I’ve been involved in data warehousing for over ten years with hands-on experience using ETL tools such as SSIS, Azure Data Factory, Synapse Pipelines and Cognos Data Manager. However, I’ve not used Dynamics to ingest, manipulate and interrogate data, so I was curious to see what Dynamics 365 Customer Insights could offer.
My personal development plan at risual incorporates passing the Microsoft Certification Exam MB-260: Microsoft Customer Data Platform Specialty, which is a relatively new certification that came out of beta in Spring 2022. As I had no prior experience of the product, my strategy to pass the exam was combining a mixture of theory and actual product development:
a) Revise the Microsoft Learn Documentation (I went through this three times!) Exam MB-260: Microsoft Customer Data Platform Specialist – Certifications | Microsoft Learn
b) Hands-on development in the Dynamics Customer Insights 30-day Trial Customer Insights Tool Free Trial | Microsoft Dynamics 365
This is what I learnt about this product and how it helped me pass the exam.
What is Dynamics Customer Insights?
Dynamics 365 Customer Insights is a product within the Dynamics 365 suite that provides insights into customer profiles and tracks engagement activities. It can therefore potentially increase customer retention and enhance customer experiences.
The key selling point Customer Insights provides is a consolidated view of customers by centralising data from various business silos to provide a ‘single view’ of a customer. This happens by understanding customers’ digital touch points within an organisation (i.e. email, SMS, Website, Surveys, Social Media etc.) and capturing information about their activities to provide a clearer image of customer engagements. Microsoft refer to this as building a ‘Digital Twin’ as shown in the image below.
Data Ingestion
The data ingestion section is set-up using a source connector, with data then pulled into a Customer Insights data set. This could potentially be purchases coming from a PoS system or customer profile data coming from a CRM application. Customer Insights encompasses an impressive array of available data sets using Power Query. These include simple Text/CSV files, Synapse Analytics, Web APi or Azure HDInsights Spark as defined below:
Power Query is then used to clean up and transform the data as required. This revolves around choosing the correct columnar data type, removing bad rows, removing/renaming columns, etc. The data then gets refreshed and viewed from the Entities page in Customer Insights. The screenshot below shows eight entities ranging from eCommerce, Customer Loyality, PoS and Website sources.
Data Unification
This section allows mapping, matching and merging of customer data sources recently imported into the data sets. The helps an organization unlock insights, builds deeper understanding of customer traits and unifies customer transactional, behavioral, and observational data sources into a single 360-degree view of the customer.
- Map: Entities and source fields get combined to create a unified customer profile. As an example, it’s likely the customers’ full name, address, DOB and email address are candidates for their personal profile, whereas reward points would be taken from a separate transactional data set. Both of these sources would then get mapped together (using a customer id as a primary key) to create a unified customer profile.
- Deduplication: It’s rare that customer data doesn’t exist more than once in a source system. This causes issues with the unification process and data quality. Customer Insights has functionality that indicates the level of precision when determining if duplicate records exist (Low (30%), Medium (60%), High (80%), Exact (100%)). Depending on how strict the business rules are, the majority of duplicates get identified and merged into one customer record.
- Match: Matching is where we define rules for matching records between entities. An example would be adding two conditional rules against a surname and an email address from two different data sources. Again, we can use the level of precision to generate exact matches or be more lenient for fuzzy matches (i.e. if the email address was john.smith@universe.com in one source system, but johnny.smith@universe.com in another source system then that would be a matching record in a more lenient model (but would NOT match if the business rules were a 100% match).
- Unify: This stage reviews, edits, removess or ranks customer profile fields after the matching phase is completed.
Working with the Insights
We now can start to dig deeper into the insights and analyse the data that is related to an organisation.
- Define Activities: Activities help consolidate customer activities across data sources and put them into a timeline view. These activities might represent interactions or purchases.
- Define Relationships: Relationships connect entities together and generate a graph of customer data.
- Enrichment: Enrichments help tracking of brand affiliation and loyalty across hundreds of different brands and several interest categories.
- Define Measures: The KPIs that best reflect the performance and business health. These measures might represent satisfaction levels, revenue targets, or performance levels.
- Create Segments: Segments group customers based on demographic, transactional, or behavioral customer attributes. An example segment could be ‘All customers in London who have spent more than £1000 in a monthly period’.
We can now see a unified 360-degree digital customer view. Functionality for searching and drilling-down to get more details into specific customers is shown in the fake data below.
Predictions and AI
Customer Insights provides multiple ways for organizations to use AI to make data related predictions. This could be predicting the value of a missing field such as customer gender, or likelihood of a specific outcome. Customer Insights can also represent more advanced scenarios that use custom Microsoft Azure machine learning models from within Customer Insights.
Exporting Data
The Exports feature shares specific data like customer profiles, segments, entities, schemas, and mapping details with various external applications. Each export requires a connection setting up that manages authentication and access to the application being exported to. Dynamics 365 Marketing and Dynamics 365 Sales are examples of where Customer Insights data can be exported to take further actions on the generated insights. There is a plethora of target applications for exporting Customer Insights. These include Azure Data Lake Gen 2, Synapse, LinkedIn Ads, Dynamics 365 Marketing, Dynamics 365 Sales, Google Ads or Microsoft Advertising.
Final Thoughts
Dynamics 365 Customer Insights gets a thumbs up from me. The UI isn’t overly complex and source data ingestion is relatively simple. The ability to manipulate the data using Power Platform doesn’t require too much technical wizardry. The trickiest part is the Unification process (mapping, matching, merging). This doesn’t come as a massive shock as the most difficult part of any ETL process is the transformation process! Once the Unification process is complete, then creating the measures, customer segments and using the AI functionality reveals the true power of this product to harness a greater understanding of customer traits across your organisation.
Would you like to know more?
risual have deep technical knowledge of the Dynamics 365 suite and are currently building these platforms for multiple clients right now! Contact us on the link below if your organisation wants to take advantage of a modern cloud technology approach to solving data problems.
