We are used to seeing business plans that are many pages thick and have taken the startup team many weeks to build. That would probably be a good way to go if we knew exactly who the customer was, what they wanted and how we were going to build it. We would also know what our organization structure would look like and exactly how much money we need. Unfortunately in a startup we don’t really know any of these things with any certainty. A major problem with this planning approach, aside from over analysis, is that we tend not to want to change it once we have put it on paper. We are destined to ride this one right off the cliff if that is where it takes us.
We need a different way to plan our business that acknowledges what we know and don’t know and gives us direction but with the flexibility to adapt as we learn. This post will layout a new approach to planning that builds in active learning.
We have started by looking at jobs to be done and the people who do them. Our plan will have a model of who the person is, how they do the job now and how we believe they can improve on it. This is the beginning of the planning process. We will constantly add to this model as we learn more. We may also add new user profiles as we find other potential users over time. We should have a belief that there will eventually be enough users to build a sustainable business around.
We have also assessed whether there is any other provider of these services to determine if it is a disruptive innovation. If it is not then we need to look at different jobs to be done, different target users or abandon the venture. We need to be constantly aware of competitor responses and new introductions and adjust parts of the plan accordingly.
Out study of the potential users will also give us some insight into where they might purchase the product, where they might use it, what assistance or training they may require. This part of the plan helps us plan for the distribution channel and part of the organization structure (i.e. do we need people to train). This part will also evolve as we learn more.
Now we are ready to create a model of the business that will pull together all of the pieces we have assembled so far. We want to build a model in Excel that demonstrates what success will look like. How much revenue we need, what our margins will be, what our costs structure has to be, etc. It is essential that the whole team be involved in building this model since the degree of learning that falls out is significant. I am always interested in how much the team does not know about how the business works or needs to work until we put the model together. There are many instances where it is necessary to go back to the product, customer or channel and rethink it based on what the model is demonstrating.
The model is a representation of the business strategy and is the bridge that links our customer/product/channel/organization planning to the execution. We will use the assumptions in the model to develop the product in a way that let us test them out and evolve the product.
Once the model is constructed the team then has to separate facts from assumptions and then rank the assumptions based on how critical to the success they are. Having identified the key assumptions we will test them using a structured experimentation process. As in a scientific process the experiment will have a purpose (i.e. we want to test the assumption that users will download the product from our website), a hypothesis (a prediction taken from the model on the user acceptance rates) and a procedure which is what we will include in our launch product. We create features, A, B and C to determine if the users will find them useful and download them from our site.
In Lean Startup, Eric Ries identifies two types of hypotheses, a value hypothesis test whether the product or service actually delivers value that the customer is willing to pay for, and a growth hypothesis tests how new customers will discover the product. We will need to make sure we are moving in both directions by making the product more valuable and growing the user base.
We will also decide on how we will measure the results and include that in our experiment description. The plan will have a series of these experiments along with the results. The results may support or refute our hypothesis. If they support it we will then move on to subsequent assumptions creating experiments for each of them as we go. In each instance we will build only enough into the product to test the specific assumption we want to learn about. We will also be updating the earlier segments of the plan such as product and customer for what we have learned. The model should remain fairly stable at this point unless some of our learning significantly changes one or more of our assumptions.
If the results fall short of what we expected or demonstrate an unexpected turn they should be followed up with users to understand why. This is a deeper learning tool that will help us to get the product back onto a user acceptance path. What we learn from these investigations will inform the earlier parts of the plan.
The company could reach a point where there is no significant traction from the product as it is being delivered. The company then has to revisit whether there needs to be a shift in the strategy by making changes to the major assumptions and redrafting the model or whether they should abandon the project. The results from the testing and discussions with potential users will provide direction for these decisions. We will discuss these types of decisions in future posts.
We can see by looking at this approach that the planning phase was not some big investment of time up front but was built as the company progressed through strategy and execution. Also, by incorporating learning early it was possible for the company to identify missteps and correct them before they has exhausted their resources. If the business ended up not proceeding then this process will let the company fail early before they spend excessive time and resources trying to prop up something that isn’t working.
A reason why people use the old way of planning is that some investors expect to see things in that format. Using the learning approach will often mean fewer resources are needed up front and investors will have better insight into how the product is progressing before any revenue or profits start to accrue. If you find yourself being pushed into the old planning model by investors then it may suggest you have the wrong investors. Having the wrong investors for your business is something you want to avoid at all costs but that is a topic for another day.