And thanks to Peter Grillet, we’ve got a TL;DR version!

Michael's business worked because:

Ask yourself

How serious is your problem?
How specifically are you solving the problem for first?
Solve a problem people need to solve regularly
How easy are your customers to find?
Does your MVP actually solve the problem?
- Get users on your product early (you are not an artist)
Who are the most desperate customers, sell to them first. If it takes 10 months, they aren't desperate
Who's business is going to go out of business, without your software?
Be weary of 'customers' who are taking the piss and avoid them
Be cautious with your discounts. Use discounts for urgency but don't devalue your product / service

Metrics

Super important: ensure your stats are part of the build process
Should be a sources of ideas for features and solutions
Google Analytics is not optimal, you need an events based analytics solution (Mixpanel...) as well
Pick 5-6 simple stats to track, don't overload yourself
Ensure you are tracking if people are using product or not
Maintain a clear spec that you are building that all team members can refer to

Dev cycles

Keep them short
Have a single KPI that reflects how you are doing (Money if you charge / Usage if you don't)
Ensure everyone in the company knows what the KPI is and was
Brainstorm solutions (with metrics to support / destroy ideas)
Categorise: New Features & Optimisations / Bug Features / Tests
Prioritise: Easy / Medium / Hard - Decide
which hard item with impact the KPI the most - Which Medium item... - Which Easy item...
Create the spec for each item
Distribute tasks
Meet once a week or bi-weekly, enough time to get shit done

Pivot or iterate

Give your product time be properly validated
Pivot: Changing problem or Customer
Iterate: Changing the solution
Identifying the problem is the genius
Don't be fake Steve Jobs; iterate and talk to customers
Ask a specific customer what they want and make it (if it makes sense for your KPI)