Microsoft Purview Pricing: Two Bills, One Platform, No Clear Map
When Microsoft rebranded Azure Purview to Microsoft Purview in late 2022, it folded two different products into one name: the Azure-hosted data catalog and governance plane (formerly Azure Purview) and the M365-hosted compliance suite (formerly Microsoft 365 Compliance Center). They share a portal at purview.microsoft.com. They do not share a billing model. Customers who come asking "how much does Purview cost?" are usually asking about one and accidentally buying the other.
This note explains both bills, where each surprises you, and how to build a credible estimate before you commit.
The Split That the Marketing Ignores
The product has two sides with genuinely different cost structures.
Side A: Data Governance (the old Azure Purview). This runs in your Azure subscription. It scans data sources, builds a Data Map, classifies assets, and surfaces lineage. You pay Azure invoices, metered on Capacity Units and scan compute.
Side B: Compliance (the old M365 Compliance Center). This covers Information Protection, Data Loss Prevention, eDiscovery, Communication Compliance, Insider Risk Management, and Audit. It runs inside Microsoft 365 and is controlled by your M365 licensing tier, not your Azure subscription.
Conflating them is where most scoping calls go wrong. A CISO asking about "Purview for DLP" wants Side B. A data engineering team asking about "Purview for data catalog" wants Side A. Budget them separately from the start.
Data Governance Pricing: Capacity Units Do the Work
The Data Governance side meters two things: the Data Map (always-on, per Capacity Unit) and scan compute (per scan vCore-hour).
A Capacity Unit is roughly 25 assets under management or 10 GB of metadata storage, Microsoft documents the exact conversion, and it changes when the asset count model is in Elastic mode. The standard Data Map runs at a flat hourly rate per CU. Elastic Data Map auto-scales between 1 and 16 CUs and charges only for what it uses in a given hour. For greenfield deployments with unpredictable asset counts, Elastic is almost always cheaper to start.
Scan compute is a separate meter. Every time you run a scheduled or manual scan against a source (ADLS Gen2, SQL, Fabric, Databricks, S3), you pay per vCore-hour of scan time. A full scan of a large ADLS container with hundreds of folders and tens of thousands of files can run for hours. We have seen scan costs exceed Data Map costs in the first month for orgs that turned on daily full scans across all registered sources without thinking about incremental scanning.
The concrete mistake: registering 30 sources, leaving each on a daily full-scan schedule, and never enabling incremental scanning. That multiplies scan hours by the asset count on every run. Switch to incremental after the first full scan; the delta scans complete in a fraction of the time.
Compliance Pricing: Start With Your M365 License
Side B is not an Azure meter. It is unlocked by your M365 or Office 365 SKU, and the step between E3 and E5 is the one that matters.
E3 gives you: sensitivity labels (manual only), basic DLP for Exchange and SharePoint, Standard Audit (90-day log retention), and eDiscovery Basic. If your requirements stop there, you pay nothing extra beyond your per-user M365 seat cost.
E5 adds: auto-labeling, advanced DLP across all workloads, Premium Audit (1-year retention, more event types), eDiscovery Advanced (with review sets and analytics), Communication Compliance, and Insider Risk Management. E5 runs roughly $25-$40 per user per month above E3 pricing, though Microsoft's current list prices and your EA discount will move that number. If you are not buying full E5, Microsoft sells compliance add-ons (the "E5 Compliance" add-on is the common one) at a lower per-user cost than the full E5 bundle.
Communication Compliance specifically, the feature that monitors Teams, Exchange, and Viva Engage traffic for policy violations, requires E5 or the E5 Compliance add-on. It does not exist in E3. This catches teams off guard when their Sentinel SIEM integration surfaces an insider threat use case and compliance realizes they have no monitoring policy to back it.
Where the Actual Surprises Live
A few cost drivers that do not appear in the summary pricing pages:
Scan frequency on large sources. A 100TB ADLS account with nested folder structures scanned daily at default parallelism will run for several hours per scan. Multiply by 4 vCores default parallelism and you have the math. Set scan trigger to weekly for bulk cold storage, daily only for active hot zones.
Elastic Data Map spikes during initial bulk import. When you register all your sources and run discovery for the first time, the Elastic map scales up to handle classification load. That first week can cost 3-5x what steady-state looks like. It is not a billing error; it is the one-time classification pass.
eDiscovery holds storage. Advanced eDiscovery stores content sets and review sets in Microsoft-managed storage. Long-running litigation matters accumulate GB of indexed content that you pay for until the case is closed and data purged. Large legal departments with many open matters have found $8,000-$15,000/month surprise line items here.
Third-party connector ingestion. If you pipe HR data, Slack exports, or Bloomberg data into Purview for Insider Risk or Communication Compliance, those connectors meter on data volume ingested. The published rate is low per GB, but high-volume connectors add up.
Estimating Before You Commit
Build a cost model in a spreadsheet before provisioning. The inputs are: asset count (from existing catalogs or rough data estate mapping), source count, scan frequency plan, user count by license tier, and any third-party connectors.
For the Azure side, query your existing Azure Cost Management to see if you already have a Purview account running:
// Azure Cost Management workspace query, isolate Purview meters
// Run this in Cost Management > Cost Analysis > Open in Query Editor
AzureCosts
| where ServiceName has "Microsoft.Purview"
or MeterCategory has "Azure Purview"
| summarize
TotalCostUSD = sum(PreTaxCost),
HoursBilled = sum(Quantity)
by bin(UsageDate, 1d), MeterSubCategory, ResourceId
| order by UsageDate desc, TotalCostUSD desc
If the table comes back empty and you know you have a Purview account, check that the account was created under the current subscription scope, it is easy to create a Purview account in a dev subscription and forget about it.
For a net-new greenfield estimate: start with 1 Elastic CU for the Data Map (you can always scale), plan incremental scans for all sources after the initial full pass, and model E5 Compliance only for users in roles that actually need Communication Compliance or Insider Risk monitoring (security analysts, compliance officers) rather than blanket-licensing the org.
What a Realistic Mid-Market Deployment Runs
These are directional ranges, not quotes. Verify current list prices against the Azure pricing calculator and your EA terms.
A 500-user financial services firm with 15 registered data sources (ADLS, SQL MI, Fabric, Databricks), weekly incremental scans, and E5 Compliance add-on for 50 compliance and security staff: expect $2,500-$5,000/month in Azure Purview Data Governance costs (heavily scan-dependent) and $8,000-$12,000/month in M365 add-on licensing for those 50 seats, depending on volume discounts.
A 5,000-user enterprise with 80 sources, daily incremental scans, full E5 licensing, and two active eDiscovery matters: the Data Governance bill can run $10,000-$25,000/month depending on scan tuning, and the licensing cost will dwarf it.
The lesson: scan tuning has more leverage on your monthly Azure bill than any other single configuration choice. Compliance licensing is a headcount calculation, optimize by identifying who actually needs the advanced features, not by blanket-assigning E5 to all users.
Tying It to a Governance Practice
Purview pricing decisions do not live in isolation. Data Map design (which sources you register, how you structure collections, what classification rules you enable) directly controls the scan footprint. On the compliance side, your detection engineering practice shapes how aggressively you configure Communication Compliance policies and how many Insider Risk cases stay open at any given time, both of which feed back into cost. We have seen orgs cut their Purview bill by 40% simply by auditing which policies were active and which were open but producing no actionable alerts.
Before you sign an EA amendment or expand Purview into a new business unit, do a scan-tuning audit and a policy rationalization pass. Both are faster than they sound and the combined savings usually pay for the engagement.
When to Call Us
If you are sizing a Purview deployment, navigating the E3-to-E5 compliance licensing decision, or trying to understand why your Azure Purview bill spiked, reach out at /contact and we will scope a fit-for-purpose architecture session.