Dynamics 365 Customer Insights uses AI in ways that are actually useful, not just marketing language. The platform applies machine learning to unified customer data to surface predictions and recommendations that business teams can act on without a data scientist in the room.
This article explains the specific AI capabilities built into Customer Insights, how they work in practice, and what CRM professionals need to understand to make the most of them.
Predictive Models Built Into the Platform
Customer Insights ships with pre-built AI models that require no machine learning expertise to run. You configure them with your data, run a training cycle, and get predictions you can use in segments and journeys.
Churn Prediction
The churn model scores every customer with a likelihood-to-churn percentage. It trains on your historical data — specifically the patterns in customer behaviour that preceded previous churns. Sales and customer success teams use this score to prioritise outreach before contracts lapse.
Customer Lifetime Value (CLV) Prediction
The CLV model estimates the total value a customer will generate over a defined future period. This is useful for tiering customers, allocating sales resources, and setting retention investment thresholds.
Product Recommendation
For organisations with product catalogues, the recommendation model suggests next-best products for each customer based on their purchase history and behavioural patterns. The output can feed directly into personalised email journeys or website experiences.
Need Microsoft Dynamics 365 talent?
We place certified Dynamics 365 professionals across the US and Canada.
AI-Generated Segments
Building segments in most marketing platforms requires you to know what you are looking for. AI-generated segments in Customer Insights work differently. You describe an outcome — such as "customers likely to convert in the next 30 days" — and the AI identifies the defining characteristics automatically.
The platform also offers suggested segments based on anomalies it detects in your data. If a subset of customers starts showing unusual behaviour — increased engagement, sudden drop-off, repeat contact — Customer Insights surfaces it as a segment worth investigating.
For CRM professionals, this changes the workflow from "build a segment and hope it is right" to "review what the AI found and decide what to do with it." That is a meaningful shift in how time is spent.
Copilot for Customer Insights
Copilot brings natural language interaction to the platform. Users can type a question like "which customers in the northeast have not purchased in 90 days but visited the website last week?" and get a segment or answer without writing filters manually.
Copilot also generates summaries of customer profiles. When a sales rep opens a customer record, Copilot can produce a plain-language summary of recent activity, predicted churn risk, and lifetime value score. This saves time and surfaces information that might otherwise get missed in a dense data record.
As of 2026, Copilot capabilities in Customer Insights have expanded significantly through Microsoft's broader Copilot rollout across the Dynamics 365 platform.
Need help with staffing?
We place CRM and Marketing Automation professionals across the US and Canada.
What AI Cannot Do for You
The AI features in Customer Insights are only as good as the data they train on. A churn model trained on six months of incomplete data will produce unreliable scores. A product recommendation model with poorly mapped purchase history will make poor recommendations.
AI also cannot replace the judgment that comes from knowing your business. A score that says a customer has a 70% churn probability does not tell you why. That context — a price increase, a competitor's offer, a bad support experience — still comes from people who know the account.
Use the AI outputs as inputs to decisions, not as the decisions themselves. The best teams use Customer Insights predictions to focus attention, then apply human judgment to determine the right action.
Getting Ready to Use AI Features
The most common barrier to using AI features in Customer Insights is data quality. The predictive models need enough historical data to train on, and that data needs to be clean and consistently structured.
Before activating AI features, audit the data you plan to use: How complete are your customer records? How far back does reliable purchase or interaction history go? Are your data sources correctly mapped and deduplicated in the unified profile?
A good data foundation does not happen automatically. It is built intentionally, often with the help of a Customer Insights specialist who knows how to configure the data unification layer correctly. Getting this right at the start saves significant rework later.
