Do You Still Need a SIEM If You Pay for MDR?
MDR covers detection and response. It often doesn't cover 12-month log retention, forensic reconstruction or cloud blind spots. Here's when a separate SIEM pays for itself, and when it's money you don't need to spend.
Usually Not, Unless Something Forces Your Hand
If you already pay a managed detection and response provider, the obvious question is whether the SIEM line in the budget can go. For many companies under roughly 500 employees, it can. The MDR's own telemetry from endpoint agents, identity logs and a few cloud connectors is enough to detect and contain most intrusions. A SIEM bolted on top often adds cost without adding coverage.
That's only half the SIEM vs MDR question. MDR is built to catch attacks happening now. A SIEM is built to keep records and connect them across systems. Four things usually justify keeping both: retention mandates longer than your MDR keeps data, forensic needs after a breach, estates spread across several clouds and SaaS platforms, and audits that require you to show who accessed what, months back.
If none of those apply, MDR alone is a defensible position. If one or more does, you probably still need a SIEM or a cheaper log archive. What's left to decide is how much of it you run yourself.
What Your MDR Provider Actually Keeps, and for How Long
Most buyers assume the MDR keeps everything. It doesn't. Providers built on EDR platforms such as CrowdStrike Falcon, SentinelOne, Microsoft Defender for Endpoint or Sophos commonly hold raw endpoint telemetry for anywhere from a few weeks to a few months, depending on the tier. Alerts and case notes often last longer, sometimes for the life of the contract. The raw process, network and authentication events investigators need usually don't.
Coverage is narrower than people expect, too. A typical MDR ingests endpoint data, your identity provider, email security and perhaps one cloud control plane. It may not ingest firewall logs, VPN logs, database audit trails, ERP access logs or admin logs from your SaaS applications. Those are often exactly the records an auditor or forensic team will ask for.
Before you decide anything, ask your provider for three things in writing: raw telemetry retention by data type, alert and case retention, and the log sources they ingest today, not the ones they support in theory.
SIEM Log Retention Requirements That Outlast MDR Telemetry
Retention rules are the most common reason a SIEM survives an MDR rollout, and they differ sharply by country and industry. Some examples of how they commonly show up:
- Payment card data (global): PCI DSS requires at least 12 months of audit log history, with the most recent three months immediately available for analysis.
- India: CERT-In's 2022 directions require organizations to keep ICT system logs for a rolling 180 days, within Indian jurisdiction.
- Saudi Arabia: the National Cybersecurity Authority's controls are widely read as requiring security event logs to be kept for a year or more, depending on the control and asset class.
- EU financial entities: DORA and NIS2 don't set a single number. They do require documented logging and the ability to reconstruct incidents, and supervisors often expect a year or more.
- Regulated sectors elsewhere: banking regulators in Singapore, the UK and Brazil, among others, set their own expectations through supervisory guidance and contract terms.
A 30-day MDR data store satisfies none of these. Data residency can matter as much as duration, so ask where the provider stores your logs. Confirm your obligations with counsel or a qualified compliance adviser, because interpretations vary by regulator and contract.
Forensics and Insurance Claims Need History You Can Reconstruct
Intrusions are rarely found on day one. Attackers often sit in an environment for days or weeks before anyone notices, and ransomware crews routinely stage access well before they encrypt anything. When a breach coach and a forensic firm appointed by your cyber insurer arrive, they'll want to establish initial access, lateral movement and data exfiltration. That requires logs from before the MDR first raised an alert.
Short retention costs you here. If you can't show what data left the network, counsel may have to assume the worst for notification purposes. That widens the scope of regulator and customer notifications, and the claim gets more expensive. Some insurers also ask about centralized logging and retention on applications and renewal questionnaires, so being able to answer "12 months, centrally stored, immutable" helps your underwriting position.
This doesn't require a full SIEM with correlation rules. Often a cheap, immutable log archive that forensic teams can query is enough.
Multi-Cloud and SaaS Estates Where Endpoint Telemetry Goes Blind
MDR grew out of endpoint protection, and it's strongest where an agent can be installed. A company running workloads across AWS, Azure and Google Cloud, plus Salesforce, Workday, GitHub and a dozen other SaaS tools, has large areas where no agent exists. Attacks there show up as a misused API key, a rogue OAuth grant or a new admin-created access token, not as a malicious process.
Some MDR providers now ingest cloud and SaaS logs directly. Others charge extra per source or cover only the major cloud control planes. If your crown-jewel data lives in SaaS applications, check whether the MDR writes detections for those logs or just stores them.
Our view: an organization with two or more clouds and significant SaaS data usually needs a SIEM-like layer to normalize and correlate those sources. Whether the provider runs it or you do matters less than making sure someone writes the detections.
Microsoft Sentinel vs MDR Is the Wrong Framing
Buyers often compare Microsoft Sentinel vs MDR as though they were substitutes. They aren't. Sentinel is a platform that stores, correlates and alerts on logs. An MDR service is people who watch alerts and respond. You can run Sentinel with no one watching it, and that's a common and expensive failure.
For Microsoft-heavy organizations, the most practical setup is often an MDR that operates inside your own Sentinel workspace. You own the data and the retention settings. The provider brings analysts, detection content and response playbooks. If you change providers, your logs and history stay with you.
This works well because some Microsoft sources, such as Microsoft 365 audit logs and certain Defender alerts, have historically been free or discounted to ingest (check current licensing terms), and long-term retention can sit in lower-cost archive tiers. High-volume non-Microsoft sources are where the bill grows. Similar provider-operated models exist on Splunk, Google Security Operations and Elastic.
What Running Both Actually Costs
SIEM cost is driven mostly by data volume, not seat count. Firewall, DNS and proxy logs can account for most of your ingestion while producing few useful detections. The table shows typical 2026 market ranges for a mid-size organization. They vary widely by contract, region and volume, so confirm with vendors.
| Approach | Typical added annual cost (USD) | Best fit |
|---|---|---|
| MDR only, provider retention | None beyond MDR | No retention mandate, mostly endpoint and identity risk |
| MDR plus low-cost log archive | $5,000–$25,000 | Retention or forensic needs, limited correlation needs |
| MDR operating in your own SIEM | $20,000–$120,000 in ingestion and storage | Regulated, multi-cloud, or want data ownership |
| MDR plus separate in-house SIEM team | $150,000–$500,000+ including staff | Large enterprises with internal detection engineering |
Hypothetical worked example: say a 300-person company pays roughly $10–$30 per endpoint per month for MDR, or about $36,000–$108,000 a year. It generates 30 GB of logs a day. At an assumed $2–$5 per GB in analytics tiers, sending everything to a SIEM would add about $22,000–$55,000 a year. If it routes only identity, cloud and critical server logs to analytics (about 8 GB a day) and archives the rest, the SIEM bill could fall to roughly $6,000–$15,000 while still meeting a 12-month retention rule. Archive storage adds a small amount on top.
Red Flags That Quietly Push You Toward a SIEM
Watch for these in MDR proposals and contracts:
- Retention described as "available on request" with no number of days per data type.
- Your data held in the provider's proprietary tenant with no export format or exit clause. If you leave, your history goes with them.
- "Supports" integrations with no detection content. Ingesting a log is not the same as monitoring it.
- Per-source surcharges that make adding cloud or SaaS logs cost more than a SIEM would.
- No commitment on forensic data access during an incident, especially for third-party responders appointed by your insurer.
- Storage regions that conflict with local residency rules in India, the Gulf states, the EU or elsewhere.
Any two of these together mean you should keep logs in a store you control.
A Decision Rule for Your Next Renewal
Ask four questions. Do you have a retention obligation longer than your MDR's raw telemetry window? Would a forensic team need logs your MDR doesn't ingest? Do you run two or more clouds or hold sensitive data in SaaS? Will an auditor ask for access history older than 90 days?
If every answer is no, run MDR alone and revisit the decision each year. If one is yes, add a low-cost immutable archive. If two or more are yes, choose an MDR that operates inside a SIEM you own, and route only high-value logs to the expensive analytics tier.
Your next step: send your MDR provider the retention request described above, and ask your compliance adviser and broker which retention periods actually apply to you. With those answers, you can settle the SIEM vs MDR decision on facts rather than vendor positioning.
Frequently asked questions
Can my MDR provider just extend its retention instead of me buying a SIEM?
Often, yes. Many providers sell extended retention, commonly in the range of several months up to a year, as an add-on. Check that the extension covers raw telemetry, not just alerts, and that you can export the data if you leave. Compare the add-on price against a cheap archive tier in your own cloud.
Is SIEM cost mainly licensing or data volume?
For most modern cloud SIEMs, data volume is the biggest cost driver, usually priced per GB ingested or per daily capacity commitment. Engineering time to tune rules and manage parsers is the next biggest cost and is often underestimated. Filtering noisy sources before ingestion is the most effective way to control spend.
Does using Microsoft Sentinel mean I don't need MDR?
No. Sentinel provides detection and storage, but someone still has to triage alerts around the clock and respond. Without an internal 24/7 team, most organizations pair Sentinel with an MDR or managed SOC provider that works inside their workspace.
Will my cyber insurer require a SIEM?
Few insurers require a SIEM by name. Many ask about EDR, MDR or 24/7 monitoring, and some ask about centralized logging and retention periods. Requirements differ by carrier and market, so ask your broker what your insurer's application and renewal questions expect.
What's the cheapest way to meet a 12-month log retention rule with MDR?
Send the required logs to immutable, low-cost object storage or a SIEM's archive tier, and keep about three months in a searchable tier if your rules require immediate access. This usually costs a fraction of full SIEM analytics ingestion. Confirm with your compliance adviser that the format and access controls satisfy your regulator or auditor.
Written by the Sentinel Underwrite Ops editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.