There were a bunch of great speakers at Momentum’s Lean Startup Michigan Conference last week in Grand Rapids.
Here is a fairly random collection of the ideas I took away.
Eric, one of the gurus of Lean Startup, kicked off the conference with a great talk.
- Entrepreneurship is management, where the unit of measure is validated learning.
- The cycle is: build-measure-learn (then rinse and repeat). You want to minimize total time through this loop: ideas, build code, measure data, learn, repeat.
- A startup is a human institution designed to deliver a new product under conditions of extreme uncertainty. A startup is an experiment
- Planning and forecasts (ala business plans) work only when you have a long and stable history to draw upon.
- All successful startups have in common that they have pivoted – changed strategy – without changing the vision of where they wanted to go. Pivot = change direction but stay grounded in what we know.
- Reducing the time between pivots increases odds of a venture’s success, before it runs out of money. Runway should be measured by the number of pivots that the funds allow to be performed, not just the payroll/expense time.
- The waterfall method of design doesn’t work becasue – you can be on time, under budget, high quality, beautiful design – but still successfully be executing a fatally flawed plan.
- what is the minimum effort that can be expended to learn what the specification should be (what the customer wants) – NOT to execute the specification (that we don’t know yet is right).
- Eric suggests we do “innovation accounting”. 1. Establish the baseline, build mvp and measure how customers behave right now. 2. Tune the engine, experiment to see if we can improve metrics from the baseline to the ideal. 3. Pivot or persevere. When your experiments reach diminishing returns, it’s time to pivot.
- If you don’t know who the customer is, you don’t know what quality means.
Brant Cooper and Patrick Vlaskovits
My apologies to Brant and Patrick for boiling down a good talk into one priceless takeaway.
- The “Old Yeller Award”. Who can kill their bad concept the fastest? Take it out behind the barn and shoot that thing!
To make it up to them, let me urge you to go buy their book, The Entrepreneur´s Guide to Customer Development for Tech Startups – the Cliffs Notes for Four Steps to the Epiphany.
Rob, the author of Start Small, Stay Small: A Developer’s Guide to Launching a Startup, likes small, bootstrapped ventures. He knows this because he has taken his own advice:
- Know yourself.
Rob likes Strengths Finder 2.0, a book that will help you find out how you are wired.
- Tell your stories like this: feature x will affect metric y.
- Use metrics for weekly cohorts of users. Metris have got to be getting better – or something is wrong.
- If users activate but don’t come back to your site and actually use, then something is wrong – you are not solving a problem worthy of their coming back.
- Traction = activation and retention.
- Use activity streams: log, time stamp, and color code (cool idea!) the actions users are using in your app.
- Three kinds of code: 1. temporary, 2. sustainable, 3. half-assed (temporary code shipped, etc.).
- Temporary code is hard to throw away, but you have to throw it out and go make something proper.
- Code for quality and maintainability – automate your tests
- Try pair programming – 2 people, 1 computer – it sounds weird, but it is very useful. Hint – programming is more about thinking than typing.
Jeffrey Schox (patent attorney)
- Provisionals are not searchable – if you abandon it, it can’t be found.
- You can create a provisional patent application in a day for $110 – file early and file often.
- Too general a concept = not patentable. Too specific a product = not valuable. In the middle = the sweet spot.
- Minimum Viable Patent strategy: 1. Before seed: ($10K) file a provisional application and conduct a preliminary freedom to operate analysis on your main competitors. 2. Before Series A: file a full patent application and conduct a preliminary freedom to operate analysis on all relevant patents.