Apply three tests. Frequency — how often does it recur, since a weekly task compounds far faster than an annual one. Dependency — does anything else wait on it, because fixing something downstream of a broken input just moves the problem. And judgment content — how much of it genuinely requires you, since the tasks worth handing off first are those that are mostly assembly with a small decision at the end. For most small businesses that points at the month-end close, which is heavy in recurring effort, sits upstream of nearly everything else, and is roughly ninety percent assembly with one real decision at the end.
What this actually feels like
Everything feels like it should be automated, which makes it hard to start anywhere. So either nothing happens, or the most annoying task gets picked.
The most annoying task is rarely the most valuable one. Annoyance tracks unpleasantness, not cost, and the two are only loosely related.
What it costs to do by hand
Working out where to start takes an hour of honest thinking: listing what recurs, what depends on what, and how much of each task is genuinely judgment.
Most owners skip that hour and choose by feel, which is how businesses end up with an automated marketing calendar and books that still do not reconcile.
Building a weekly business dashboard
A worked example of the principle — six numbers, chosen by what changes a decision rather than by what is easy to measure.
Read the guide →Input, skill, output
- Plain English — What you actually need.
- A worry — Or the thing that is bothering you.
- Or a file — Something concrete to work from.
- Read intent — What you actually need.
- Pick the path — The right one of 31.
- Walk you through — Step by step, with checks.
- A recommendation — Which task to start with, and why.
- The reasoning — Frequency, dependency and judgment content.
- A first step — Something concrete to do this week.
Who this is for
- You know several things could be improved and cannot decide where to start.
- You have tried automating something before and it did not stick.
- You would rather fix one thing properly than five things partially.
If none of those describe you, this is probably not your first priority — and the rest of the library may point somewhere more useful.
Common questions
Should I start with what annoys me most?
Usually not. Annoyance tracks unpleasantness rather than cost, and the two correlate only loosely.
What are the three tests?
Frequency, dependency, and how much genuine judgment the task contains. The best first candidate scores high on the first two and low on the third.
What does that usually point at?
The month-end close, for most small businesses. It is heavy in effort, upstream of nearly everything else, and mostly assembly.
Why does dependency matter?
Because fixing something downstream of a broken input moves the problem rather than solving it. Clean data feeding unreliable numbers helps nobody.
What if I get it wrong?
The cost is mostly the second attempt being harder, because a failed first automation makes everyone skeptical of the next one.
See it on your own books, once
The First Close is one real month of your books, closed properly and configured to your chart of accounts, delivered inside two weeks. Full reconciliation against every processor, every exception flagged with the transaction behind it, and the narrative written. $497, refundable if the packet is not delivered, and it credits in full toward the first month if you continue.
Get Your First Close — $497 See a sample packet