Created on 2017-12-17 23:18
Published on 2017-12-18 22:59
Do an internet search for "engineering productivity measures". Unfortunately, you'll get a mixed bag of speculation, naysayers, some complex theories, and not much else. The world is full of engineering work, and it cannot be measured properly?
For the past 15 years, I led a global team that successfully developed engineering productivity metrics for automotive projects in seating, interiors, fabrics, electronics, and batteries. More recently, we did the same for HVAC controllers and fire-suppression sprinklers. A total of 14 were developed. 11 went beyond engineering and included resources for program management, advanced quality, finance, advance manufacturing engineering, purchasing, sales, and launch managers.
Thorndike was right and the benefits exceeded expectations. We could accurately predict future projects and avoid the perils of under-estimating and over-estimating (yes, both are bad). We quantified our performance gaps and set reasonable targets to achieve benchmark performance (like stepping on a scale in public). We stopped arguing about engineering cost and balanced our project portfolios to avoid over-stuffing the pipeline and overloading people. Throughput, gates-on-time, and quality all improved at the same time. Our models were declared a "small miracle" by our CEO for finally de-mystifying engineering costs.
Accuracy was statistically validated against data from completed projects and based on the core concepts of complexity, newness, and design responsibility. The concepts applied as well to software development and mechanical engineering. Below is a typical result showing a 85% correlation between engineering equivalent units (EQUs) and Hours spent on projects.
I love this song by Johnny Cash, but it's only about US cities. I led the development of a sophisticated Excel workbook that calculated the size of programs and translated this into detailed budgets for 10 different product groups spread across more than 50 technical centers around the world (see map below).
We created quick budgets for more than 2,000 projects (with multiple iterations and scope changes) and more than $4,000,000,000 in engineering spend with 87% accuracy. That's right, 4 billion dollars!
As the business world changed, the models were adjusted to account for new realities. Adjustment factors were developed and validated to account for global projects, developing countries with lower performance requirements, and customers or suppliers doing design work. All adjustments were based on data and validated to improve correlation with real-world results and reduce noise. For example, global projects are more complicated and require more coordination, but how much more effort is required? We were told to add at least 50% for global projects based on some extreme cases. Instead, we identified all the projects in each region that were global-leads and global-followers. Data from some projects were rejected because they were part of a steep learning curve and not suitable for future estimates. The remaining data showed that global-leads needed 10% more hours and global-followers needed 5% more hours. When the model was adjusted for past projects, our correlation improved!
Equivalent Units (EQUs) provide a top-down, holistic model for sets of projects. It is not a single project, bottom-up method like work breakdown structures (WBS) and earned value management (EVM) because it does not deal with work packages. In the Agile world, EQUs combine the best elements of Story Points which are quick & easy for a single development team and Function Points which can be compared across many teams. EQU's are intentionally standardized across development teams to allow direct comparison.
Please connect with me if you'd like more information at: firstname.lastname@example.org
Go to the top of the page to Share, Comment, or Like if you found this interesting.
This is part of a series of articles about applying continuous improvement processes to measure and improve the performance of product development projects. Here is the complete list in recommended reading order.
Copyright 2018 Richard Crayne