Ask questions

Unlocking precious insights for your business is just one question away. Interacting with the advisor is as easy as asking a question to a colleague - a very knowledgeable one, in our case!

The advisor learns about your data the more you interact with it, so the best way to get the most out of Crystal is to experiment with how she learns from your data as you go.

In order to make sure the advisor correctly understands your questions, especially in the first period of use, make sure you formulate queries in a clear and concise way.

A good guideline to keep your question straight to the point. Bear in mind that, while the system is very flexible in keeping most parameters optional, mentioning the name of the topic or entity along with the objective is highly recommended.

If you enter a partial topic title, Crystal can recognize and respond to your request without requiring the complete title, making it more convenient for you. Then, you can drill down into the data more with several other optional filters.

However, to ensure a precise recognition of your question, you should follow these guidelines. Here is how to structure your questions for an optimal result:

1. Request, command, or greeting (optional) Start with a request, a command or, why not, a greeting!

  1. Can I see the …

  2. Show me the …

  3. Hey Crystal!

2. Name of Topic, entities, objectives (mandatory)

Next, specify either the name of the Topic you want to ask for, or the entities or the value aliases with the objectives which have been set up:

  1. Comparison between Sales and Leads …

  2. … Breakdown of Users by Country …

  3. Ranking of Sellers by Revenue …

  4. Average Revenues over time (by month)

3. Time range (optional)

After specifying the Topic you wish to ask for, you can add a time range to capture a specific moment in time of your data. Bear in mind that if you do not include a specific time interval in your request, your data will be displayed following a default time range set up by your admin, but if you’re interested in a specific period, you can add, for example:

  1. in the past six months …

  2. … for the last year …

  3. … in 2019 ...

4. Filter (optional)

And finally, if you want to further narrow down your results, again optionally, add a filter among the ones that your admin has configured and made available for your selected Topic:

  1. … for Spain

  2. … for smartphones

  3. … for the Main Office.

So, to recap:

Managing conversational complexity

Crystal’s mission is to simplify the interaction with your data and, let’s face it - most times data is complicated. For this reason, we added an additional layer of simplification to the conversation.

If your project contains entities, Filters or objectives which are too close between them, Crystal will automatically detect similarities and ask you to disambiguate between uncertain values through an easy-to-use selection component, to ensure you get exactly the answer you expect. Disambiguate will be provided both for when you ask a simple question and when you ask Crystal to set up an alert.

In order to further detail the conversation, applying Filters to Topics can help a lot.

For example, if the Topic is “Sales overview”, a Filter entity could be “Country”, which would retrieve the sales for a single country. To go further in detail, a Filter value is a specific item in the entity’s data column. For example in the above case, “Italy” would be a specific value for the entity, Country.

Filters can have different data types applied to them. In the case of a Filter with numerical data it can be difficult to apply a single value in the conversation. To address this, you can apply them to your Topic in a more natural way: for example, instead of having to specify the exact values, 100,78 you can ask for “between 100 and 101”. Numerical dynamic Filters can also be retrieved with comparisons (e.g. filter all values greater/lower than 100).

Additional terms understood by the Advisor

Since our app relies on advanced Natural Language Recognition to allow users to communicate with the Advisor in the most natural way possible, we have also included more advanced terms you can use in your interactions with Crystal. As we learn from users’ usage, the list will grow, allowing an ever-growing number of recognized terms to ensure an even smoother experience.

Bear in mind that these terms are strictly related to the objective of the Topic you are requesting (read more about objectives here) and that they can (optionally) be used as variations of the terms used in your questions to the Advisor.

We hope this article was helpful. Check out other tutorial articles to know more about Crystal!

If you have any questions about Crystal’s features, you have encountered a problem or you would like to share your feedback, contact us at support@igenius.ai.

Last updated