153. The Pursuit of Truth to Master Complexity #3 - Measure!

In a hyper hyper-competitive market of digital transformation in mainland China, especially in Financial Services industry, there was not a single conversation that I had with agile coaches, where they don't talk about the importance of measurements and metrics.

At the same time I often heard the term Involution, or “Nei Juan”, "laying flat", among IT professionals, a term first introduced by Anthropologist Clifford Geertz, where he analysed overpopulated peasants overworked their land for little additional gain in Java Island, Indonesia. Although peasants invested a significant amount of time and energy, creating very sophisticated rice terraces, the field could only produce a few more grains and support a few more residents due to its finite potential.

In economic terms, the marginal cost put into the production was much higher than the marginal profit, and such investment was a waste. But even so, the baseline of productivity and habit of measuring performance is quite remarkable, in this article, I will briefly summarize few takeaways from the TiD event from 12th to 16th August.

You can't manage what you can't measure. - Peter Drucker

No.1 Flow Predictability combined with Demand Rate

Every month, on enterprise level, each product needs to provide their flow predictability (aka success rate of planned deliverables) as well as the demand change rate, for each product that is not able to meet their goals, identify improvement areas and provide support, while at the same time, identify trends of how often demands are raised, where they are raised and how quickly can they be processed.

No.2 Measure Cycle Time to Identify Wastes

People ask themselves the following 2 questions:

• How does the demand throughput and delivery throughput change over time?
• And which part of the process has the highest delay?

In order to identify the phase that needs the most improvements, especially when it comes to automated testing, preparation of test cases and integration test.

No.3 Flow Distribution and Change Rate of Demand

How often does demand change and what is the rate of rework? How are the work distributed among bugs, change requests, technical debt and new functionalities, a classic example according to Lean approach, guidance provided by Scaled Agile Framework, such as how complete is the work, so that the next step can proceed?

No.4 It is the dawn of AIOps to accelerate Performance Test

Statistics from a foreign e-commerce headquarters show that for every 1 second latency in the response time to customer access, the website's sales will decrease by US$1.6 billion a year. For every 100ms increase in the opening time of the first page, the website's sales will decrease by 1%.

It is very hard to design the test, with high quality standard, not to mention lack of experience and hard to identify abnormalities. In the case study, presenter has shown the penetration test result for million per second transaction such as Chinese New Year, apply user complete sequence and high frequency subsequence, personas cluster analysis to detect potential risks, and improve performance drastically thanks to AIOps models.

No.5 accelerated Infrastructure as Code use cases

AWS Expert of Cloud formation has given a set of best practices across suppliers, business scenarios, business cases. Compared to Germany or US, the adoption of IaaS in China is much more broad than PaaS, due to the following reasons:

• limited capabilities in managing and developing cloud technologies;
• lack of expertise and cloud computing migration experience;
• Businesses want to build custom applications from scratch rather than using external applications;
• In order to avoid being locked in by suppliers and achieve quick profits, enterprises would rather sacrifice PaaS services;
• constrained by time and migration budgets;
• the business ecosystem, especially in banks, are dominated by the 5 biggest banks and 12 private banks, who have abundant amount of resources to be invested

No.6 how to avoid misinterpretation on metrics to avoid blaming and partial truth?

Eventually, agile ways of working improve productivity and team engagement, but how to identify the correlation between different factors. There is a very clear focus on balanced measurement on a regular basis per sprint, to get a holistic view about how the team is going, we use the PI Planning simulation to diagnose and guide teams how to improve their iterations, across quality, value, team, efficiency, product looking into customer's NPS, self-assessment and real data from tooling.

To find more about measurement, pls. check out my previous blogs here:

Light side of Measurement

Dark side of Measurement

Source: The WIRe, TiD Conference, scaledagileframework.com