Data Availability in Luma AI

Luma has access to much of the same data you are already used to working with in Decile. However, Luma does not have access to all data available in other parts of the app, or may present slightly differently. If you are ever unclear about what a metric or dimension represents in Luma, you can always ask for a definition.

Supported Dimensions

Luma can answer questions using a wide variety of attributes related to customers, orders, and product transactions. While this isn’t a full data dictionary, these examples illustrate the types of fields Luma currently understands:

Customer-Level Dimensions

• Location (city, state, country, ZIP)
• Acquisition date or cohort
• Subscription status
• Customer tags
• Common enrichment attributes such as:
  • Age and age ranges
  • Gender
  • Household income ranges
  • Marital status
  • Home ownership
  • Education level
  • Presence of children
  • DMA or population density groupings (urbanicity)

Order-Level Dimensions

• Order date, created date, canceled date, updated date
• Order tags
• Shipping location
• Subscription order flags
• UTM marketing dimensions (source, medium, campaign, term, content)
• Discount codes
• Order line item details

Transactional Product Dimensions

• Product titles and categories
• SKU and variant information (title at purchase, vendor)
• Product tags
• Quantity and pricing details

These fields allow Luma to answer questions about who is buying, what they’re buying, when they’re buying, how they were acquired, and how products perform over time.

Supported Metrics

Luma can calculate a wide range of standard ecommerce and performance metrics, including:

• Revenue, net sales, gross sales
• AOV (Average Order Value)
• Order count
• Units purchased and units per transaction
• Discount amounts
• Refunded quantities and amounts
• Customer lifetime revenue and order frequency
• ...and more

These metrics can be aggregated, segmented, trended, or compared based on your prompt.

What’s Not Yet Available

While using Luma during the preview period, please note a few types of data are not yet supported or may be partially limited:

• Custom product or customer attributes that have been enabled elsewhere in Decile are not yet available to Luma
• Certain Decile-specific fields, such as CAC, Personas, or the full Acxiom enrichment catalog
• Decile Audiences
• Ad platforms and other non-ecommerce data source integrations

As data coverage expands, Luma will be able to answer more questions with increasing depth and flexibility.

However, there is some Luma functionality that may help you work around these limitations:

• Luma can dynamically search and group product dimensions, without requiring additional product hierarchy fields.
  • Example: A level of your custom product hierarchy includes a categorization called "Pants" that groups all products with 'pants' in the title. While that class attribute is not currently available in Luma, you can still ask a question such as "How many orders of pants did we have last month?" and Luma can dynamically identify all those product titles including pants.
• Luma can create rich personas, segments, and lists based on a description or analysis
  • Example: Ask Luma "What is the profile of my high value customers?"
• You can export customer lists from Luma and re-import to Decile to use as audiences and sync with outbound destinations

How Data Availability Affects Your Questions

In most cases, Luma will simply answer your question using the fields and metrics it already knows.

If you reference a dimension or field that isn’t yet supported, one of the following may happen:

• Luma may interpret your question using the closest applicable fields
• Luma may simplify the response or return a higher-level insight
• Luma may indicate that the data is not available
• Luma may request clarification

This behavior is expected during preview and helps ensure Luma stays accurate.

Tips for Better Results

Asking strong questions becomes easier once you understand what Luma can see:

• Use timeframes (e.g., “last 90 days,” “2024,” “month-over-month”)
• Reference customer state or lifecycle (e.g., “new vs. returning,” “subscribers only”)
• Segment by product attributes
• Apply question context if relevant (e.g., "We're trying to decrease the time it takes new customers to come back and make a second purchase")

If you know a field isn’t supported, ask the question more broadly or remove that specific dimension. Not sure if a field is supported or how to correctly identify it? You can ask Luma that too!

• Example: You know you have a trade program tracked through customer tags, but don't know or can't remember how those tags have changed over time. You can ask Luma "What customer tags do I have that are related to our trade, wholesale, or artist program?" and Luma can provide a list of tags for you to better ask your question.

Questions or Feedback

If something appears missing or inconsistent, or if you would like to request a field, contact your Decile Customer Success Manager. Your feedback directly helps improve Luma’s data coverage.