Learning Cycles

Many projects we work on involve bleeding edge innovation. The most common push-back to planning these projects is that, “Innovation can’t be managed.” The proponents of this notion believe that, “Ideas happen when they happen, and that it is impossible to program these out over time with any reasonable degree of confidence.” Therefor, it is a “waste of time” to plan.

The problem with this logic is that these projects have investors or shareholders that expect some degree of predictability about when they will see a return on their investment. Not planning is not an option in the real world. Further, we reject the view that innovation can’t be planned. We know this to be false because we have worked on many projects around the world where we did in fact program innovation and accelerated market entry of cutting edge technology. It can be done.

We have even achieved breakthrough cycle time on advanced semiconductor node development (i.e. <=20nm) where the manufacturing process was being developed while basic materials research was still being done. Simultaneously, tools (semiconductor manufacturing equipment) were being developed and a customer’s product was being designed using the new manufacturing process. A new manufacturing process, new tools, new product design, new design software (PDKs), and a new manufacturing facility was being started-up all at the same time. Even in this environment, we were able to plan (predict) and then manage the project to an accelerated pace.

Smith and Reinertsen discussed this idea in their book “Developing products in half the time.” They point out that 90% of “innovation” in projects is known and most of the time, less that 10% of the project involves real “new” innovation. The trick is to isolate this 10%, get it focused and properly resourced with the right skills, and to make sure it has a wide window of time; while you manage the remaining 90% using conventional methods. We also have observed teams becoming transfixed on the 10%, and wait for the solution; and when it comes, they end up behind schedule because they failed to get the easy part (i.e. the other 90%) done on time.

Learning Cycles

Innovation can be planning using the Learning Cycle concept. I critical component to innovation is learning from failure; the faster the failure, the faster the learning. You have heard people say, “Fail fast,” and this is the idea in rapid innovation projects. So how does one plan what they don’t know?


We can break a learning cycle down into some generic steps; plan, do, check, act (above). How do we know how long each of these steps will take if we don’t know the results of the experiments we have yet to perform


We’ve used this basic metric to assign a time value to a typical learning cycle. We first guess at how many learning cycles it will take to find the solution. Most professionals can be within an order of magnitude accuracy when making this guess. This metric is a way of quantifying the ranges. Most learning cycle problems can be categorized in three buckets; easy, hard, and difficult.

Use your own terminology if you like. But the concept is to define the difficulty levels of the problem you are trying to solve and then quantify the number of learning cycles needed to solve the problem. You can also assign an average duration per learning cycle. Most teams are able to categorize their problem, even though they don’t know at the time how they are going to solve it. Yes, this is a series of imperfect guesses, but then again this is all a plan is in the first place! Quantified guesses are better than having nothing but “hope.”


The final step is to estimate a learning cycle duration. In the example above, I've determined that my problem is “hard,” but on the “easy” side of hard, so I estimate 3 learning cycles are required. The overall duration is 45 days (assuming a 7 day work week, since my experiments run through a manufacturing line 24/7).

Refresh Planning

The only thing certain is uncertainty, especially in advanced development projects. We know it will change once we get the results of the first set of experiments in learning cycle #1. I leave learning cycle #2 and #3 as 15 day summary tasks and only breakdown what I know; learning cycle #1. Once I get through learning cycle #1 I can breakdown learning cycle #2 into more detail. I only breakdown what I know at the time. I may also not need LC #2 and #3 if the results from LC #1 are successful. This is also one way to accelerate an innovation project by optimizing the number of learning cycles.

It's amazing how many projects we see where the schedule has ZERO learning cycles built in. They expect it to be “right fist time.” They say, “This is our corporate policy…” We ask, “Have you ever done it right the first time?” They respond, “No.” Yet they plan this way because; a) adding the real learning cycles will cause a large slip in the schedule, and b) if they plan for multiple learning cycles then this somehow becomes a self-fulling prophecy. The reality is that hard problems are rarely solved “right the first time,” and planning for learning cycles don’t cause them to happen.

By expressing the number of expected learning cycles you can show the “real” schedule gap and can use this knowledge to create near-term urgency and/or to make critical scoping decisions that could narrow project deliverables to meet the expected time frame.

Our schedules are Refreshed weekly, sometimes many times a week. The Refresh Planning process is the tool we use to continually refine the schedule. This is the mechanism to align the schedule with the reality of innovation progress and/or a lack of innovation progress. It is a way of approximating a rough time frame for accomplishing a breakthrough and it also gives you some idea of the trend; are we trending away or towards our target time frame for delivering the solution?

Innovation can be planned using this approach. It is not a perfect science, but it is roughly right. In our opinion it is a “cop-out” to not plan learning cycles. Most teams know if a problem is easy, hard, or difficult. The time it takes to solve a problem can be approximated and the Refresh Planning process is the way to keep the schedule and reality in synch. Innovation can be managed.