Automate vs hire

Every operator faces the same fork: we need capacity, do we hire a person or automate the work? Most teams default to hiring because hiring feels safer and automation feels speculative. In reality, the answer depends on three variables, variability, volume, and reversibility, more than it depends on cost.

The three variables

Variability

How much does the work vary between instances? High variability favors humans. Low variability favors automation.

• Low variability, same inputs, same outputs, same steps every time → automate
• Medium variability, same steps but inputs vary → automate with human in loop, or AI-assisted
• High variability, every instance looks different → hire

Volume

How many instances per unit time?

• High volume (hundreds/day+) → automate almost regardless of complexity
• Medium volume (dozens/day) → automate if variability is manageable
• Low volume (handful/week) → usually hire; automation ROI isn't there

Reversibility

How bad is a mistake?

• Reversible, customer email gets typo'd → automate with retry
• Expensive, wrong invoice sent → automate with audit trail
• Irreversible, firing someone, signing a deal, sending a refund → human in the loop, always

The decision matrix

                    HIGH VOLUME              LOW VOLUME
                 ┌──────────────────────┬──────────────────────┐
HIGH VARIABILITY │  Hire + use tools    │  Hire                │
                 │  (e.g, senior CSM)   │  (e.g, executive)    │
                 ├──────────────────────┼──────────────────────┤
LOW VARIABILITY  │  Automate            │  Contract / offshore │
                 │  (e.g, data entry)   │  (e.g, quarterly     │
                 │                      │   reports)           │
                 └──────────────────────┴──────────────────────┘

The costs people forget

Costs of hiring

• Recruitment (3, 6 months work, 20, 30% of salary if using recruiter)
• Onboarding (2, 3 months to productivity)
• Management overhead (~20% of a manager's time per direct report)
• Benefits + overhead (40% load on top of salary)
• Opportunity cost of the seat (someone else could have been hired)

Costs of automation

• Upfront build (weeks to months of engineering)
• Maintenance (10, 30% of build cost annually)
• Brittleness when the underlying process changes
• Edge-case handling complexity
• Tooling + infrastructure subscription costs

The hybrid path, the best answer most of the time

Most real work doesn't neatly automate or fully require a human. The hybrid answer:

• Automate the 80%, the repetitive, low-variance parts
• Escalate the 20%, exceptions, judgment calls, customer-facing complexity
• Human reviews automation output before final action on anything irreversible

A single CSM with good automation serves 3x the accounts of a CSM without it. The question isn't "automate vs hire", it's "what's the right ratio of automated work to human work?"

The automation staircase

Automation isn't binary. The ladder:

1. Documented, process is written down
2. Standardized, same steps every time, different people can run it
3. Templated, inputs and outputs use shared templates
4. Tool-assisted, software does part of the work (e.g., pulls data, generates drafts)
5. Semi-automated, pipeline runs; human reviews and triggers
6. Fully automated, no human intervention except for alerts/exceptions

Climb the staircase. Skipping levels breaks things.

When to hire despite the math

• Customer-facing roles where presence matters (enterprise CSM, exec sales)
• Roles requiring judgment under uncertainty (hiring, strategic partnerships)
• Roles that generate the automation (ops engineer, RevOps, automation engineer)
• When the business is early and process is still being discovered, automating too early calcifies a bad process

What good looks like

• Hiring requests include a "why can't this be automated?" answer
• Automation projects include a "what's the human escalation path?" answer
• An automation backlog exists with ROI estimates for top candidates
• Ratios like revenue per employee are tracked quarterly

Related: Process mapping · Vendor management · Hiring, signal vs noise

What to do with this

• List every recurring task and tag it automate-first vs hire-first, the mix reveals where you're over- or under-investing
• Automate before hiring for repetitive rules-based work, one-time tool setup compounds vs permanent salary
• Hire for judgment-heavy roles, automation is bad at empathy + novel problems
• Track ratios like revenue per employee quarterly, it tells you if automation leverage is working
• Re-audit annually, tools get better and tasks that needed humans last year may be automatable this year