show me a list of distinct acquisition sources"
  • This will return a unique list of acquisition sources without duplicates.

Optimising Saved Insights & Refresh Rates

When saving an Insight for repeated use, set an appropriate refresh rate to maintain system efficiency.

• Each account is allocated a predefined set of resources. Setting excessively low refresh rates for multiple insights can slow down overall performance.
• Saving insights minimises the need to re-enter queries and reduces additional query costs incurred when executing natural language queries.

✅ Suggested read: How to effectively write natural language queries.