At the Financial Times’ last count, less than half of Northern Ireland’s new businesses survive until their third birthday. This is hardly a problem isolated to our own context; starting your own enterprise is a notoriously risky endeavour, at least from a statistical point of view. Taking inspiration from Eric Reis’ book The Lean Startup, Niall Crozier asks could 5 steps help?

The Lean Startup by Eric Reis is based on the premise that without proof from ‘real’ customers that business ideas have genuine potential, founders risk building a new organisation around a product or service which nobody wants or will pay enough for.

Reis pitches his approach to starting a new business as an alternative to ‘just doing it’ on the one hand and traditional, inflexible business planning on the other. It has broad applicability – he defines ‘startup’ as any human institution designed to create a new product or service under conditions of extreme uncertainty.

I don’t have time to read it – what are the book’s main messages?

Ries proposes:

1. Methodically identifying the assumptions founders have about:

a. The best offerings to present to consumers (Value Hypotheses), and

b. The best ways to scale the business (Growth Hypotheses).

2. Testing those assumptions in an experiment with real customers in a real context. This is done by investing as little resources as possible – creating a Minimum Viable Product  ( MVP).

3. Capturing the results of these experiments (‘validated learning’) via a method called ‘innovation accounting’ which:

a. Determines whether the assumption was accurate;

b. Records any implications for changes to how the organisation should operate, and

c. Indicates the highest priority hypothesis to test next.

4. Pivoting or persevering based on the outcomes of those ‘build-measure-learn’ experiments. The pace at which this testing loop is repeatedly run increases the chances of uncovering the changes required in order to build a start-up’s vision into a sustainable business which has ‘product-market fit’ (where a widespread set of customers resonates with the product).

5. Structuring the organisation around this approach, to accelerate the rate at which the ‘build-measure-learn’ loop is run, and to scale beyond ‘early adopters’ into more mainstream market segments.

1.  What does the identifying assumptions stage look like?

Before investing huge amounts of time and money on developing an incredible product or service for your target customers, it makes sense to first have proof of what those customers regard as valuable, and whether new users will discover and buy the product.

This helps prioritise which features of the product or service to improve first, how to improve them, and which customers are likely to be most valuable (and therefore whose insights are most worth listening to). The starting point becomes identifying the highest priority/most risk ‘unknowns’ that need to be investigated, and the experiments needed to generate those results.

Creating ‘archetypes’ – pen portraits of typical customer profiles – allow whole customer segments to be humanised. These iterate over time as experiments, in-person observations and other research provide more insights.

Not all assumptions require testing – ‘analogues’, where the same hypotheses have been proved in other contexts, may be enough. When creating the iPod, the assumption that people would use headphones to listen to music in a public place had already been proven – by Sony’s Walkman. On the other hand, Napster provided an ‘anti log’ – proof that people would download music, but not pay for it – hence that assumption being tested by Apple.

IDEO, the agency famous for pioneering design thinking summarised the crucial hypotheses that need to be proved as the intersection of three things – desirability (‘Do they want it?’), feasibility (‘Can we build it?’) and viability (‘Is it worth it?’)

In the excitement of the challenge of building something impressive (‘feasibility’), there is a danger of skipping ‘desirability’ and ‘viability’; success is not building something, but learning how to solve the customer’s problem in a sustainable way.

Finally, if you begin with no initial hypothesis, there is a danger of declaring success retroactively; no targets were set so none were missed, but ultimately the end result is failure.

2. What’s a minimum viable product (MVP)?

The MVP is the lowest cost prototype required to test the core underlying hypotheses that the startup relies on. Its aim is to rapidly establish whether the product should be built as planned and whether a sustainable business could be built around it. Unlike market research, it is ‘live’, collecting ‘real’ data about ‘real’ customer behaviour when they are presented with this initial iteration of the product.

The MVP includes everything customers experience as part of their interaction with the company (e.g. website, professional services). It is aimed at visionary ‘early adopters’, not the mainstream. Consequently, it has the fewest number of features needed to appeal to early adopters – they want it most, are happiest to give feedback and are most forgiving of mistakes. By the time the later versions are launched, the product already has a backbone of real, established customers, who are bought into its success.

MVPs take a variety of forms. These can include a ‘smokescreen’ of promotional materials offering unbuilt products/features for sale, a ‘concierge’ approach where the smallest possible number of customers is served before expanding, or a ‘Wizard of Oz’ approach, where what seems like a sophisticated technology solution is initially just manual work ‘behind the curtain’.

3. What is ‘innovation accounting’ for?

These are clear, concise and credible metrics which record ‘validated learning’. They indicate the extent to which the startup has succeeded or failed in proving hypotheses. These metrics demonstrate increasing understanding about which aspects of the offering are (and are not) helping realise its vision.

Metrics of this nature also drive accountability, and are a way of ascertaining progress toward a sustainable business at a time when revenues may be low, and expectations might otherwise be vague.

Innovation accounting is made up of a range of methods, including:

A/B split testing of different versions of the offering, to uncover the improvements which matter most to customers,

Funnel metrics, such as conversion rates from one stage to another (e.g. trial to purchase),

Cohort analysis, independently analysing on their own merits each set of customers following a new version of the product, using a (percentage) cumulative flow diagram,

Net promoter score, a customer insight measure not subject to short term fluctuations, and even

Revenue, charging customers from Day 1, even with a low quality MVP – albeit with low targets – immediately drives accountability, as opposed to a target of zero which allows the imagination to run wild, while also giving more leeway.

Ultimately, as Shawn Carolan said, ‘Startups don’t starve, they drown’. These metrics exist to ensure that time, money, and energy are focused on the most crucial issues.

4. Why ‘pivot’ rather than ‘persevere’?

Iterative ‘tuning’ of a startup has its limitations. A pivot should come at the point where momentum is being lost, experiments are decreasing effectiveness, and improvements are delivering diminishing returns.

A pivot implies having one foot rooted in what has been learnt so far, while changing direction completely with the other. This is particularly difficult when experiencing moderate success; a regular meeting to consider it, attended by outside influences, can be helpful.

A customer segment pivot away from early adopters – of whom there are a limited number – toward the mainstream is often required post-MVP. This generally requires a move to increased quality rather than speed, with major overhauling required rather than fine tuning, testing head to head against the original.

Other types of pivots include:

‘Zoom in’ (where part of a product becomes the whole), or ‘zoom out’ (the reverse)

Customer or channel segment (e.g. from B2B to B2C)

Customer need pivot (solving a different problem for existing customers)

Platform pivot (e.g. from selling an app, to being a vehicle for third parties to sell apps)

Technology pivot (solving the same problem for the same customers in a different way)

Business architecture (from high margin/low volume to low margin/high volume, or the reverse)

Engine of growth and revenue model (see below).

5. What implications does this approach have for structure?

Ries touches on a range of topics which relate to adopting this approach:

– Building cross-functional teams who are accountable for big-picture ‘learning milestones’ rather than siloed into their own specialist world,

– Using Kanban, a lean tool with:

  • capacity constraints to ensure the most important things are focused on, and
  • a set of simple process steps to ensure validated learning is evidenced for every change to the offering (e.g. a split test to show change in behaviour)

– Delivering improvements in ‘small batches’, adopting the continuous discipline of getting products in front of customers as early as possible in order to spot serious errors earlier, iterate based on customer insight earlier, and receive revenue earlier

– Getting clear on the type of growth engine being used, and how to increase its rate of improvement most effectively. Sustainable growth – not one-off bursts – come from these feedback loops, whereby new customers come from the actions of past or existing customers, by word of mouth, repeat purchase, advertising funded by previous revenue etc.

  • A ‘sticky model’ is focused on ensuring the acquisition rate remains higher than the churn rate (the fraction of customers in a time period who fail to remain engaged).
  • In a ‘viral’ model customers do the vast majority of the marketing as a necessary consequence of product usage, not just as deliberate endorsement e.g. luxury goods, Paypal, Tupperware. The ‘viral co-efficient’ measures this model – how many new customers will use a product as a consequence of each new customer who signs up. Many viral products do not charge directly but rely on indirect sources of revenue such as advertising
  • A paid engine of growth focuses on ensuring the average customer LifeTime Value (LTV) grows more quickly than the average Cost Per Acquisition (CPA).

– Asking ‘5 whys’ (i.e. ‘why?’ repeatedly) to get to root causes of issues, and targeting proportional levels of investment to address them

– Recognising that where a lean startup approach is being taken as a subset of a wider organisational, a ‘sandbox’ approach is required, holding entrepreneurial managers – with a personal stake in outcome – accountable to independently use scarce but secure resources to experiment with a limited percentage of the parent’s customers/offering for a set time period

– Matching talent and management approaches to modes of work, recognising that the entrepreneurial flair required to build initial innovations fundamentally differs from the more operational aspects of scaling. This portfolio approach allows team members to follow the same product as it enters a scaling mode, or stay in the same mode while pursuing the creation of a new product.

What case studies are used as examples?

– Intuit’s online offering Snaptax, which – protected from the core business – focused on the simplest customers first, and in one state, then rolled out more broadly;

Facebook, which convinced investors to back them long before they had large revenues by demonstrating their proven value hypothesis (metrics to show time spent on site each day) and growth hypotheses (the rate at taking over other college campuses and rates of customer acquisition driven by existing users for free);

Groupon, which launched with an MVP built on a WordPress blog, manual pdfs and apple mail, to prove people wanted it, and

Wealthfront, an automated investment vehicle, which began as kaChing, a fantasy league for amateur investors. It started with customers unable to participate in the mainstream market, but then pivoted to use that access to deliver an entirely different kind of value.

 

Final reflections?

Regardless of the context, clarifying the key unproven assumptions before investing your time, effort or money heavily is a really helpful discipline!