"Annoying" and "worth automating" are different measurements, and confusing them is how people lose a month. Score every automation before you build it — the arithmetic takes about eleven seconds and it will save you more time than any tool in the book.
The calculator, the threshold, and the worked example are all on this page. No email needed for any of it.
First-year return on the hours you invest.
Payback = (minutes saved per week × 52) ÷ (hours to build × 60)
Estimate the build time, then double it. Everyone's first estimate is optimistic and everyone believes they're the exception. The calculator below doubles it for you — enter your honest guess.
Anything under 10× doesn't get built. Not "gets deprioritized" — doesn't get built.
Sorted by payback. Build the top three and close the rest — not defer, close.
Nothing scored yet. Add a task above.
Some automations aren't worth saved minutes — they're worth prevented failure. The chain that emails a customer their login when Stripe fires, at 3am, without you, isn't saving six minutes. It's preventing the version of your business where someone pays $149 and waits eleven hours in the dark.
Score those on consequence, not on clock, and build them regardless of ratio. Flag them separately in your ledger so you don't talk yourself out of them at 1.2×.
What yours should look like by Sunday night. Two of these are genuinely painful and still don't get built.
| Task | Min/wk | Build ×2 | Payback | Build? |
|---|---|---|---|---|
| Reformat monthly reports for banks | 126 | 4h | 27× | Yes — first |
| Answer the same four setup questions | 48 | 3h | 13.9× | Yes |
| Rewrite one post for three platforms | 55 | 5h | 9.5× | No |
| File orders into the tracking sheet | 22 | 2h | 9.5× | No |
| Pull Stripe payouts into a summary | 3.5 | 3h | 1× | No |
| Send login credentials after purchase | 4 | 3h | 1.2× | Build anyway |
The two tasks at 9.5× are genuinely painful and still don't get built this month — that's the threshold doing its job. In month three, when volume had doubled, the platform-rewrite task re-scored at 19× and got built immediately. That's the system working, not the system being wrong the first time.
The three he did build cost about ten hours and returned 178 minutes a week, permanently — just under three hours, recovered every week, for about a weekend and a half of work.
This page scores tasks. The sheet keeps them — and pairs with the time audit that produces the list in the first place.
It's a spreadsheet and a base template — neither pastes into a web page, so tell me where to send them.
One email a week about building things that run without you. Unsubscribe link at the bottom of every one, and it works.
© 2026 Jeff Sheikowitz. All rights reserved.