The “2-Hour Rule”

TL;DR

  • If a recurring task takes more than about two hours per month and mostly follows a predictable script, you should not be doing it manually.
  • Audit the last 1–2 weeks of your work and list every “ugh, this again” task you keep repeating.
  • For each task, estimate time per month; anything over two hours goes on your “Do Not Do Manually” list.
  • Use the 2-Hour Rule Worksheet to make this a weekly habit, not a one-time brainstorm.
  • Capture the logic for each task (trigger, inputs, steps, exceptions, output) so a system can run it reliably instead of you.
  • For every task on that list, decide whether to automate it with tools, delegate the first pass to AI, or delete it entirely.

The “2-Hour Rule”: How to identify tasks you should never do manually again

If you are a founder, operator, or team lead, you are probably doing at least two jobs:

  1. The one on your LinkedIn
  2. And the secret one: Chief Human Glue

You know that second job.

You are:

  • Manually updating tools so they match reality
  • Nudging people who forgot things
  • Rebuilding the same reports every week
  • Copy-pasting the same info between apps
  • Clearing an inbox that mysteriously refills every few hours

You do not need a giant AI strategy deck to change this.

You need one brutally honest rule:
If a recurring task takes more than about two hours per month and mostly follows a script, you should never do it manually again.

This is the 2-Hour Rule. It is a simple way to start thinking like an operator designing systems, instead of a person numbing out and “just getting through” the week.

Let’s run it on your actual schedule.


Step 1: Run a “tiny annoyances” audit

Forget your job title for a minute. Look at where your time actually went.

Open:

  • Your calendar
  • Your task manager
  • Your email
  • Your DMs or team chat

Scan the last 7–14 days and mark anything that made you think:

“Ugh. This again.”

Write it down, even if it feels small or stupid. The tiny things are usually where the hidden hours live.

Common suspects:

  • Re-typing notes from calls into your CRM or project tool
  • Updating task statuses after every meeting
  • Pulling the same five numbers into a “quick weekly update”
  • Sending nearly identical follow-up emails after calls
  • Manually tagging or filing incoming messages
  • Rebuilding the same filters and views over and over

You are not solving anything at this stage. You are just surfacing all the invisible jobs that quietly eat your week.


Step 2: Apply the 2-Hour test

For each recurring “ugh” task, do some quick back-of-the-envelope math:
Time per occurrence multiplied by how often it happens in a month. I like this Chrome extension.

You are aiming for a rough estimate, not a spreadsheet-level analysis.

Examples:

  • Inbox triage
    15 minutes a day times 20 workdays is about 5 hours per month
  • Weekly metrics report
    30 minutes times 4 weeks is about 2 hours per month
  • Post-meeting admin (notes and tasks)
    10 minutes times 25 calls is more than 4 hours per month

Anything that crosses your “about two hours per month” line and is not genuinely creative, relational, or strategic goes on your Do Not Do Manually list.

The exact threshold is flexible. Some people use one hour, some use three. The point is to draw a line between:

  • “This genuinely needs my brain”
  • “This is operations pretending to be my job”

Step 3: Decide the fate of each task – automate, delegate to AI, or delete

Now you decide what happens to everything on that list.

There are only three options.

1. Automate (let tools do tool work)

These are tasks with clear triggers and predictable steps.

Think in terms of “when this happens, do these things”:

  • When a form is submitted, add a row to a sheet, create a CRM record, and send a notification
  • When a payment fails, send a reminder email, tag the customer, and notify the account owner
  • When a project moves to “Complete,” send a feedback survey, archive assets, and update status

This is the territory of your automation tools, whatever you use: Zapier, Make, n8n, or built-in workflows inside your CRM, help desk, or email platform. These flows become the hands of your system. They move data, update records, and send standard messages so humans do not have to babysit every step.

A simple rule of thumb:
If a human is just moving information between tools or clicking the same buttons repeatedly, it is a strong automation candidate.

2. Delegate to AI (let AI do the first pass)

Other tasks are a bit more “thinking heavy,” but still follow patterns. This is where AI is useful.

Examples:

  • Drafting follow-up emails after a sales or client call
  • Summarizing meetings into decisions and action items
  • Categorizing support tickets by topic and urgency
  • Sorting your inbox into “do now,” “do later,” “delegate,” and “ignore”
  • Turning long-form content into shorter posts, emails, and summaries

In these cases, AI acts like a junior ops assistant, not your replacement.

  • It reads, drafts, categorizes, and suggests
  • You skim, edit, and handle edge cases
  • You keep the relationship, judgment, and final say, but you lose the slog

The goal is not to remove you from the loop. The goal is to turn a two-hour grind into a ten-minute review.

3. Delete (the powerful “no”)

Some tasks do not deserve automation or AI.
They deserve to disappear.

Ask yourself:

  • If we stopped doing this entirely for a month, what would realistically break?
  • Is anyone actually using this report, dashboard, or update?
  • Does this still support our current strategy, or is it a leftover from a previous version of the business?

If nobody can give you a good answer, the most efficient option might be to stop doing it.

AI Operations is not “automate everything.”
It is “stop institutionalizing nonsense.”


Step 4: Capture the logic, not just the task

Automating something once is nice. Designing a reliable, inspectable flow is better.

For each task you have chosen to automate or delegate to AI, capture:

  • Trigger: what starts this?
    For example: “When a new lead submits the ‘Book a demo’ form.”
  • Inputs: what information has to be present?
    For example: name, email, company size, territory, product interest.
  • Steps: what happens, in order?
    1. Create a lead or contact record
    2. Assign an owner based on rules (territory, segment, product)
    3. Create a follow-up task
    4. Send a confirmation email
  • Exceptions: when should a human step in?
    For example: enterprise-size deals, unusual regions, or key accounts.
  • Output: what should be true when it is “done”?
    For example: the lead has an owner, an open task, and a confirmation email, and the CRM is up to date.

You can capture this in a simple document, a note, a Notion page, or a whiteboard tool. Use whatever you will actually open again.

Why this matters:

  • Tools and AI need clear, explicit instructions to behave well
  • You or someone else can debug and improve the flow later
  • You can change platforms in the future without losing the logic behind how things work

You are not just automating a task. You are documenting how work should move through your business.

That is AI Operations in miniature.


Step 5: Start tiny, but make it permanent

It is tempting to try to fix everything at once. A better approach is to start with one thing.

Choose a task that:

  • Hurts every single week
  • Has a clear trigger and simple steps
  • Is low-risk if the first version is not perfect

Design a simple flow. Implement it. Live with it for two weeks.
Then smooth out the rough edges.

Most importantly, make a rule for yourself:
This task is now illegal to do manually.

Tell your team. Write it down. Put it on a sticky note if you have to.

The win is not “we built a cool automation.”
The win is “this task will never quietly steal two hours from me every month again.”

Do that a few times, and your calendar and your nervous system will both look very different.


Putting the 2-Hour Rule to work (grab the worksheet)

Reading about the 2-Hour Rule is nice. Using it for 15 minutes is where it actually starts to change your week.

So instead of telling you to open a blank doc and build your own system from scratch, I made you a simple worksheet.

It walks you through:

  • Listing your “ugh, this again” tasks
  • Estimating how many hours each one quietly steals every month
  • Tagging each task as Automate, Delegate to AI, or Delete
  • Choosing one task to fix this week, with space to sketch the basic flow

Here is how to use it:

  1. Download the 2-Hour Rule Worksheet.
  2. Set a 15-minute timer.
  3. Fill in as many recurring tasks as you can remember from the last week or two.
  4. Circle anything over about two hours a month.
  5. Pick one circled task and commit to never doing that one manually again.

That is it. No big transformation, no giant implementation project.

Just one clear rule, one page, and one task at a time.

You can grab the worksheet below and run this experiment once a week. A month from now, you can either have the same calendar you have today… or a short list of things you never touch manually again.

The 2-Hour Rule is how you quietly choose the second option.

Leave a Comment

Your email address will not be published. Required fields are marked *