Start Smaller Than You Think
A guide to how to build great things, lessons learnt from working with Olympians and other elite organisations
We’ve all been trained to think big.
Big vision. Big systems. Big impact.
It’s how strategy decks are written, how consultants pitch, and how executives reassure boards that the future is under control.
The trouble is, I now think that this approach is starting to look … really really slow.
The contrast between the leaders I work with that are moving fast right now is that we are doing something that would be considered amateurish five years ago.
We are starting small. Ridiculously small.
In fact, if you wrote it down on paper, it would barely pass the test of looking like a plan at all.
When we started our boxing project 5 years ago, our job was to analyse boxing. Boxing has a lot going on. We thought small. We figured out if we could tag and detect one single punch, then we could tackle other punches.
Then we could build up to a combo.
Then we could build up to a fight.
Eventually that grew into a whole AI model.
That project is still going to this day.
High Performance Sport New Zealand published an article about what we are up to for the LA 2028 Olympic Games here.
If I pitched you what we are doing now, you would never sign it off. The risk would be far too big. The way we have got there though? Step by step.
I think starting small is critical as it gives you speed, clarity and flexibility.
Another example
This was a recent one from our business. We wanted a cashflow report that is automated. Now one way to do this is to think big:
Full integration to Xero
Full database
Live syncing
Currency adjustments
While that might be good, it takes ages.
Thinking small is what we did. We just made a simple 4 step-model of cash in the business.
Step 1 - Get your current cash balance
Step 2 - Calculate how much money is due in the next 30 days
Step 3 - Average your expenses over the last 12 months
Step 4 - Plug in your revenue forecast
This is an extremely small way to get started. It takes about 5 minutes to do this.
Then guess what? It speeds up your development immensely. You also pick up patterns about what you need in the final system.
This works for all clients. This will work for you
We work with some absolutely massive datasets. Businesses that are 10/10 on the complication scale. This method works for every single one of them.
So I’ll teach you how to think small.
1. Define the “Smallest Useful Version”
The goal isn’t to strip something down until it’s useless. It’s to cut it to the smallest version that still gives you a meaningful signal.
Ask yourself:
If I had to deliver something in 48 hours, what would it be?
What’s the minimum set of inputs I need to get a real answer or reaction?
Could I solve 80% of the problem with 20% of the effort?
Example: Instead of a fully integrated CRM overhaul, start with the data you do have and make a customer journey for one subset of customers.
2. Focus on the Question, Not the Tool
Big builds often start with the tool — “We’ll integrate into X” or “We’ll roll it out in Y system.” That locks you into complexity before you’ve proved the value.
Instead, focus on the question you’re trying to answer.
Example: Don’t ask “How do we build a market analysis dashboard?”
Ask “What do we actually need to know about our market this quarter?” Then build a system to answer just that.
3. Use AI to Remove Friction
The main reason people avoided starting small before was that prototypes still took too long to make. AI removes that excuse.
Drafts, mock-ups, and summaries can be ready in hours.
Data can be cleaned and analysed without touching a formula.
You can create something presentable enough to share without a designer or developer.
If you aren’t a developer then give Replit or Lovable a go.
4. Test in a Safe Slice
Don’t unleash a small version everywhere. Pick a slice where feedback is quick and the stakes are low.
Example: If you’re rethinking onboarding, test your stripped-back process with one department or one new hire group before you roll it across the whole organisation.
Everyone usually jumps to these massive complicated roll out plans. That’s thinking big. We want to think small. Usually the first person you show something to finds the same thing the next 10 people do and they usually find the same things that 100 people find.
5. Measure What Matters Now
A small version isn’t worth anything if you don’t measure the right signals. Decide upfront:
What question is this version trying to answer?
What’s the simplest way to measure that?
Example: If you’re trialling a lightweight reporting tool, the metric might be “Does this reduce the time to prepare the weekly update by at least half?” not “Is it fully automated yet?”
6. So What?
My big thing I challenge you on is jumping to the end.
Let’s imagine you have your magic system. What’s next? What changes about the world? What would you then do? What does it unlock?
Why This Wins Today
Before AI, a small-first approach was mostly a “lean startup” thing — nice for entrepreneurs, but often seen as unprofessional in big organisations.
I think AI has fundamentally changed feedback loops forever. The game is now speed of feedback and getting as many rounds as you can.
The new saying we have at Cub is
“Make it exist, then make it good”
Hint, it should be existing quickly. So make a good small thing first. Then make a good big thing.
Here is an awesome $4 book on Kindle to read by a former employee at Google.
I’ll help you build something
If you are a paid subscriber or a current client and have an idea that you want to make, then send me an email. chris@cubdigital.co.nz I’ll help you build a prototype and even set up a little website. Just be aware, it’s going to be small.
Further Reading
It’s how ChatGPT works using the metaphor of a post office - read it here.
Podcast Version
I’ve started making audio books for sleep. But to be honest they are actually pretty interesting and I listen to them when I’m awake.
Brilliant