Automation that pays for itself in a month
The best automation targets are boring, repetitive and slightly embarrassing. Here is how to find them, and how to know when automating is the wrong answer.
By Shashi Kumar ·
Most automation projects fail the same way: someone automates the interesting problem instead of the expensive one.
The interesting problem is usually rare, complex and full of judgement calls. The expensive one is usually boring, frequent and mildly humiliating to describe out loud.
Automate the second one.
How to find it
Ask the people doing the work one question:
What did you do more than five times this week that a reasonably smart person could have done with no training?
You will get an answer immediately. It will be something like:
- Copying form submissions into a spreadsheet, then into the CRM
- Renaming and filing invoices
- Chasing the same three people for the same status update
- Reformatting a report that arrives in the wrong shape every Monday
None of these are interesting. All of them are pure cost, every week, forever.
The arithmetic that decides it
Two numbers:
- Cost to automate — hours to build, plus a realistic estimate of ongoing maintenance. Be honest here; integrations break.
- Cost of not automating — minutes per run × runs per week × 52, plus the error rate, plus the fact that nobody enjoys doing it.
If the payback period is under three months, it is nearly always worth it. Over a year, usually not — the process will probably change before you break even.
The trap is that people estimate the first number and feel the second one. Write both down.
When automation is the wrong answer
Genuinely, sometimes it is:
- The process is about to change. Automating a process that is being redesigned next quarter is building something to throw away.
- The volume is low and the judgement is high. Three complex cases a month should stay with a human. Automation there costs more than it saves and removes the person who understood the exception.
- Nobody owns it. An automation with no owner becomes an unexplained system that everyone is afraid to touch. That is worse than the manual process, because at least the manual process is legible.
- The real problem is upstream. If you are automating the cleanup of bad data, fix the form that produces it. I have watched teams build elaborate normalisation pipelines for a problem that was one required field away from not existing.
CRM work specifically
Most CRM customisation requests are actually process requests wearing a technical costume. "Can we add a field?" usually means "our sales process changed and nobody updated the system".
The useful move is to ask what happens after the field is filled in. If the answer is "nothing, we just like having it", you have found a field that will be empty within two months and a report that will quietly lie to management.
The pattern that works
Start with one process. Instrument it so you can see it running. Give it an owner. Prove the saving. Then do the next one.
Automation programmes that try to do everything at once produce a lot of half-finished pipelines and a team that no longer trusts any of them.
Contact Acquiretech Software Solution: info@acquiretechsoftwaresolution.in · WhatsApp +91 80786 33912 · Send an enquiry