Knowing When You Need Help: MSPs, Contractors, and the Second Hire
Every one-person IT department has a ceiling. Not a ceiling on skill — on hours, on simultaneity, on the plain arithmetic of one human and a growing business. The dangerous thing about the ceiling is how it announces itself. It brings no clean signal, just a slow accumulation of slipped maintenance and permanent project deferrals. And there is a person who has quietly stopped taking real time off. By the time the announcement is unmistakable, the fix has become a crisis purchase, made quickly and badly. The announcement might be a burned-out resignation, or an outage that one exhausted responder could not catch in time.
This guide is the capstone of the solo-admin series. And its job is to help you see the ceiling early and act before the crash. You will get the objective signals that one person is no longer enough, each one measurable from data you already collect. You will get an honest comparison of the four ways to add capacity: MSPs, contractors, managed-service slices, and the second hire. You will get the structure of a proposal management can actually approve. And you will get the mechanics of the transition itself, where your solo rigor finally pays its dividend. At the end: a scorecard you can fill in honestly in ten minutes.
In this guide
- The signals you have outgrown one person
- The options, compared honestly
- Building the case for management
- Transition mechanics, whichever path you choose
- The self-assessment scorecard
- Where to go next
The signals you have outgrown one person
Feelings are unreliable here. Solo admins habitually normalize overload. So use signals that can be counted, and count them from records you already keep. Five carry most of the weight:
- After-hours pages trending up. Not one bad month — a slope. Your alert history has the exact numbers per quarter, and the trend line matters more than any single figure.
- The maintenance calendar slipping, quarter after quarter. When restore tests, threshold reviews and firmware passes keep getting bumped by urgent work, operations has eaten your margins. The calendar's completion record is the measurement.
- Single-point-of-failure risk the business now notices. When your manager, an auditor, or a cyber-insurance questionnaire starts asking "what happens if you're hit by a bus?", the risk has become organizationally visible. Write down each time the question is asked, and by whom.
- Project work permanently displaced by operations. Compare the project list from a year ago with today's. If the same items sit there, aged and untouched, while new operational load arrived on top, capacity ran out some time ago. The deferred list, with dates, is the exhibit.
- The vacation test failing. You built the coverage plan from the vacation guide. The honest conclusion was: there is no viable delegate, and two weeks away is a genuine business risk. A failed vacation test is a failed availability test for a system the business depends on — and it is documentation, not a complaint.
Each signal maps to a data source you already have, because you built it in the earlier guides:
| Signal | Where the data lives | What "worrying" looks like |
|---|---|---|
| After-hours load | Alert history; page-lane log | Rising quarter over quarter |
| Maintenance slippage | The maintenance calendar | Same tasks bumped 2+ quarters running |
| SPOF visibility | Notes of who asked, when | Outsiders now raise it unprompted |
| Project displacement | The deferred-projects list | Items aging past a year; list growing |
| Vacation test | The coverage plan exercise | No credible plan without you |
| Your own load | The time log; ticket volumes | Fragmentation up, focus time near zero |
Two or three of these trending the wrong way at once is the ceiling talking. The next question is what kind of help actually fits, because the four options solve different problems. Buying the wrong one is how companies conclude "we tried getting help and it didn't work."
The options, compared honestly
Read each option below for three things. First, look at what it is for: the specific ceiling it lifts. Next, look at its cost character: the shape of the spend, not a figure. Prices vary too much by region and scope for numbers to be honest here. Finally, look at its fit signals: which of your worrying trends it actually answers. Matching the option to the signal is the whole game. Every option on this list has made some company miserable by being bought for the wrong problem.
The MSP: breadth, coverage, and a relationship to manage
A managed service provider (MSP) is a company that runs IT operations for many clients at once. That is exactly why it can do the things one person cannot. It can answer at 3 AM, absorb a flu week, and bring specialist depth on demand. MSP arrangements come in two broad shapes. There is full management, where they take over the environment, and co-managed, where they work alongside an internal admin. For a solo shop that is not actually drowning, co-managed is usually the sweet spot: they take the after-hours pager and the overflow; you keep the architecture, the judgment, and the institutional knowledge. That single change — someone else on nights and your vacations — removes the two heaviest stones in the solo backpack.
Here are the frictions, stated plainly so you can plan for them rather than be surprised. An MSP runs on standardization, so responses to anything off their script can feel procedural where you would have used judgment. They do not know that the finance server is fragile about reboots or that the warehouse switch lies about its uptime. The tribal-knowledge gap is real and closes only through your documentation and briefings. And the relationship itself is a standing item on your calendar — scope reviews, escalation-path checks, holding response expectations to what was agreed. An MSP is not a fire-and-forget purchase. It is a vendor relationship that pays off roughly in proportion to how well you manage it. Fit signals: after-hours load and the vacation test are your worst signals, and the work overwhelming you is routine operations rather than deep environment-specific projects.
The contractor: for the spikes
Some ceilings are not about steady load at all. They are about spikes: the server migration, the phone-system replacement, the office move. A contractor or project specialist exists for exactly this. Using one well is how projects stop eating operations (and vice versa). An admin doing a migration between interrupts does both jobs badly, slowly, at night. Costs are project-shaped: a scoped engagement rather than a subscription. The discipline that determines success is almost entirely yours. It means a written scope with a definition of "done," and knowledge transfer as a deliverable, not a hope. The contract should say the project ends with documentation you accept and a handover session you schedule. Otherwise you have bought a system only the contractor understands. That is to say you have bought yourself a second SPOF with an hourly rate. Fit signals: the deferred-projects list is your worst signal while daily operations remain manageable.
Managed-service slices: shrinking the surface
The quietest option is not adding people at all but subtracting surface: moving whole functions to services that run them for you. Examples include hosted email security, backup-as-a-service, hosted phone systems, and cloud-hosted line-of-business apps. Each slice moved is a set of servers not patched, backups not babysat, and expertise not maintained. That is in exchange for a subscription and some loss of control at the edges. The strategic value for a solo admin is that slices are incremental and reversible in planning terms. There is no reorganization, no big proposal. There is just one function at a time, chosen by the ratio of "hours this costs me" to "how standard is this need." Rule of thumb: the less differentiated the function — spam filtering being the classic — the better it slices. Fit signals: your load is spread across many routine systems rather than concentrated in a few, and budget conversations go better in operating-expense-sized bites.
The second hire: when sustained load justifies a person
The full answer to sustained, growing, business-hours load is another human — and it changes more than arithmetic. With two people, rotations exist, vacations are real instead of engineered, sick days stop being emergencies, and the bus factor finally halves. The design question is junior or peer. A peer doubles your ceiling fastest but costs the most and overlaps your skills. A junior costs less, takes the routine load off you quickly, and grows. And a solo shop that followed this series is unusually well set up to grow one. That is because the runbooks, checklists, and monitoring you built are a ready-made training track. The junior runs the documented procedures. The documentation improves every time they stumble, and competence transfers system by system. For most solo shops the junior is the right first hire. Hire the peer when what you lack is not hours but a second opinion at your own level. Fit signals: every signal in the table trending wrong at once, and the load clearly permanent rather than a phase.
| Option | Best for | Cost character | Watch out for |
|---|---|---|---|
| MSP (co-managed) | After-hours coverage, overflow, vacation resilience | Monthly retainer, scales with scope and seats | Tribal-knowledge gap; the relationship needs managing |
| Contractor | Project spikes; keeping projects off operations | Per-project | Scope drift; knowledge leaving with them |
| Service slices | Shedding routine functions one at a time | Subscription per function | Edge-case control loss; slices still need an owner |
| Second hire | Sustained, growing, business-hours load | Salary and everything around a person | Onboarding takes real time before it pays back |
These combine well. A common evolution is a slice or two first, then a co-managed MSP for nights, and the hire when growth makes the load case undeniable. The comparison is not a bracket where one option wins; it is a menu you will likely order from more than once.
Building the case for management
Management does not approve "I'm overloaded." Management approves risk reduction and capacity purchases they can defend upward — so build the case in those materials. Three ingredients:
- Risk. State the single point of failure without melodrama, backed by receipts. Those are the near-misses in your incident record, the coverage question the insurance form asked, and the failed vacation test. If you have been keeping the uptime evidence file, this section nearly writes itself. The record shows what almost happened and who caught it. The answer to "who caught it" is always the same name.
- Capacity. The deferred-projects list with business impact attached in the business's own words — not "no time for the wireless refresh" but "the warehouse dead zones stay until this is staffed." Plus the time-log tally showing where the week actually goes.
- Cost honesty. Present the options with their cost character side by side. Name the cost of doing nothing in qualitative but concrete terms. Burnout-driven turnover means losing the one person who knows everything, then hiring their replacement into an undocumented environment anyway. That is the expensive version of this proposal, purchased at the worst moment. No invented figures; the logic lands on its own.
Then the proposal itself has four short sections in order. Start with current state (the evidence, one page). Follow with risk statement (what the current design bets the business on). Then come options (two or three, with rough cost character and what each does and does not solve). End with recommendation (one option, a first step, and a review date). Short enough to read in five minutes; every claim traceable to a record.
Timing matters as much as content. The ask lands best after the evidence file has a few months of receipts. A trend is persuasive where a bad week is dismissible. The ask also lands best before a crisis forces the decision, because crisis purchases skip evaluation and shop badly. The worst time to choose an MSP is the week after the outage that proved you needed one. If the signals are trending wrong now, start the file now and set yourself a date two quarters out. The discipline of a date keeps "I should raise this" from becoming another deferred project.
Tip: rehearse the counterargument you will definitely hear: "but everything is running fine." It is the visibility paradox of good solo work — the better you do the job, the less necessary help appears. The answer is the evidence file itself. Fine-so-far is the record of one person catching everything. The proposal is about what happens when that one person is on a plane, in a hospital, or in a new job. You are not arguing the present is broken; you are pricing the fragility of how the present is achieved.
Transition mechanics, whichever path you choose
Approval is the halfway point. The transition is where the value is won or lost. Here the habits from the rest of this series stop being virtuous and start being negotiable assets. The documentation and monitoring you built is the onboarding pack. An MSP's first question is "what does the environment look like?" And your answer is an auto-discovered inventory, a network map, runbooks, and a monitoring history instead of a shrug and a whiteboard session. The same pack cuts a new hire's ramp-up from months of oral tradition to weeks of guided reading. Solo rigor, it turns out, was never just for you.
Access discipline matters most in the first week, when enthusiasm wants to hand over the keys. Use named accounts for every outside person (never shared logins), least privilege scoped to their actual duties, and logging on from day one. And — for contractors especially — set an expiry date on the account the day it is created. That is because "we'll disable it when the project ends" is how ghost accounts are born. None of this signals distrust; done matter-of-factly, it signals an operation worth joining.
Then give the integration a structure instead of a vibe — checkpoints at roughly one, two, and three months:
- ~30 days: access is correct and least-privilege is verified. The pack is read. First responsibilities are live (for an MSP, the after-hours pager on a trial basis; for a hire, the first runbook-driven duties). You are still the safety net.
- ~60 days: they own their lane end to end. Incidents in scope are handled without you. The documentation is updated where reality disagreed with the pack (it will; that is the pack improving).
- ~90 days: the honest review: what do they now own outright, what still bottlenecks on you, and what does the next quarter transfer? For a hire, this is where the training track gets its next milestones; for an MSP, where scope gets tuned against actual ticket flow.
The quiet goal across all three: you stop being the single point of failure on purpose, one responsibility at a time, with a record of the handoff. That is the ceiling actually lifting.
The self-assessment scorecard
Ten honest minutes. Score each signal 0 (fine), 1 (wobbling), or 2 (failing), and resist the solo-admin reflex to round down:
| Signal | 0 — fine | 2 — failing | Your score |
|---|---|---|---|
| After-hours pages | Rare, flat trend | Regular and rising | |
| Maintenance calendar | Mostly on schedule | Same tasks slipping for quarters | |
| Vacation test | Plan exists, tested | No credible two-week plan | |
| Deferred projects | List short and moving | Aging past a year, growing | |
| SPOF visibility | Nobody asks | Raised by others unprompted | |
| Ticket backlog age | Stable, honest SLAs kept | Promises quietly slipping | |
| Focus time | Real blocks most days | Fragmentation total; log proves it | |
| Your own honest fatigue check | Tired on bad weeks | Every week is the bad week |
Read the total in bands. 0–4: the solo design is holding — keep tuning, keep the evidence file warm, re-score twice a year. 5–8: the ceiling is in sight. Start the receipts deliberately, and open the cheap conversations — a service slice, a vacation-coverage retainer — while they are still optional. 9–12: build the proposal this quarter. The data you need is already in your records, and waiting only moves the decision closer to crisis pricing. 13–16: you are past the ceiling and running on reserves. Take the case to management now, imperfect, because the version of this conversation that happens after you break is worse for everyone, you included.
Where to go next
There is a quiet arc across this series. The monitoring setup watches so you don't have to. Next, the automation returns your hours. The vacation machinery proves the operation survives your absence. The pace keeps your judgment sharp. Finally comes this guide, where all of it becomes the case for the help you have earned. Being a department of one was never supposed to be a life sentence. Done well, it is the best possible audition for running a department of two. The whole series lives at The IT Department of One.
Put this into practice with SysPeep Monitor
Whichever path you take, the handover leans on the records you kept. SysPeep Monitor auto-discovers and maps your network. It keeps the metric and alert history that becomes your evidence file. It gives an MSP or a new hire one web console that explains the environment on day one. It runs on a single Windows machine — the free trial is the easiest first receipt.
Download Free TrialLearn About SysPeep