The Leverage Audit
A leverage audit scores how much of your working week could run without you. You log the hours you spend on 15 recurring tasks, each is weighed against how much of it automation can genuinely absorb, and you get a Leverage Score out of 100 plus a ranked list of what to automate first. A high score means your time is already concentrated in work only you can do.
It takes about five minutes and runs entirely in your browser — nothing is submitted, stored or sent anywhere. Built for one-person businesses, where the constraint is never headcount and always hours.
Put hours against the tasks you actually do. Leave the rest blank.
Every number this tool uses, and why
Each task carries a ceiling: the share of that work current tooling can realistically take, assuming you still review the output. They are deliberately conservative — a vendor demo will always quote a higher number than the thing achieves in a real week. Reclaimable hours are your hours multiplied by the ceiling; the score is 100 minus the reclaimable share of your total. A year is counted as 48 working weeks, because nobody works 52.
Communication
A drafting layer, not a sending layer. Give a model your last hundred replies as context and let it write the first version; you approve or rewrite. The approval step is what protects quality, and it still cuts the time by roughly three quarters.
Purely mechanical and long since solved. A scheduling link plus an automated reminder sequence removes nearly all of it. If you are still negotiating times in a thread, this is the cheapest hour you will ever buy back.
Transcribe, then summarise against a fixed template. The follow-up message writes itself from the transcript, and it is more accurate than what you'd remember an hour later.
Workflow and admin
The highest-yield automation in any solo business, every time. n8n, Zapier, or twenty lines of script. The rule: if you have copy-pasted between the same two systems three weeks running, it was never a task — it was a missing integration.
Template the generation and automate the chase sequence. The chasing is the part people avoid, which is exactly why handing it to a scheduled job improves cash flow more than the time saving suggests.
A dashboard that regenerates on a schedule doesn't speed up reporting — it deletes it. The work was never the analysis, it was the assembly.
Knowledge
Models are strong at the sweep and weak at the judgement. Automate the collection and the first-pass filter; keep the conclusion. The failure mode is trusting a summary of sources you never checked.
Capture should be instant and unstructured; structuring is a batch job a model runs later. Trying to file at capture time is why most note systems die in month three.
Record yourself doing the thing once and have a model write the procedure from the transcript. Nobody writes SOPs from a blank page, which is why nobody has any.
Planning and judgement
What automates is the input assembly — pulling together what happened, what's open, what's due. The decision itself stays yours, and the time you spend on it is not overhead.
This is the job. A model can argue with you usefully and should, but if you automate the choice you have automated the only role a one-person business actually has. Low ceiling here is the correct answer, not a limitation.
Collection and presentation automate almost completely; interpretation doesn't. The gain is that you review weekly instead of quarterly, because the friction is gone.
Execution and delivery
Repurposing and second drafts automate well. The original angle does not, and the ceiling here is lower than almost anyone selling you a content tool will admit. Automate the distribution, not the thinking.
Templates and generators. If you have built the same structure three times by hand, the fourth should be generated — and building the generator usually costs less than the fifth build.
Entirely dependent on what you sell, so assume low until you've proven otherwise on your own work. The parts that reliably automate are setup, QA and handover — rarely the substance.
What the bands mean
- 0–39Running on manual labour
- Most of your week is work a machine could be doing. That's not a criticism — it's the normal starting position, and it means the first two automations you build will return more than everything after them combined.
- 40–59Half machine work
- You're carrying a substantial automatable load alongside real work. The usual cause is that the automatable parts are spread thin across the week, so none of them ever feels big enough to fix. Batch them and they are.
- 60–79Reasonably leveraged
- The obvious wins are mostly done. What's left is the harder tier — the work that needs a system built rather than a tool bought. Returns get smaller from here, so be selective about which ones you chase.
- 80–100Highly leveraged
- Your week is concentrated in work that genuinely needs you. At this point more automation is not the constraint — demand, positioning or pricing almost certainly is. Chasing the last few hours is the wrong problem.
If the ranking looks daunting
The top two items on your list are almost always worth more than everything below them, and both are usually buildable in a weekend. If you would rather have them built than build them, that is roughly what I do for a living.
How I work with people