Resources / Blogs
Resources / Articles

Articles

Understanding Lead Time Distribution, WIP Age and Inventory Age and How They Affect Delivery Performance

Understanding Lead Time Distribution, WIP Age and Inventory Age and How They Affect Delivery Performance

In the previous article in this series we explored the Top Symptoms of Delivery Underperformance.


In this article we will explore the first of a number of key indicators to help you diagnose problems in your delivery ecosystem.

Distribution of lead time, WIP Age and Inventory Age, including trend analysis

Understanding how long it takes for your delivery ecosystem to deliver value consistently is an excellent place to start your diagnosis,  especially if you have a clear idea of what success looks like.  Service Level Expectation (SLE) is a great place to start. If there is a difference between your Lead Time and your agreed SLEs this will usually mean your service is not fit-for-[customer]-purpose. 

When looking at Lead Time for diagnosis it is important to look at the lead time distribution. Every work item completed gives you one sampling of lead time. As you continue to complete more and more work, you start getting more and more samplings of lead time, usually with different durations. If you sort your sample from the shortest to the longest duration, you have your “lead time distribution”. This is an important metric as it represents your time to market or your time to value. 


By analysing your distribution, you can understand, what is the usual time it takes for you to deliver, or when things don’t go to plan, what your customer can expect. When analysing delivery performance, there are key points on the lead time distribution that help you understand how your system is behaving. They are:


Mode: The mode is the lead time in your distribution that occurs most frequently. 

Average: As in mathematics the average of the data set. If your distribution has low variability, averages can be used to forecast with confidence.

50th percentile: Same as the median. It’s the central number of your lead time distribution. If there’s a lot of variability in your distribution, then the 50th %ile better represents central tendency.

85th percentile: Represents the point of the distribution that 85% of work items fall within. When we say the 85 %ile of the lead time distribution for new features is 90 days, it means that 85% of all features completed in the past took 90 days or less. If the future mimics the past, you can have confidence that 85% of new features will also be completed within that time frame.

95th percentile: As per above, intended to catch the maximum lead time for almost any work item entering your system. This is often used to communicate service level expectations when the situation requires a higher level of confidence as the impact caused by delays are considerably high. 

Max: Represents the maximum representation of lead time in the distribution.

The lead time distribution helps you understand how predictable your system is, and also helps you to negotiate expectations with your customers.

In a practical example, let’s say your lead time distribution for delivering new product features is as follows:

Mode: 5 days

50th %ile: 16 days

Average: 37 days

85th %ile: 60 days

95th %ile: 80 days

Max: 145 days

The above example shows a system that is fairly unpredictable, as the range between the Mode and Average is 5 to 37 days, and with outliers taking between 80 and 145 days to transition through the system.

If you understood this and a stakeholder asked you how long it will take to deliver a new product feature, most people would likely use the 50th percentile, and say “it will probably take 16 days!”.   

As we saw, lead time is a distribution, not a single metric. When setting expectations around lead time, a good practice is always to use a range of dates associated with a probability level. In this example, “More than likely 16 to 37 days, but up to 60 days with 85% of confidence”.  

Lead time is a very important indicator but it is also a lagging indicator.  In order to anticipate problems, you should also look at the age of your work in process (WIP). We term it ‘Work in Process’ rather than ‘Work in Progress’, because more often than not, the work is in process, meaning it has entered your delivery system but is not actually progressing; what we consistently see in enterprise organisations, is that approximately 85% of the time work is stuck in queues. WIP Age is a leading indicator, which can help you predict your potential to meet your customer SLEs.

Inventory Age is the time work spends waiting to be worked on and is another essential leading indicator. We term it ‘inventory’ rather than ‘backlog’ to call attention to the fact that it comes with a cost. As in manufacturing, knowledge work inventory (backlog items), incurs a considerable cost in its development, and an additional cost related to maintenance (organising, grooming, sorting, triaging, re-understanding, etc.). This also has the tendency to increase as items age in your inventory, people lose context, the environment changes.  Additional work and cost are required to bring them back into relevance.  

This article is part of a series of articles on 

Diagnosing and Rectifying Delivery Underperformance.

If you are currently facing similar challenges, check out Flomatika. Our platform can help you achieve faster and more predictable value delivery!

Find out More

More

Articles