Sunday, September 8, 2013
Lean Startup principles and mobile app development
Towards the end of my post “APIs to leverage Metcalfe’s law” I mentioned the Lean Startup principles evangelised by Eric Ries. I find those principles very useful — not just for launching a startup but also for mobile app or software development, project management, general work practice and personal efficiency. Thus, I wanted to summarise these principles here.
Eric Ries is serial entrepreneur and coined the Lean Startup concept. This concept is basically around two ideas: eliminate waste wherever possible and be as close to your customer as soon as possible. He describes a couple of methods how to achieve that: customer development, minimum viable products, the importance of measuring, split tests, and continuous deployment.
There are some substantial differences to the classic ‘Silicon Valley’ approach which usually followed the pattern of convincing a venture capital firm to invest into the business with large sums, then develop the product behind the curtains and launch it with a big bang hoping that it generates revenues quickly – and a lot of it. So what is it that the Lean Startup principles suggest?
Many startups are technology driven. Ries suggests integrating customers right from the beginning. You need to know who your customers are and for which service they are actually willing to pay for, how they are willing to pay and how much. For this, you need to ‘get out of the building’ (i.e. gemba) and collect facts. Steve Blank (Professor at Berkeley University) wrote a lot about customer development (see his book or his blog).
The image below gives an overview over the customer development process and some pointers for further reading.
(Source: Steve Blank’s blog)
Minimum Viable Products (MVP)
Ries suggests going to the market as early as possible in order to get customer feedback as early as possible and by this let your product grow gradually but by exactly meeting your customers’ demands. This bears the risk that you lose customers but the ones you gain later on are more likely to be more loyal because the product has evolved directly addressing validated customer needs. An MVP to test the market can be as simple as a mock-up website or a video that introduces a potential service. This, in fact, was the approach that dropbox took.
Related to MVP is the idea of many and short iterations, with every iteration producing a functional product – or a subset at least. Ries refers to this as the build—measure—learn cycle (see below).
Measure – measure – measure
Ries uses the notion of vanity metrics. These are the classical and standard metrics used to assess progress. The problem with these is that most of them are not useful for the very business at hand because they are not actionable for the concrete context of that business. Examples of vanity metrics are web traffic, revenue, or number of customers. They are too broad and too general.
An example for a better metric (because targeted, context specific and actionable) would be, for instance, after adding a new feature to your product or service, measure exactly how customers react to it and then draw your conclusions from it. This is easier achievable in some industries than in others. With web sites or mobile apps you can pretty much freely define what you want to measure and how frequently.
A very powerful method of collecting expressive data is measuring based on split tests (i.e. A/B experiments). That means that a new feature is initially only introduced to a subset of your customers. Then you can measure the reactions and compare the collected data with data from customers who did not get access to the new feature.
The origins of the Lean Startup principles
These ideas behind the Lean Startup principles are not new but enriched with further ideas and packaged in a meaningful and useful way.
Ries bases his concept on Lean Thinking borrowed from operations management — e.g., eliminate waste or “gemba” (=go out and see for yourself). Some of the principles are very similar to the Agile Manifesto (or Srcum as an example of an implementation of the Agile Manifesto), which in itself is based on lean operations management.
Another field that suggest an iterative and incremental approach is innovation management, especially the works of R. G. McGrath and I. MacMillan about Discovery-driven Planning. The challenge in innovation management is that a priori you have no idea if potentially huge investments in innovations eventually pay off. The authors suggest that every project should start small but define for every iteration the financial requirements and returns. Assumptions are crafted, tested with quick prototypes on the market and evaluated after each iteration.
Finally, the customer development aspect is mainly coming from Steve Blank’s works and evolved into the Lean Startup. In fact, Blank and Ries worked together a lot. Steve Blank BTW offers an interesting free online lecture about “How to build a startup” on Udacity.
My key learnings
A lot of business ideas fail because of lack of a market, i.e., customers. Peter Drucker said in his book “The practice of management” (1954) that the only purpose of business is to create a customer. It is paramount to find your customer archetypes and actual customers as soon as possible. Following the Lean Startup ideas forces you to do that early. It prevents you from wasting your time and encourages you to focus and if necessary pivot into a direction which based on experiments and validations seem to be more likely to lead to success.