An Early Look at Natural Language Queries for Dynamics 365 Marketing

Natural Language Queries for Segments

Natural Language Queries are now available as a new preview feature, to create audience segments in Microsoft Dynamics 365 Marketing.

These are designed to simplify the process for building segments by entering regular search expressions. Natural language processing combined with machine learning will seek to match these terms and phrases, to apply logic and attributes across common entities.

Let's take a look...

Firstly, this feature must be enabled by an administrator. In Marketing settings, navigate to Other Settings > Feature Switches > Segmentation to enable the Natural Language Queries feature.

Segmentation Setting

Once activated, the natural language input will be shown in a Natural View on the segment builder form when a new query block is added.

Using this, an expression can be quickly entered that will target the relevant audience profile. This has the potential to save time as users don't need to use drop down lists, to define each query line.

Segment NLQ

To test this we'll use an expression that references a single field:

"contacts anniversary in the next 30 days"

As shown below, this query has added two filter rows which reference the Anniversary field to define the 30-day term.

NLQ Example

Natural language queries can reference associated entities. In another example, we'll build a segment that will target contacts who are based in London working in finance. From this one line expression, the segment query is immediately populated. In addition to referencing the 'city' field on Contact entity / table, this involves hopping to the associated Account entity to reference the Industry field that will identify Contacts who work in the financial sector.

Contact and account query

As a preview model this can feel limited at times but it is undoubtedly improving through increased adoption and feedback, that will help to train the model and better understand each query context. After running a query, users are recommended to click the like or dislike icons to provide feedback on the result, to help the model improve the relevance of results in the future.

Each query block can be easily amended or expanded. If the natural language expression hasn't correctly applied any filter, this can be quickly corrected before the segment is applied.

Microsoft has yet to announce a timeline to release this feature for general availability during the 2020 Release Wave 2, so it may be rolled out during the next release wave from April 2021. We also look forward to seeing how similar natural language query capabilities will be extended to other Dynamics 365 apps and search features.

In the public preview, this feature currently only supports English language queries and is available in US, European and UK Microsoft Azure areas.

READ MORE: Natural Language Use in Segments

Never miss a post! Subscribe to our blog

Receive our emails about the latest Microsoft Dynamics 365 updates and events

Sign-Up Now