The Next Step In Optimization – Iteration Focusing! Part I

Add bookmark

Dan Morris

shutterstock_212278543One of the key advances in BPM is the concept of design iteration, says PEX Network Columnist Dan Morris in the first instalment of this two-part blog on the future of optimization.

It presupposes that any single design can be improved – and experience has proven that to be right. In BPM the new business operating design is improved through multiple "iterations" with each iteration being improved to create the next one. This is continued until an "optimal" solution is reached.

The issue, however, is in the way each iteration is improved. Commonly accepted practice says that if we use Lean, we are going to be OK. But practical experience says that Lean, while helpful, will only take you so far. It does help you find the obvious places for improvement and it does provide structure. Both good things and worth the time to do. However, the problem is that in each iteration you don’t really know if you have actually improved the operation. It is an "educated guess". The fact is that you really don’t know what the design may hide in terms of choke points, break points, decision selection and work volumes, and what happens with increased workload.


What we have been doing with design iteration and Lean streamlining has given us a lot better results than we had before we integrated them into our approach to business change. However, now that we have reached this point in our change approach maturity, we should consider ways to continue improving our business redesign approach.

So, what is new?

The problem is that any business outcome target can be achieved in a hundred different ways. The options however, are not limitless - they are limited by a few things. These include:

· The creativity of the people on the team

· Resource availability

· IT constraints

· Corporate and business unit cultures

· The willingness of management to change

Even taken together these limiting factors still allow a lot of room for work and workflow design options – all of which may be able to deliver the target benefits. So what is wrong?

The design problems are:

1. You won’t know if the workflow is better or worse until you actually use it

2. You won’t know for certain if the design will deliver target goals until you build it

3. Even given constraints, there is an ability to create an almost endless set of different business designs – many of which can actually cause harm

4. Uncontrolled iteration in the construction of the solution components can change the actual design and workflow in sections of the business operation – creating overall operational inefficiency

While some business designs are better than others, it is tough to anticipate which really help the effectiveness of the business operation and its efficient operation and which will be best accepted by the customer. For many companies this is still a guess. The really good news is that most companies now try to consider all these aspects in their new designs – not just how much cost can we wring out of the operation sponge. That difference actually represents a significant step forward in business redesign – cost reduction is important but not as important as business/revenue growth and customer satisfaction. We lost sight of that for about 10 years, but many have rediscovered that truth.

But back to the central issue – how can we know which new design is best? The real answer is that we need to try each one, but we need to do that without having to build the solution. And that can be done with simulation modeling and automated simulation tied to the BPMS and modeling tools.

These tools have also evolved in their ability to deal with complexity and in their ease of use. Many Business Process Management tools and a few specialty software tool vendors now offer advanced simulation capabilities. These capabilities have seldom been taken advantage of in business operation improvement solution modeling in the past. But, it is the key to the next step in our business redesign evolution.

Simulation – You need to try this!

The problem with iteration modeling is "how do you figure out what to change in each iteration?" How do you know where to focus your changes to improve a design? For me, the answer is simulation. This will show you the flow, the bottlenecks, the holes and more. It will also show you what happens under load, where you can reduce support, and where you will need to add support. In addition, it will allow you to anticipate if a change will likely provide the improvement you are looking for.

Note: The simulation tools today also provide visual feedback on the efficiency of the solution and the potential problems with the design – in some tools you can actually watch the data as it moves through the models. Also, you can use real data for current state modeling and then in the iteration model checks.

Having talked to several more technical colleagues, I find that there are any number of excuses that will come up for not trying simulation. In my opinion all of them represent "old thinking". In a previous column I talked about the problem of closed minded people in our companies. They are there and they really do need to be dealt with. Adding simulation to business design iteration is just one area that should be addressed, but it is an important one.

The fact is that it takes very little effort or cost to try adding simulation to a project. I strongly suggest you try it. I predict that you’ll find it much more than helpful. Does it take extra time? Not really if you collect the time, volume, and other needed information during your "current state" or "As Is" discovery work. The actual information that is collected should be defined by the simulation tool you choose. If simulation modeling is not imbedded in your BPMS tool, try an inexpensive tool like ProModel. There are several to choose from. So cost shouldn’t be the issue. And if the information items are simply added to the list of information needed for each activity in your workflow models, there will be little overhead in obtaining the information.

Start with the "As Is" modeling to establish the simulation baseline. This will be used to judge operational improvement from iteration to iteration. Each iteration will first be compared to the best iteration (based on its simulation score) and then if better, it will be compared against the "current state" or "As Is" baseline model. The result will be an optimized business workflow when several attempts (new iterations) fail to beat the "best iteration" as identified through simulation.

The creation of this current state baseline for simulation will actually be the hardest part. Few business operations have information on volume, time, level of effort, or decision probability. However, a count for a week or two with estimates of peak and low time differences will be a good starting point. While these numbers may not be right on target, they will be at least at the 80% on target point. They can also be improved if measurement continues during the discovery and redesign phases of the project. But, let me caution you here. These initial metrics will feed the "current state" model. They will also be a learning exercise for everyone. During this learning time, do not go overboard and measure everything. Follow the data needed by the simulation tool and limit how you will use it. Later you will know more and you can expand use in future projects. However, if the initial approach is draconian and turns everyone off, you will be dead in the water.

Once the information has been consolidated in the BPMS tool or business modeler and run through the simulation tool, you can see what really happens in the workflow. This initial use may take a few iterations itself and you may need to get help from the vendor or consultants to get this going. It should be noted that these initial models will not be accurate. They are a starting point. However, these initial models will evolve and the business managers will become comfortable with the measurements. This will give you a baseline to measure all improvement against. It will also get business and IT managers used to looking at productivity and performance metrics – and most importantly, it will give you a set of volume and performance measurements that managers and staff will accept.

Note: While creating these workflow models and collecting the information needed to drive the simulations, you will have an opportunity to engage managers and staff at all levels and gain concurrence on the approach and the data. This is the foundation for a general "buy-in" and an agreement on the measures. By including these managers and staff, collaboratively, in the evolution of these models and data, you will also be setting the foundation for agreement on the approach, the simulations, and the way the information will be used.

To read the second instalment of this two-part blog, click here