Manual work is a recurring cost that never appears as a line item
Software shows up in the budget. Contractors show up in the budget. The four hours a week someone spends copying leads from a form notification into a CRM shows up nowhere — it is absorbed into a salary that was going to be paid anyway, which is exactly why it survives for years.
Pricing it changes the conversation. Four hours a week at a fully-loaded $48 an hour is just under $10,000 a year, for one task, for one person. Most teams have five or six of those running simultaneously and have never added them up.
Getting the hourly figure right
Using salary alone is the most common error, and it understates the answer by a quarter to a third. The fully-loaded cost of an employee includes payroll taxes, benefits, equipment, software seats, and the share of management time they consume.
A practical approach: annual salary ÷ 1,880 working hours gives the raw rate, then add 30% as a general-purpose overhead multiplier. A $70,000 salary lands near $48 an hour. Use the rate of whoever actually does the task — if a senior person is doing data entry, the arithmetic gets worse, and that is genuinely the situation you are trying to surface.
What this model excludes, on purpose
The estimate counts labour hours and nothing else. Three real costs are deliberately left out, which is why the output should be read as a floor:
- Error cost. Manual transcription has a non-zero error rate. A mistyped email address is a lead you paid for and cannot contact.
- Delay cost. A person processes a queue on their schedule; a workflow processes it on arrival. For anything time-sensitive this dominates the labour saving — see the speed-to-lead calculator for what that gap is worth.
- Opportunity cost. The work that did not happen because the time went elsewhere. Real, and not honestly quantifiable in a calculator.
The honest counter-argument
Saved hours are not saved dollars unless something changes. Automating twelve hours a week almost never reduces headcount, and presenting it that way to a finance team invites a fair objection. What it produces is released capacity, and the value is entirely determined by what that capacity gets pointed at.
Automation also has real costs the enthusiasm tends to skip: build time, tooling subscriptions, and maintenance when a system changes underneath a workflow. Net savings and payback period are the numbers to present, not gross hours reclaimed.
Sequencing the work
Automate high-frequency, rule-based tasks first, weighted by what the delay costs. Resist automating a process that is broken — a bad workflow executed reliably at scale is worse than a bad workflow executed occasionally by someone who notices. Document the process, fix it, then automate it.
If you are not sure whether the foundations are in place, the readiness scorecard checks the ten prerequisites that decide whether automation compounds or just automates chaos. For the return side of the same equation, run the marketing ROI calculator.