MB-260, Microsoft Dynamics 365 Customer Insights (Data) Specialist, retired on November 30, 2024. That means the exam is no longer a certification target, but the product area it measured is still active. Dynamics 365 Customer Insights – Data continues to receive product investment in 2026, and the underlying work\u2014ingesting customer data, unifying profiles, building segments and measures, enriching data, applying predictions, and governing customer information\u2014remains relevant for organizations using Microsoft\u2019s customer-data platform.<\/p>\n
The retired Microsoft MB-260<\/a> blueprint is therefore best treated as a historical skills map rather than an exam-preparation checklist. Microsoft\u2019s final 2024 version gave the largest weight to creating customer profiles through data unification, followed by measures and segments, ingestion, AI predictions, third-party connections, and administration. Those areas still provide a useful framework for understanding the product even though candidates can no longer earn the old specialty credential.<\/p>\n There is no reason to keep studying MB-260 as though an exam date is approaching. The better use of old material is to identify the durable Customer Insights – Data skills that still matter in current implementations and then update the product-specific details against the 2026 service.<\/p>\n Microsoft\u2019s 2026 release planning continues to include Dynamics 365 Customer Insights – Data. The product remains the data side of the combined Dynamics 365 Customer Insights offering, while Customer Insights – Journeys handles real-time customer engagement and journey orchestration.<\/p>\n This distinction is important because the retirement of MB-260 did not mean Microsoft retired customer-data unification. Organizations still need to combine records from multiple systems, resolve identities, build customer profiles, create segments, calculate measures, and activate data in downstream systems.<\/p>\n What changed is the certification portfolio around that work, not the business problem itself.<\/p>\n A useful customer profile begins with reliable source data. Customer Insights – Data can work with data from Microsoft and non-Microsoft systems, but the engineering questions are familiar: which source is authoritative, how often does it refresh, what schema does it use, and what happens when records are missing or malformed?<\/p>\n MB-260 expected candidates to understand Power Query, Dataverse, Azure Data Lake Storage, Azure Data Factory pipelines, and other source-integration concepts. Those skills remain useful because Customer Insights implementations still depend on data arriving consistently and in a form that can be unified.<\/p>\n Teams should avoid treating ingestion as a one-time import. Source ownership, refresh behavior, schema changes, privacy constraints, and operational monitoring all need to be designed from the beginning.<\/p>\n Customer data is rarely clean enough to unify without preparation. Names vary, addresses are incomplete, identifiers are inconsistent, duplicate records appear, and values may use different formats across systems.<\/p>\n Power Query and other transformation processes can standardize data before unification. This is where field mapping, data types, null handling, normalization, and source-specific cleanup matter.<\/p>\n A customer-data platform cannot create trustworthy insight from systematically poor inputs. If two source systems disagree about the same customer and there is no clear business rule for resolving the difference, the technical platform cannot invent the correct answer.<\/p>\n The final MB-260 blueprint devoted 35\u201340% of the exam to creating customer profiles through data unification. That weighting made sense because identity resolution is the central problem the product is designed to solve.<\/p>\n Organizations may have the same person represented differently in ecommerce, CRM, support, event, loyalty, and billing systems. Unification attempts to determine which records belong to the same real-world person or organization and then create a usable consolidated profile.<\/p>\n Good matching requires careful selection of source fields, normalization, deduplication, match rules, and confidence in the identifiers being used. A loose rule can merge unrelated people; an overly strict rule can leave one customer fragmented across several profiles.<\/p>\n There is no universally correct matching strategy. Email may be a strong identifier in one business and unreliable in another. Phone numbers can be shared. Names can change. Household accounts may intentionally contain several people.<\/p>\n The implementation team should decide what constitutes the same customer and test the rule against representative edge cases. Review should include false merges as well as missed matches.<\/p>\n The durable skill from MB-260 is therefore not memorizing a wizard. It is understanding how data quality and matching decisions affect the customer profile the business will trust.<\/p>\n Once profiles are reliable, organizations can calculate measures such as lifetime value, average order value, frequency of interaction, service history, engagement level, or other business-specific metrics.<\/p>\n A useful measure needs a clear definition. \u201cHigh-value customer\u201d is meaningless until the organization defines the period, revenue basis, exclusions, currency treatment, and other rules behind it.<\/p>\n Measures should be designed so downstream teams can explain them. If marketing, sales, and service teams use the same metric but calculate it differently, a unified customer platform has not solved the governance problem.<\/p>\n Segments group customers according to attributes, behavior, measures, or other criteria. Their purpose is usually activation: selecting customers for engagement, analysis, service treatment, or another business process.<\/p>\n A segment should have an owner and a clear reason to exist. Creating dozens of overlapping audiences without lifecycle or naming standards makes the environment harder to operate.<\/p>\n Customer Insights – Data skills also connect naturally to Customer Insights – Journeys skills<\/a>, because unified customer data and segments often become inputs to engagement and journey scenarios even though the old certification paths were separate.<\/p>\n MB-260 included built-in predictions and integration with machine-learning models. The important concept remains valid: prediction should solve a defined business problem rather than exist because an AI feature is available.<\/p>\n Examples might include churn propensity, customer value, product interest, or another measurable outcome. The organization needs sufficient historical data, an appropriate target variable, and a way to evaluate whether the prediction is useful enough to affect decisions.<\/p>\n Responsible AI also matters. A prediction can be statistically accurate while still being inappropriate if the training data introduces bias or the organization uses the output for a decision that requires stronger explanation or human review.<\/p>\n Customer profiles can be enriched with additional information where supported. Enrichment can add useful context, but it should not be used to hide weaknesses in source data.<\/p>\n Teams should understand where enrichment data comes from, how often it changes, which profiles it applies to, and whether the organization is permitted to use it for the intended purpose.<\/p>\n That governance question is increasingly important as organizations combine first-party customer information with external data and AI-assisted insight.<\/p>\n The final MB-260 audience profile explicitly included privacy, compliance, consent, security, and responsible AI. Those are not side topics.<\/p>\n A customer-data platform can centralize highly sensitive information about individuals and their behavior. Access should be limited according to job responsibilities, and data retention should follow business and regulatory requirements.<\/p>\nCustomer Insights – Data is still an active product in 2026<\/h2>\n
Data ingestion remains the first engineering challenge<\/h2>\n
Data preparation determines the quality of every downstream result<\/h2>\n
Identity resolution remains the core Customer Insights – Data skill<\/h2>\n
Unification rules should reflect business reality<\/h2>\n
Measures turn unified data into business signals<\/h2>\n
Segments should represent actionable audiences<\/h2>\n
Predictions are useful only when the outcome is defined<\/h2>\n
Enrichment can improve context without replacing first-party data quality<\/h2>\n
Privacy, consent, and security are part of the data model<\/h2>\n