Q.How is MSP pricing typically structured?+
A.Three common models. Per-user or per-device pricing (a monthly rate per seat) is simplest and most predictable. Per-server or per-instance pricing is common for infrastructure-focused MSPs. All-you-can-eat block pricing is a flat monthly fee covering everything within a defined scope. There is no published market rate: MSPs quote, and the rate moves with what it includes (ticket allowance, after-hours cover, security tooling, project work). Ask every bidder to normalise its price to a per-endpoint monthly figure with the same inclusions listed, so the bids compare.
Q.What should a SOC RFP for an MSP include?+
A.Six requirements: SOC staffing hours and whether the SOC is the MSP's own or a named subcontractor; the log sources ingested, the SIEM or XDR platform and who owns the data at exit; monthly time-to-detect, acknowledge and contain reported from raw alert data; a runbook of response actions the SOC may take without approval; detection engineering mapped to MITRE ATT&CK; and an incident notification timeline that meets your own regulatory obligations. Ask separately for the MSP's SOC 2 Type II report, which covers the MSP's own controls.
Q.Should I require a co-managed or fully outsourced model?+
A.Depends on internal IT capacity. Fully outsourced (MSP owns everything from helpdesk to architecture) suits organisations with no internal IT or with a small team that wants to focus on strategy. Co-managed (internal team owns user experience and strategy; MSP owns tooling, monitoring, after-hours, and specialist domains) suits larger organisations and is the more common 2026 model. The RFP should specify which model and let MSPs that do not operate in that model self-disqualify.
Q.What security framework should the MSP follow?+
A.NIST CSF 2.0 is the practical default. It covers Identify, Protect, Detect, Respond, Recover, and Govern. CIS Controls v8 maps neatly to it and is more prescriptive for implementation. ISO 27001 is common in EU contexts. Ask the MSP which framework they operate against, where they self-attest vs where they are independently audited, and how they report security posture to clients. Generic 'we follow best practices' answers are red flags.
Q.What SLA targets are realistic for help desk and incidents?+
A.First response: P1 (system down) 15 minutes, P2 (major impact) 1 hour, P3 (minor impact) 4 hours, P4 (request) 1 business day. Resolution: P1 4 hours, P2 8 hours, P3 24 hours, P4 5 business days. Treat these as a starting position for your draft, not a market benchmark: bidders will counter-propose, and tighter P1 and P2 targets usually cost more.
Q.How long should the transition / onboarding take?+
A.60 to 90 days for a typical mid-market organisation (100 to 500 endpoints). Phases: discovery and documentation (2 weeks), tooling deployment (2 to 3 weeks), parallel run (2 to 4 weeks), cutover (1 week), stabilisation (4 weeks). A 30-day transition usually means the MSP skipped discovery and will be relearning your environment during incidents.
Q.How do I evaluate MSPs that all claim 24/7 support?+
A.Ask: where the after-hours team sits geographically, whether after-hours is staffed by the same engineers as business hours or a separate team, what the after-hours escalation looks like, and what percentage of tickets are resolved at the L1 vs escalated. Many MSPs technically offer 24/7 via a small night-shift team that only triages, then waits for the day shift to resolve. Honest disclosure of staffing patterns is more useful than a 24/7 marketing claim.