Video walkthroughs · Claude Code Club

The Five Rules That Turn AI Work Into Real Results

12 minute read · Companion guide to “How to Turn AI Into Results Faster Than 99% of People”

TL;DR

Results with AI show up late, so judge the week by what you shipped. Stick with one thing and make it a little better each day, drop the rest, ship it before it feels ready, then count how many people saw it. The five rules are an order, and the last one, traffic, is the one most people skip. The Friday scorecard on this page turns them into five questions you answer in five minutes.

For a long time I used AI to do more: more projects, more tools, more content. None of it turned into results. The biggest shift for me was not a better prompt. It was how I think about the work. This guide walks through the five rules I use to turn AI work into real results, in the order they depend on each other, and ends with a one-page scorecard you fill in every Friday.

Watch the full walkthrough, then use this page as the checklist.

Watch the full walkthrough, then use this page as the checklist.

The five rules are an order, and the last one is the one you skip

The five rules are an order. Each one only works once the one before it holds. Showing up every day only pays off if you keep getting better at it. You cannot get a little better every day at one thing if you are juggling ten. One project only teaches you something once it is out in the world. And shipping only helps if enough people actually see it.

That chain is the idea to take away. Skipping ahead wastes the effort, because a rule built on top of a missing rule has nothing underneath it.

The last rule is the one almost everyone skips. Claude makes building so easy that you keep building, and building feels like progress, but nobody ever sees the thing. Nine times out of 10 for the people I talk to, traffic is the problem, and traffic is the last rule.

Rule 1: Judge your week by inputs, not outcomes

Here is the trap. You build something with Claude, you post it, and you check the numbers the next morning. Nothing happens, so you decide AI did not work and jump to the next tool. Results with AI show up late, so a next-morning check always reads as failure.

I once saw a TikTok video with a line I think about all the time: post once a day, every day, on TikTok for a year and never work another day again in your life. The video was not about chasing virality, views or sales. The only question was whether you can show up every day and do the thing.

My own channel followed that shape. My first 1,000 YouTube subscribers took three to four months, and I kept showing up. Three months later I had 5,000, and three months after that I had 15,000.

Navy graphic of a root growing under the ground line while the day counter above the ground shows almost nothing yet, titled Results show up late.
Most of the growth happens below the line, long before anything shows above it.

So the rule is inputs, not outcomes. Focus on the things you control, like how much you show up and how you get better. Views, sales and followers are outcomes. That is the approach I took, and the outcomes followed.

Here is the step to try this week. Pick one thing you can finish with Claude, like a video script or a small tool. Ask Claude to help you cut it down until you can finish it by Friday. Then on Friday ask one question: did I ship it? If yes, you won the week, whatever the numbers say.

Rule 2: Get 1% better every day

AI work rarely clicks on the first try. The trap is building something for three weeks, not seeing results fast enough, jumping ship to the next thing and starting over from zero.

Navy graphic of a small piece that climbed a staircase and then jumped to a new, tiny staircase, with the counter reset to week 0, titled Start over, start at zero.
Every jump to a new thing resets the counter.

The rule is to get 1% better every day. You stick with one thing and make it a little better each day.

I do this with my community. I added a video sales letter, which is a short video that explains what the community is, to the About page, so people who land there can see what is inside and decide quickly whether it is for them. Then I made a small adjustment: I matched the one line in my YouTube video descriptions to the top line of the About page, so people see the thing they came for. I keep adding new lessons and resources based on what members are interested in, and I keep trying to make better YouTube content.

To start, pick one thing you have already built and improve one small part of it with Claude each day. Tell Claude what the thing is, name the one part you want better today, and ask for the smallest change that improves it. Naming the part before you open the chat keeps the day small.

Rule 3: Do less, because AI makes starting too easy

AI makes it far too easy to start new things. You can spin up a new app or a new agent in one afternoon, so most people end up working on too many things and wondering why they get no results or have no time.

Navy graphic of ten test tubes sharing one liquid, none of them reaching the fill line, titled Too many things.
One supply of effort spread over ten projects fills none of them.

I did this too. I tried to maximize every social channel, like Instagram, X and TikTok. Now I pay people to run my Instagram, which is a privileged position to be in, and if I could not pay them I simply would not do it. I learned to focus on the highest leverage activities.

So the rule is do less. There is something in your life or your business with the greatest impact for you, and all your time, energy and effort goes into that one thing. For me that was YouTube. I stopped worrying about the other platforms and asked whether I could make more YouTube content consistently. Then I used rule two on it. Can I make it better? Can I package it better, with stronger titles and thumbnails? Can I write better hooks and better intros?

Here is the step to try. List every AI project you have going right now. Circle the one closest to a real customer or a real client. Pause the rest for 30 days and close all of those Claude chats. When two projects look equal, pick the one a real person could use or pay for soonest.

Rule 4: Ship before you're ready

AI makes it tempting to keep polishing forever. Claude can always make the thing a little better, so it never feels done, and real users never touch it.

I learned this the hard way with a product called Showplace. You pasted in a Zillow listing link and it made a 40 second cinematic video of the home. It went in front of real customers and a few of them signed on. Then it became clear that the founding team, me included, were not the right founders to grow it, so the product was shut down. I am still proud of the build. But it took months of building and thousands of dollars, and it should have shipped much earlier. Shipping early would have brought the customers sooner, shown that it was a product I was never going to grow, and let me move on to the next thing.

Light graphic of a timeline where the ship flag moves from the far end to the start, followed by the words customers, learn, move on, titled Ship earlier.
Move the flag to the start, and the learning starts with it.

So the rule is ship before you're ready. You will never feel ready. Stop trying to perfect it and ship messy, get feedback from real people, iterate and make it better. Take risks, put yourself out there and take on users.

Here is the step to try. Take the thing you are building with Claude and send it to five real people. Ask each one what confused them. Then give Claude the answers and ask it to fix the top one.

Rule 5: Find your bottleneck, because fixing the wrong problem gets you nowhere

There are three things you bounce between in your business. Traffic is how many people look at your thing. Conversions is how many of those people buy. Churn is how many people stay, especially with a subscription or agency clients. One of them is usually your bottleneck. Once you solve that one, you look at the next, and I solve them in that order.

Navy graphic of a pipe in three sections labelled traffic, conversions and churn, with an orange ring on the narrow first section, titled Find the bottleneck.
The narrowest section sets how much gets through, so fix that one first.

For most people, the narrow section is the first one. People ask me to fix their About page all the time. Then I look and they have 22 members in a free community. Nobody knows you exist, and that is okay. It is fixable.

It was obvious in my own case what the reason was: I was not driving any traffic to my community. So I started putting the community link in my YouTube descriptions, and people started showing up. I also started saying it out loud in the videos: do you want to join the community? Check out the link.

For a sense of scale, as of this video in November 2026 the Claude Code Club About page has had 137,000 visitors in the last 30 days, and the community has over 11,000 members. Now compare that to what not enough looks like. Your website has been seen by 50 people, or you sent five LinkedIn connection requests and 20 cold emails, or you made one Instagram reel. I guarantee your traffic is not enough.

Decide which bottleneck is yours with these tells

Start with one number: how many people saw your thing last week. Write it down. The number points to the branch.

For traffic, I use Alex Hormozi's idea of more, better, new. First, can you do more of the thing you are already doing? Then, can you do it better? Only once you have maxed out how much and how good, look at new ways to get traffic. The same order applies to new offers and new products.

Conversions was the next problem for me. Once people were coming, not enough of them were deciding to click the buy button, so I made a guess and then looked at what the top people in my space were doing. Whichever branch you land on, fix that one first and then look at the next.

Know the five ways this goes wrong

Almost every stalled project traces back to one of these. Each one is a rule skipped, and the fix is the step from that rule.

Fill in your Friday scorecard

Everything above comes down to five questions, one per rule, in the same order. The scorecard below takes five minutes. Copy it into a note, or paste it into Claude.

Light graphic of five numbered tiles, each with a rule name: inputs over outcomes, 1% better every day, do less, ship before you're ready, find your bottleneck.
The five rules in order. The scorecard asks one question per tile.
textTHE FRIDAY FIVE
Week of: ________

1. INPUTS. Did I ship the one thing I picked?
   Yes means: it is finished and out where another person can see it.
   If no, Monday: cut the thing smaller until I can finish it by Friday, then pick it again.
   Answer: Yes / No

2. 1% BETTER. Did I improve one small part of the same thing each day?
   Yes means: I can name the small change I made on each day.
   If no, Monday: name tomorrow's one small part before I open Claude.
   Answer: Yes / No

3. DO LESS. Is only one project active, with the rest paused?
   Yes means: one project is circled, the others are paused for 30 days, and their Claude chats are closed.
   If no, Monday: list every AI project, circle the one closest to a real customer or client, pause the rest.
   Answer: Yes / No

4. SHIP. Did five real people see it, and did I ask each one what confused them?
   Yes means: I have five answers and Claude is fixing the top one.
   If no, Monday: send it to five people and ask what confused them.
   Answer: Yes / No

5. BOTTLENECK. How many people saw my thing last week? ______
   Yes means: I wrote the number down and named my bottleneck.
   Under 100: traffic. Monday: ask Claude to help me do more of what I already do, like one post a day.
   Plenty see it, few buy: conversions. Monday: have Claude research the top players in my space.
   They buy but do not stay: churn. Monday: ask the people who left why they left.
   Answer: Yes / No

Score: ___ of 5. The earliest No is next week's focus.

Each line has three parts: the question, what counts as a yes, and the one move to make on Monday if the answer is no. The earliest No on the card is your focus for next week, because a later rule cannot work until the earlier one holds. If you paste the filled card into a Claude chat, ask which line you keep missing and what the one smallest move is for Monday.

Go from reading to doing in one sitting

Here is the shortest route from finishing this page to having the first Friday ready.

  1. Pick one thing you can finish with Claude this week, like a video script or a small tool.
  2. List every AI project you have going and circle the one closest to a real customer or client. Pause the rest for 30 days and close those Claude chats.
  3. Name the one small part of the circled project you will improve tomorrow.
  4. Find out how many people saw your thing last week and write the number down.
  5. If the number is under 100, ask Claude to help you do more of what you already do, like one post a day.
  6. Copy the Friday Five into a note and put a reminder in your calendar for Friday.
  7. On Friday, fill it in, find the earliest No, and write down Monday's one move.

Show up every day and judge your week by what you shipped, not by the numbers. Pick one thing and make it a little better every day. Put your effort into the one thing with the biggest impact, ship it before it feels ready, and get feedback. Then find your bottleneck, which for most people is traffic.

Common questions

Why is traffic the rule most people skip?

Claude makes building so easy that you keep building, and nobody ever sees the result. Nine times out of 10 for the people I talk to, traffic is the problem. It is the last rule because the others come first, but it is the one that decides whether anyone finds your work.

What if my numbers are still tiny after a few weeks?

Results with AI show up late. My first 1,000 YouTube subscribers took three to four months, then 5,000 came three months later and 15,000 three months after that. Judge the week by whether you shipped, which is the first line of the scorecard, and check your traffic number against the 100 people tell.

Do I have to give up my other projects for good?

No. Rule 3 says to pause the rest for 30 days and close their Claude chats, while you put your effort into the one closest to a real customer or client. The point is to give one project enough attention to learn from it.

How do I know when something is ready to ship?

You will never feel ready, and Claude can always make a thing a little better, so the feeling is not a signal. I built Showplace for months and spent thousands of dollars before real customers touched it, and I should have shipped earlier. Send it to five real people, ask what confused them, and have Claude fix the top answer.

What does more, better, new mean?

It is Alex Hormozi's idea, and it is the order I use for traffic. Do more of the thing you are already doing, then do it better, and only once you have maxed out both, look at new ways to get traffic. The same order applies to new offers and new products.

Want the lessons and the people behind these rules?

The five rules are yours. The Claude Code Club is $9 a month and built to help you master Claude Code in 45 days, with lessons, resources and members building websites, apps, automations and skills alongside you.

Join the Club — $9/mo

Read this online at claudecodeclub.ai