Before automating anything, run the numbers. Here's a simple framework for estimating the real return, and spotting the automations that aren't worth it.
Not every task is worth automating. The trick is knowing which ones are before you spend time and money building. Here's a back-of-the-envelope framework you can run on any process in about ten minutes.
The core formula
The annual value of automating a task comes down to:
(Hours saved per year × fully-loaded hourly cost) + value of fewer errors + value of speed
Then compare that to the cost to build and maintain the automation. If the payback period is under a year, it's usually a strong candidate.
Step 1: Measure the real time cost
Pick a repetitive task and estimate: how many times does it happen per week, and how long does each take? Multiply out to an annual number. Be honest, include the context-switching and the "quick" follow-ups, not just the core action.
A task that takes 15 minutes, done 20 times a week, is 260 hours a year. At a fully-loaded cost of $50/hour, that's $13,000 of time on one task.
Step 2: Add the hidden costs
Pure hours undercount the value. Add:
- Error cost: what does one mistake cost to fix, and how often does it happen? Manual data entry and copy-paste work carry real error rates.
- Speed value: does doing this faster make money? A lead followed up in 5 minutes converts far better than one followed up in 5 hours.
- Morale: repetitive drudgery is what burns people out and drives turnover. Hard to price, real to feel.
Step 3: Estimate the true cost to automate
Include build and maintenance. A good automation isn't free forever, APIs change, edge cases appear. Budget for upkeep, and be wary of automating a process that's about to change anyway.
Step 4: Score and prioritize
Rank your candidates by value-to-effort. The best first projects are high-frequency, rules-based, and error-prone, exactly the work that scales badly with headcount. Save the messy, judgment-heavy, rarely-done tasks for later or never.
The honest part
Some automations don't pay off, and a quick estimate saves you from building them. The framework's real value isn't justifying the yes's, it's giving you permission to say no to the work that looks automatable but isn't worth it.