Hiring a managed security service provider hands an outside company standing administrative access to your firewalls, logs and often your identity provider, yet most MSSP selections are decided on price per device and a glossy service catalogue. An MSSP typically runs a mix of managed firewall and VPN, log collection and SIEM monitoring, vulnerability scanning, endpoint tooling and compliance reporting, which is broader and more operational than a managed detection and response (MDR) service. This vendor-neutral buyer's guide is for security leaders, IT managers and procurement teams running an MSSP selection or renewal. It covers how to define scope, the capabilities and contract terms that matter, the questions that separate providers, red flags, and a proof of concept plan. For detection-led services, read it alongside the MDR Buyer's Guide.
Key takeaways
- Scope is the product. Write down every device, log source, tenant and process the MSSP will run, and who keeps responsibility for each. Anything not in a schedule is not in the service.
- The provider becomes a privileged identity in your estate. Its engineers, tools and service accounts need the same MFA, least privilege and logging you demand of your own administrators, and you should be able to see and revoke that access yourself.
- Measure service levels from a defined timestamp. Agree time to notify, time to act and time to restore by severity, with the clock start written into the contract.
- You must own your logs, configurations and detection content. Contract for export in usable formats, a retention period you set, and a tested exit plan before you sign.
- Certification is a starting point, not evidence. Cyber Essentials Plus, ISO/IEC 27001 or a SOC 2 report show a baseline; a proof of concept and named references show whether the service works for an organisation like yours.
Define requirements
Start with outcomes and responsibilities rather than a tool list. MSSPs package services differently, so a requirement framed as "managed SIEM" can mean anything from forwarding alerts by email to an analyst team that tunes rules weekly. Build a responsibility matrix first, then turn it into the statement of work.
| Service area | Typical MSSP role | Decide before you go to market |
|---|---|---|
| Managed firewall, VPN and secure web gateway | Rule changes, firmware, monitoring, availability | Change approval route, emergency change window, who holds the root or super-admin credential |
| Log collection and SIEM monitoring | Onboarding sources, correlation rules, 24x7 triage | Which log sources are in scope, retention period, whether the SIEM tenant is yours or the provider's |
| Vulnerability scanning and patch reporting | Scheduled scans, reports, remediation tracking | Whether the MSSP patches or only reports; target times by severity |
| Endpoint protection management | Policy, agent health, alert handling | Pre-approved response actions such as host isolation |
| Identity and cloud monitoring | Sign-in anomaly alerts, privileged role changes, SaaS audit logs | Read-only or write access to the identity provider, and break-glass ownership |
| Incident response | Triage, containment support, retainer hours | Whether forensic incident response is included or a separate retainer |
| Compliance and reporting | Monthly service reports, evidence for audits | Which frameworks you need evidence for (ISO/IEC 27001, PCI DSS, DORA, NIS2) |
- Size and complexity: NCSC guidance on choosing a managed service provider is written for organisations of up to 250 people and points larger organisations to its cloud security guidance. A small firm usually needs a bundled service with clear defaults; a large organisation needs a co-managed model where its own SOC and the MSSP share tooling. See the Co-Managed SOC and SOC-as-a-Service Guide.
- MSSP or MDR: choose an MSSP when you need devices and platforms operated for you; choose MDR when you need threat detection and response on top of tooling you or a vendor already run. Many organisations buy both, so define the boundary in writing to avoid alerts falling between two providers.
- Regulated sectors: under DORA or NIS2 an MSSP is an ICT third-party provider, so register it, assess concentration risk and contract for audit rights. See the DORA ICT Third-Party Risk and Register of Information Guide.
Capability areas
Score each provider against these areas. The weights below are illustrative for a mid-sized organisation with a small internal security team; adjust them to your risk profile and keep the weights fixed before any vendor presentation.
| Capability area | What good looks like | Illustrative weight |
|---|---|---|
| Coverage and onboarding | Named list of supported log sources and devices; documented parsers; onboarding plan with dates; health monitoring that alerts when a source goes silent | 15% |
| Detection quality | Detection content mapped to MITRE ATT&CK, tuned to your environment, with a change log you can read; false positive handling with feedback loops | 15% |
| Response authority and speed | Pre-approved actions per severity, contracted time to notify and time to contain, 24x7 analyst escalation by phone | 15% |
| Provider access security | Named individual accounts, phishing-resistant MFA, just-in-time privileged access, session logging, access reviews you can see | 15% |
| Data ownership and portability | Logs, rules, configurations and case history exportable on request and at exit; you set retention; data location stated | 10% |
| Reporting and governance | Monthly service report with agreed metrics, quarterly service review, executive summary, audit evidence | 10% |
| Provider resilience and assurance | Independent assurance (ISO/IEC 27001 certificate with scope, SOC 2 Type II report), tested business continuity, own incident response plan covering compromise of the provider | 10% |
| Commercial clarity | Pricing unit stated (device, user, gigabyte, events per second), overage rules, price for adding sources, exit fees | 10% |
Where identity fits
An MSSP is a non-human and human identity problem as much as a service. Its engineers sign in to your consoles, its automation uses service accounts, API keys and multi-tenant applications, and its remote management agents run with high privilege on your hosts. MITRE ATT&CK tracks the abuse of this access as T1199 Trusted Relationship, alongside T1133 External Remote Services and T1219 Remote Access Tools. The ConnectWise Cloud Backup breach showed the scale of the risk: a stolen credential for one vendor's multi-tenant application produced confirmed unauthorised sign-ins to 162 Microsoft 365 tenants in a single security vendor's customer base, with no user phished and no MFA prompt. Ask every MSSP to list each application, service principal and API credential it will hold in your tenants, with permissions and owners. For Microsoft 365 partner access, see the GDAP and MSP Delegated Admin Access Guide, and for general vendor access controls the Third-Party Access Guide.
Evaluation criteria
Turn the capability areas into criteria that can be evidenced. Ask for evidence, not assertions, and record it in the scoring sheet so the decision can be audited later. The Buyer's Guide Methodology and Scoring Template gives a reusable scoring sheet.
| Criterion | Evidence to request | Pass threshold (example) |
|---|---|---|
| Service levels | SLA schedule with definitions of each severity and the timestamp the clock starts from; last two quarters of SLA performance for a comparable customer | Time to notify for critical alerts stated in minutes, measured from alert creation, with service credits |
| Patching and vulnerability handling | Process for devices the MSSP operates; reporting format | Critical and high-risk patches on managed devices within 14 days, in line with NCSC guidance |
| Log retention and access | Retention per source, storage location, your direct access route | Retention you choose (often 12 months searchable or longer for regulated firms), read access for your team at all times |
| Provider access controls | Access model, MFA type, privileged access tooling, sample session log | No shared accounts; phishing-resistant MFA for all administrative access; access removed within one business day of staff leaving the account |
| Incident notification | Contract clause, escalation tree, notification of incidents affecting the provider itself | Notification timeframes per severity, including provider-side compromise |
| Assurance | Certificate and statement of applicability scope, SOC 2 report, penetration test summary | Scope covers the service you are buying, not only head office |
| Exit | Exit plan, data return format, transition support days, deletion certificate | Full export of logs, rules and configuration at no extra charge beyond agreed transition days |
For reading assurance reports properly, see the How to Read a SOC 2 Report Guide. For the procurement process around this table, see the Security Vendor Evaluation and RFP Guide.
Questions to ask vendors
- Who exactly will have access to our systems? How many named staff, in which countries, under what vetting, and how will we see their sign-ins in our own logs?
- Which machine identities will you hold? List every service account, API key, OAuth application and remote management agent, with permissions, rotation schedule and how we revoke each one.
- What happens when you are compromised? Show the incident response plan for a breach of your own platform, how quickly we are told, and how you would cut your access to our tenant.
- How is the SLA clock measured? From event, from alert or from analyst acknowledgement, and what counts as "response"?
- Which actions will you take without asking us? Host isolation, account disablement, firewall blocks: list them by severity and show the approval path for anything else.
- Who owns the detection content and the SIEM tenant? Can we export rules, playbooks and cases in a usable format during the contract and at exit?
- How do you tune detections for us? Show a change log from a current customer, the false positive rate you report, and how we request a new use case.
- How do you know a log source has stopped? Show the source health alert and the last time it fired for a customer.
- What is excluded? Forensics, legal support, out-of-hours changes, cloud workloads, OT networks: what is extra and at what rate?
- Can we speak to two customers of our size and sector? NCSC advises asking for client references, particularly from organisations similar to yours.
Red flags
- Shared or generic admin accounts: a provider that asks for one global administrator account for "the team" cannot give you an audit trail.
- No customer-side visibility: the logs sit only in the provider's tenant and you receive PDFs. You cannot verify the service or investigate independently.
- Response times without definitions: "15 minute response" with no statement of what starts the clock or what response means.
- Certification scope that excludes the SOC: an ISO/IEC 27001 certificate for a sales office says nothing about the operations centre that will hold your credentials.
- Exit fees or proprietary formats: detection rules and case history that cannot leave the platform lock you in and leave you blind on the day you switch.
- Silence about their own breaches: a provider that will not discuss past incidents or its plan for a compromise of its remote management platform has not thought about it.
- Remote access tools you did not approve: remote monitoring and management (RMM) agents installed widely without an inventory, MFA or logging. See the Compromised MSP Playbook for what that costs when it goes wrong.
Proof of concept
A paid or time-boxed pilot of four to six weeks is the only reliable way to compare detection and service quality. Run it on a representative slice of your estate, not a lab.
- Agree the scope and success criteria in writing: pick five to ten log sources (identity provider, firewall, endpoint, one SaaS platform, one cloud account), the measures you will score, and who from each side attends weekly.
- Grant access the way you would in production: named accounts, MFA, least privilege, logged sessions. How the provider handles this step is itself evidence.
- Onboard and check health: confirm every source is parsed, time-synchronised and alerting on silence. Deliberately stop one source and measure how long the provider takes to notice.
- Run controlled test scenarios: with written authorisation, use benign emulation (for example a scheduled test of impossible travel sign-ins, a new privileged role assignment, a blocked outbound connection to a test domain) and record time to notify and quality of the notification. Do not use live malware or real attacker tooling.
- Test the escalation path out of hours: raise a simulated critical incident at night or at a weekend and measure whether the named contact is reached within the agreed time.
- Request an export: ask for logs, rules and case notes from the pilot in the exit format the contract will specify.
- Score and close down: score against the fixed weights, then remove all pilot accounts, applications and agents and confirm removal in your own logs. Treat a provider that leaves access behind as a fail.
Measure after go-live: track mean time to notify for critical alerts, percentage of in-scope sources reporting in the last 24 hours, number of provider accounts and machine identities versus the agreed inventory, false positive rate on escalated alerts, and percentage of SLA targets met each month. Review these at a quarterly service review, and re-run the access inventory at least every six months.
Framework mapping
| Framework | Reference | Relevance to MSSP selection |
|---|---|---|
| NIST CSF 2.0 | GV.SC-05, GV.SC-07, GV.SC-10 | Security requirements in supplier contracts, assessing supplier risk, and planning for the end of the relationship |
| NIST CSF 2.0 | DE.CM-01, DE.AE-02, RS.MA-01 | Monitoring and analysis the MSSP performs, and incident response coordinated with third parties |
| NIST SP 800-53 Rev. 5 | SA-4, SA-9, SA-9(2), SR-6 | Acquisition requirements, external system services, functions and ports used by the provider, supplier assessments |
| NIST SP 800-53 Rev. 5 | AC-6, AU-11, IR-4, IR-6, SI-4 | Least privilege for provider access, log retention, incident handling and reporting, system monitoring |
| ISO/IEC 27001:2022 Annex A | 5.19 to 5.22 | Supplier relationships, security in supplier agreements, ICT supply chain, monitoring and change of supplier services |
| CIS Controls v8.1 | Control 15 Service Provider Management | Inventory, classification, contract requirements, assessment, monitoring and secure decommissioning of providers |
Standards and references
- NCSC: Choosing a managed service provider (MSP)
- NCSC: Supply chain security guidance
- NIST: Cybersecurity Framework 2.0
- NIST: SP 800-53 Rev. 5 Security and Privacy Controls
- CIS: Control 15 Service Provider Management
- MITRE ATT&CK: T1199 Trusted Relationship
How NHI Mgmt Group can help
We provide independent requirements, RFP and evaluation support. Browse vendors in our products directory or contact us.
Related NHI Mgmt Group resources: MDR Buyer's Guide · Cybersecurity for MSPs and MSSPs Guide · Third-Party Risk Management Guide · SIEM Buyer's Guide · Choosing an IT Support Provider or MSP Securely