"How much should we spend on automation?" is one of the most common questions I get, and it's almost always the wrong question. There's no right number of dollars, no percentage of revenue, no benchmark that tells you a business your size should spend X on automation. Automation spend isn't a budget line you set in advance — it's a return you justify one task at a time. Get that framing right and the number answers itself.
Stop asking "how much" and start asking "for what"
A budget-first mindset — "let's allocate 5% to automation this year" — leads to spending the money whether or not there's anything worth automating, then reverse-engineering a justification. That's how teams end up with expensive tools nobody uses.
Flip it. The right spend on any given automation is anything comfortably less than what the manual version costs you — in time, in errors, in things falling through the cracks. Automation is a trade: you pay money once (build and maintenance) to stop paying time forever. So the real question is never "how much should we spend," it's "what is this task costing us, and is that more than the fix?" Which is really a sharper version of what to automate first.
The number that actually sets the budget
Before you can justify any spend, you need one honest figure for each candidate task: its fully-loaded cost. That's more than the minutes on a stopwatch. Add up:
Time × frequency × cost of that time
A 20-minute task done daily by someone whose hour is worth a real amount adds up to a serious annual number. Multiply it out — the yearly figure is what you're really comparing against, not the 20 minutes.
The cost of getting it wrong
Manual, repetitive work fails silently — a missed follow-up, a mistyped number, an invoice that never went out. Price in what those errors cost, because automation quietly removes most of them.
The opportunity cost
The most expensive part isn't the task — it's what your best people aren't doing because they're stuck on it. An hour of busywork reclaimed for real work is worth far more than the hour saved.
Add those up and you get the annual cost of leaving the task manual. That number — not a budget percentage — is your spending ceiling.
The payback rule
Here's the simple test I use: an automation is worth building if it pays for itself within a few months, and keeps paying after that. Take the fully-loaded annual cost of the manual task, compare it to the all-in cost of automating it — tools, build time, and the ongoing maintenance everyone forgets — and look at how quickly the savings clear the investment.
Fast, obvious payback (weeks to a couple of months): do it now. Payback measured in years, or that depends on the process never changing: leave it alone. Most teams badly underestimate the second half of that equation — automation isn't free once it's built. It breaks when a tool updates, when a process shifts, when an edge case appears. If maintenance will cost more attention than the task ever did, the "automation" is a liability with a nicer name.
Why cheap can be the expensive choice
The instinct to spend as little as possible often costs the most. A brittle, duct-taped automation that saves a few dollars up front but fails quietly every few weeks — sending the wrong email, dropping a record, corrupting a report — can cost far more in cleanup and lost trust than doing the task by hand ever would.
Spend enough to make it reliable, because an automation you can't trust is worse than no automation at all — now you've got the tool cost and the manual double-checking. This is the same principle behind good operations: the goal isn't the cheapest process, it's the one that runs predictably without someone hovering over it.
Where the money goes furthest
If you want the highest return per dollar, aim automation spend at the tasks that are high-frequency, rule-based, and directly tied to revenue or risk — the repetitive machinery of your pipeline and delivery. Automating the follow-ups and hand-offs inside a lead-generation system compounds every single week; automating a report you read twice a year does not. Spend where the task repeats and the stakes are real. Everywhere else, a human and a checklist is often the cheaper, smarter answer.
There's no benchmark for how much to spend on automation — only a return to justify. Price the fully-loaded cost of each manual task, spend up to (not beyond) what it's costing you, and only build things that pay back in months and stay reliable after. Automate the expensive, repetitive, revenue-tied work; leave the rest to a human and a checklist.
Read next
Not sure what's worth automating?
I help founders and teams find the automation that actually pays for itself — and skip the expensive tools that don't — see how I work.
Book a Call