This is the second post in our blog series, Measuring the Impact of your Agile Investments. The series focuses on measuring the impact that Agile practices have on business outcomes.

For many leaders, increasing the productivity of the development organization (delivering features the business needs and generating improved ROI) is their primary and over-riding goal. ‘Doing more with less’ is a mantra they preach to their teams continuously, and which colors every decision they make.

Yet few have a clear and consistent definition of productivity or an effective means of measuring it.

Productivity: a relative measure of the efficiency of a system in converting inputs into useful outputs.

Productivity: the ratio of the real value of outputs to the combined input of labour and capital.

Productivity: a measure relating a quantity and quality of output to the inputs required to produce it.

So, in simplest terms:

Productivity = Valuable Outputs / Costly Inputs

Be Careful What you Wish For

While I am a ardent believer in the value of metrics, it is important to keep in mind the potential for unintended consequences. Individual and group behaviors will evolve to meet your measures – not necessarily your goals. So ensure that the measures you use promote the behaviors you desire.

Most documented productivity metrics in software development define a unit of output as either a KLOC (thousand lines of code) or as a Function Point. Both have significant disadvantages.

KLOCs are generally favored for the simple reason that they’re unambiguous and easy to measure. However, they can unintentionally promote poor design and coding practices. Elegant, efficient, maintainable software takes more time to write, and takes fewer lines of code. Therefore, a KLOC metric inadvertently creates a disincentive for building high-quality software, and rewards poorly thought through designs and shoddy craftsmanship.

While Function Points don’t necessarily promote poor design and craftsmanship, they have their own unique challenge – they are famously difficult and expensive to calculate and measure.

Perhaps even more importantly, both KLOCs and Function Points ignore the core aspect of ‘Value’ inherent in our definition of output – and therefore productivity.

It has been stated that 64% of the features and function in the typical software system are rarely or never used (Standish 2002). Calling the code that delivers these features and functions “productive” may be a mischaracterization.

Similarly, those features and functions that ARE commonly used are generally not of uniform value. The Pareto Principle would suggest that 80% of the value is delivered by 20% of the effort. Yet KLOC and Function Point-based metrics treat all features and functions (and code) delivered as interchangeable. This promotes focusing on the easiest and lowest technical risk; rather than the most valuable, most innovative and greatest business risk…

Output is a measure of the value delivered, not the effort expended.

A number of people in the Agile community have written about an alternative unit of output measure – Value Points.  While the simplicity and value focus of this model is appealing, it has challenges when scaled beyond a few teams. In order to be meaningful at the organizational level, you must normalize the relative value point scale across teams and programs – which can be difficult and expensive.

Also, the Value Point approach does not easily translate to initiatives and/or divisions that may not be delivering in an Agile manner. Having a common measure of output, and therefore productivity, is critical to measuring the impact of your Agile investment.

So, the approach we recommend is to associate a percentage of the total value of a given initiative to each Minimally Marketable Feature (MMF) or production release. These percentages can then be applied to a any monetary business justifications (ROI, NPV, Discounted Cashflow, etc.) to arrive at an expected dollar value realized from each release.

Hence,

Productivity = Total Value Realized (delivered to production) /  Total Cost of production (labor)

Using this approach, your organization (be it a Scrum team, a multi-team Agile program, a waterfall project, or even an entire product development group) base-lines their productivity for a period of time and monitors the change over time.

Example:

The division has 3 initiatives in progress:

• Initiative A:  Total Expected Value $5MM; being delivered by a Scrum team with an iteration run-rate of$70,000.
• Initiative B:  Total Expected Value of $25MM; being delivered by 5 Scrum teams with a combined iteration run-rate of$400,000.
• Initiative C:  Total Expected Value of $50MM; being delivered according to a project plan and resource matrix that charges$2.5MM to the project in the 1st quarter.

In Q1:

• Initiative A: Released to production monthly and delivered a total of 60% of Expected Value; or $3MM. 25% of the backlog has been burned-down in terms of story points. Total Cost$210,000. PRODUCTIVITY = 14.29
• Initiative B: Released to production quarterly and delivered a total of 65% of Expected Value; or $16.25MM. 35% of the backlog has been burned-down in terms of story points. Total Cost$1.2MM. PRODUCTIVITY = 13.54
• Initiative C: Completed requirements definition and is 50% done with Detailed Design, and delivered 0% of Total Expected Value. Total Cost \$1.2MM. PRODUCTIVITY = 0

*undelivered work is WIP, and therefore not yet productive.

Aggregate Division Productivity for Q1 = 7.38

As you can see, it would be relatively straightforward to predict Q2 productivity – at the initiative as well as division levels – by assessing the various product roadmaps and traditional project plans. Those projections could then be used to:

• drive discussion about trade-offs on where to allocate limited capacity and maximize productivity
• staff and fund initiatives where productive potential is high, and to cancel successful projects whose greatest productive potential has already been harvested
• inform intelligent business decisions – which is WHY we’re measure outcomes

By understanding the productivity of the development organization (its efficiency in delivering features the business needs and delivering improved ROI), leaders can effectively drive improvements and business decisions that maximize productivity – Doing More with Less.

The next topic I’ll address in the Measuring Impact series is Quality. Most organizations claim to have a Quality focus, yet few take a holistic view of Quality or appreciate the strong correlation between Quality and our other Outcome Metrics – Productivity, Responsiveness, Customer Satisfaction, Employee Satisfaction and Predictability.