Quick search…
Tutorials
Data integration
Tutorials and guides
Writing good descriptions
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”
Tutorials and guides
Writing good descriptions
Having good names and descriptions for your data is crucial - the AI uses these to find the best source of data to answer questions. Better names and descriptions will lead to better accuracy and clarity for both the AI and your colleagues. We generate the first version for you, using AI, to provide a foundation, but your insights and expertise can help the system understand all those little nuances in your data.
Note - changing names and descriptions within Telescope Labs has NO impact on the underlying data, it just improves the AI.
Here are some tips on writing the best names and descriptions:
1) Use full words
Clarify any technical abbreviations with their full word counterpart. This will help the system understand the meaning in the data. The AI will still be able to interpret the abbreviated form in the question, but for added accuracy, add it in parenthesis, or into the description. Be specific.
e.g. cust_tx could be Customer Transactions.
2) Disambiguate between fields
In case of overlap or similar values, make sure to clearly clarify the differences in the names and descriptions.
e.g. Customer Revenue1, Customer Revenue2 could be Customer Revenue, Customer Revenue (including Refunds)
3) Add information about the calculation (metrics)
Add information about how the metric is calculated, and any important nuances. This helps the AI have a deeper insight of the workings of the field as well as more accurately apply filtering logic down the line.
e.g. “YTD Revenue is calculated using the xyz_column, and excludes revenue from region C”
4) Add flavor and context in descriptions
Descriptions are a great place to add details about how to use the data - both intrinsically (how it should be used) and extrinsically (how it applies to your business) - and which for analyses analyses it is particularly relevant for
e.g. “Day 1 retention is our most important retention metric…”
5) Incorporate business-specific language
Descriptions are also a great place to define any special internal terminology you may use internally, to help the LLM key into your organization’s language.
e.g. “Rex’s metric” could be “DAU/WAU” with a description of: “Also known as Rex’s metric”