ChatGPT workflows to save time every week, and how to prove they did

Monday — Clear the inbox into decisions

The cost of an inbox is not the reading, it is the re-reading. One pass, one list, then act.

Here are this morning's messages: {paste}. Sort them into: needs a reply from me today, can wait to Friday, can be answered with a saved reply, and no action. For the 'today' group, draft each reply under 100 words in a direct, plain tone. Do not invent commitments, prices or dates — if one is needed, leave {BRACKETS} for me to fill.

Monday — One input, four channels

Writing four separate posts is what makes people stop posting. Write the thing once, then adapt it — and keep the adaptation honest to the source rather than inflating it for each platform.

Source piece: {paste the article or notes}. Produce: a LinkedIn post under 180 words, an X post under 240 characters, a short email under 120 words, and three subject lines. Each must be built only from claims present in the source. No hype words, no earnings claims, no emojis. End each with a single specific next step.

Any day — Quote in ten minutes instead of an evening

Quoting drags because you are pricing and writing at the same time. Split it: decide the number yourself, then let the model write the document around it.

Job: {description}. Scope in: {bullets}. Scope out: {bullets}. Price: {number}. Timeline: {dates}. Write the quote with sections: what you get, what is not included, timeline, payment terms, and what happens if scope changes. Under 400 words. Do not add services I did not list.

Any day — Support without the evening shift

Every support reply that takes more than five minutes is a candidate for a saved answer. Build the library as you go rather than as a project.

My policy: {paste verbatim}. Customer message: {paste}. Write the reply under 120 words, then write a generalised version of the same reply I can save for future customers with {BRACKETS} where details change. Stay inside the policy. If the policy does not cover this, say so and stop.

Friday — The review that takes six minutes

This is the workflow that keeps the others honest, because it runs on numbers rather than on how the week felt.

This week: {sessions, signups, sales, refunds, hours worked}. Last week: {same}. Name the one number that moved most and the single likeliest cause. Name one thing to change next week and one thing to stop. If the data cannot support a cause, say 'unknown' instead of guessing.

Prove the time saving, or drop the workflow

A workflow that you cannot measure is a habit, not a system. The maths is deliberately simple: minutes saved per run, times runs per month, against your hourly value and whatever the tooling costs.

Anything that comes out negative, or that you ran fewer than three times in a month, gets dropped. That test is why the list above is six items long instead of thirty.

• Time it once, honestly, before and after.
• Count real runs per month, not intended ones.
• Compare against your hourly value and the subscription cost.
• Book the recovered hour deliberately, or it refills with admin.