Use a balanced ticket sample
| Case | Required observation |
|---|---|
| H01. Routine approved reset | Identity verified, permitted action and clear confirmation |
| H02. Insufficient identity evidence | Safe refusal and accepted escalation |
| H03. Workstation fault | Accurate diagnostic notes and permitted remediation |
| H04. Unsupported application | Useful evidence sent to the correct next owner |
| H05. Reopened ticket | History retained and no unexplained duplicate billing |
| H06. Branded telephone call | Correct greeting and approved customer wording |
| H07. Abusive or distressed caller | Agreed conduct and escalation procedure |
| H08. Unresolved overnight request | Accepted morning handoff and customer update |
Include a case the desk cannot solve. Its value is showing how the service reaches a boundary without abandoning the user or inventing a resolution. Match the same cases and input information across candidates; otherwise apparent differences may reflect easier tickets.
Use the identity verification case brief when specifying approved recovery, target-account authority and held outcomes. Extend the identity case with a claimed lost phone and a request concerning a privileged account. Use authorized test identities; keep the approved route, permission boundary and a held/escalated outcome when verification fails. The account-recovery requirements keep those decisions separate.
Pilot case log
Keep the first result, evidence, correction and retest together for each case. The blank log includes 8 suggested cases, the five timer labels and a separate scope-acceptance decision.
Use Word for an editable working copy with space for each observation and retest, or open Markdown in a text editor. Both contain the same blank cases. Complete your copy for your approved environment.
Score evidence, not charisma
Use a rubric with separate pass, fail and not-observed states for identity, tenant mapping, technical accuracy, permitted action, clarity, documentation and handoff. Do not calculate a performance score from not-observed cases. A critical identity failure should remain a blocking event, even if the technician was friendly and other tickets were quick.
Illustrative assessment: a test reset has clear communication and correct branding, but the technician proceeds without the required second verification step. Document identity as fail and stop enrollment until correction. This is a hypothetical case; it is not a claim about any listed provider.
Measure clocks and retained effort
Document contact time, acceptance, first useful action, restoration and final closure separately. Mark pauses and approval dependencies. Measure the MSP engineer's actual review, correction and escalation time in the pilot; outsourcing does not automatically remove that work. Ask how the sample maps to billable units and whether reopened or abandoned contacts affect the quote.
Customer satisfaction collected from your own consenting pilot users is different from public vendor testimony. A provider-hosted comment can guide questions, but it cannot replace observed identity and workflow evidence. Document how many cases and channels were tested so the conclusion stays within that scope.
Release the service deliberately
Accept only the tested scope and document unresolved exclusions. Train the retained team on escalation and emergency takeover before expanding. Use the white-label desk requirements, desk model comparison and transition plan to keep the next phase controlled. Put remediation dates and owners into the procurement RFP.