Sunday, March 19, 2006

Mary Poppendieck

I had the luxury of listening to Mary speak at SD West 2006 this week and then a follow-up coffee with her and Tom.

Her talk centered on Lean and Muda as they normally do. She left us with the usual jewels such a talk should. A couple of the stones I took away and put in my pouch of knowledge for later polishing were statements like:

Inventory in SW development is partially completed steps

Software deployment is akin to Manufacturing Setup

Inspections should be used to mistake-proof the process.

Adaptations around constraints in a process become embedded into the fabric of the process, and are hard to remove.

Chief Engineers are the necessary and sufficient factor in a successful undertaking.

These all fermented additional thoughts in my mind and each are probably worthy of an entry, but first, let’s associate these to Portfolio Management:

All organizations exist for a single reason – to create profit. Therefore, there is only one real reason to initiate a project: to create some value add for the organization itself. This in turn can only be done if something is completed and deployed. Think of a manufacturing plant. Eli Goldratt has made a life’s study out of the optimization of product production (we will certainly talk about him at some point, but in the meantime, all you who have not read The Goal, go do so). Optimization has more to do with merely making things quickly and getting them to market (Goldratt would say that that was a necessary but not sufficient factor). Please accept that as truth for now. In Lean, Mary’s favorite topic, you try hard to minimize anything that can be viewed as waste, or Muda. In a manufacturing plant, a big source of waste is the setup time necessary to changing components of the machinery necessary to make the next critical part. There has been much writing and more proof that smaller batches, though non-intuitive, optimize throughput. Throughput is profit. Therefore it is reasonably easy to show that it follows that minimizing setup times helps to maximize throughput and profit.

Ok, keep following. What is the key to recognizing value from projects? You are right! Getting them deployed. Therefore, by minimizing the deployment effort, or at least streamlining it, you can increase your throughputs and your profits. Think about a SW Product house. If you create defect free systems every week with more and newer features that customers are going to love, but you can’t get them deployed, you can’t get compensed and you can generate value or profit. Your throughput is stalled cause your product is sitting somewhere, lonely and unused.

In Manufacturing, inventory is the stuff that is piled up in front of the very important machine waiting for the technicians to finish the Setup Process. Once that is complete (and verified), they will be able to begin processing the built up backlog and create throughput. Going back to our projects, all those releases stuck awaiting deployment, are inventory. Stale inventory at that.


Post a Comment

<< Home