A 90-day automation plan started this week finishes right as the year closes, which is exactly the point. September is the last real window most SME owners get before Q4 fills every spare hour with the busiest quarter of the year.
Start now, and you walk into January having actually fixed something. Wait until “things calm down,” and you’ll be having this same conversation with yourself next September.
This isn’t a plan that requires hiring anyone or signing anything. It’s a staged way to spend the next twelve weeks finding and fixing whichever manual process is quietly costing your business the most right now.
TL;DR: Weeks one and two: measure, don’t guess, using a simple one-week self-audit. Weeks three to six: automate the single highest-frequency task the audit surfaces, and only that one. Weeks seven to twelve: measure the result, then extend to a second task if the first one is working. Ninety days, three stages, one thing fixed properly rather than five things half-started.
Check the 90-day automation plan
The 90-day automation plan, stage by stage
The plan itself is deliberately simple, because complicated plans are the ones that don’t survive contact with a genuinely busy Q4. Three stages, four weeks each, one clear output per stage.
Weeks one and two: measure before you touch anything
The temptation in September is to start automating immediately, riding the motivation of a fresh quarter. Resist it for one week. Instead, run a simple self-audit: keep a tally across four categories, finance, sales and leads, customer support, and operations, and every time you or your team do something manual in one of those categories, mark it down.
Do this for a genuine working week, not a quiet one, and involve anyone else who touches these processes daily, since a business owner’s own week rarely reflects what’s actually happening on the front line.
By Friday, you’ll have a real answer to a question most owners are only guessing at: which category is actually eating the most hours.
It’s rarely the one you’d have named from memory. Owners tend to overestimate whichever category caused the most recent frustration and underestimate the quiet, constant drain of something nobody’s complained about lately, simply because loud problems get remembered, and quiet ones don’t.
Use week two to confirm the pattern held and to pick the single highest-frequency task within that category, the one that showed up in the tally more than any other.
Resist the urge to pick the most interesting task, or the one that would make the best story later. Pick the one with the most tally marks next to it, full stop. The whole point of the audit is to replace instinct with a number, and abandoning the number the moment it produces an unglamorous answer defeats the purpose of measuring anything at all.
Weeks three to six: automate exactly one thing
This is where most 90-day plans quietly fail, not through lack of effort, but through trying to fix everything the audit revealed at once.
Pick the single task from week two and automate that one, fully, before touching anything else. One process, finished properly and actually working, beats five processes half-automated and none of them reliable, and it’s a great deal easier to tell whether one thing is working than to untangle which of five overlapping changes caused which result.
Four weeks is a realistic window for one process, provided you use it properly. Spend the first week defining exactly what “done” looks like for this specific task, including what happens when something goes wrong, beyond the happy path alone.
Spend the middle two weeks building and testing it against real examples from your actual business, not hypothetical ones. Spend the final week watching it run before declaring it finished, because a process that works once in testing and a process that works reliably every day for a week are genuinely different claims.
Whether you build this yourself, bring in a freelancer, or work with a done-for-you provider is a separate decision from what to automate.
Make that decision based on your own technical confidence and available time, not on which option feels most exciting in week three. The task you chose in week two doesn’t change depending on who builds it, and neither should your standard for calling it finished.
Want a second opinion on where to start? The 2-Minute Assessment gives you a personalised recommendation in under two minutes.
Weeks seven to twelve: measure, then decide whether to extend
Once the first automation has been running for a few weeks, go back to the same measurement discipline you started with.
Is it actually saving the time you expected?
Has anything broken that needed fixing?
Only once you can answer both honestly should you consider automating a second task, using the same one-at-a-time approach that made the first one succeed.
If the honest answer is that it’s saving less time than hoped, or breaking more often than expected, spend this window fixing that before adding anything new. A second automation built on top of a shaky first one just doubles the number of things that can go wrong at once.
If the first ninety days go well, you’ll close out Q4 with one genuinely fixed process, a proven method for finding the next one, and a much clearer sense of whether you want to keep going alone or bring in help for what comes next.
That’s a considerably better position than the one most businesses are in every January: the same vague sense that something should be automated, and no method for deciding what.
Not sure where your 90 days should start?
The 2-Minute Assessment asks a few quick questions about how your business actually operates and points you at the highest-impact starting point, no call required.