Prioritizing to Minimize Cost of Delay

WRITTEN BY Derek Huether

What is Cost of Delay?

Cost of Delay is a way of sharing and understanding the impact of time against forecasted outcomes. It provides the means to calculate and compare the cost of not completing something now, by choosing to do it later.

I was recently watching an episode of Shark Tank. I loved the unfiltered statement from Kevin O’Leary (Mr. Wonderful) toward an entrepreneur seeking an investor in his company.

I’m here to make money!

If you’re a fan of Shark Tank, you’ll notice something about Mr. Wonderful.  He keeps the conversation focused on the money.  When will he get his money back? How many multiples of his investment should he expect to get back? Other investors (and many of our stakeholders) don’t focus enough on the money.  Particularly, what is the cost of delaying the implementation of one feature over another.

Why is it so important to understand?

If you want to save or make the most money, you need to prioritize your backlog… by money.

One way you can prioritize work by money is by using cost of delay.

Sounds kind of weird, right?  Aren’t we always telling people to prioritize their backlogs by customer value?  When you ask customers, or the business, which features are the highest priority; all too often they say all of them.  (Jim Hayden makes reference to this in a recent podcast,when teams don’t prioritize or limit their work in process.  People are really good at starting things but not necessarily finishing them.)  Don’t just ask what is the most valuable. Ask the question, “what will cost us the most, by delaying its delivery?”

That’s really what we’re doing.

We’re not profiting from a feature that is not in production, so therefore, we are losing money every day it’s not there.

If I have 4 features to choose from, each with a different worth to the business and each taking a different amount of time to implement, how do I make the best economic decision on what to finish first?  I use Cost of Delay.

How can I quantify the Cost of Delay?

Step 1 – Compare Features

Let’s put the features in a table and compare them.     What we care about is the amount of time it will take to deliver each feature, and the overall value of that feature afterward.

The time to deliver a feature is measured in sprints, and for our purposes I’ll take that to mean 2 weeks.

The value of a feature is can mean an amount saved (for an efficiency improvement) or additional customer value received due to the feature being in production.  That value continues to be delivered for each week after the feature is released.  This number needs to take into account the business value, time criticality, risk reduction and opportunity enablement values, but what matters more is how they stack up against each other and the specific needs of your application.

In general, it’s a good idea to not spend too much time thinking about the exact value of each feature.  Instead, I use a simple comparative value, typically using the Fibonacci sequence to estimate (1, 2, 3, 5, 8, 13, …).

For our purposes here, I intentionally tried to keep this simple by multiplying the value of the features by $1,000.  Imagine what this would look like if your features were worth tens of thousands or hundreds of thousands of dollars?

Donald Reinertsen, in the book “The Principles of Product Development Flow”, suggests we take into account the value delivered when we prioritizem by calculating a “weighted value” defined as that value delivered divided by the time it will take to deliver that value.  We’ll calculate that and show it as well.

Feature Sprints to Deliver Value (Approx) Weeks to Deliver

(Sprintsx2)

Cash Value

(Value x $1000)

Weighted Value (V/T)
A 2 8 4 $8000 4
B 3 5 6 $5000 1.67
C 4 13 8 $13000 3.25
D 6 21 12 $21000 3.5

Step 2 – Visualize Scenarios

Taking what we have learned in the table of Step 1, let’s visualize different scenarios, showing when we could get a return on our investment, given a choice of priority.

  1. No priority at all. Do all at the same time.
  2. Complete the features that take the shortest amount of time first.
  3. Do the features that are the most valuable first.
  4. Weighted Shortest Job First (WSJF) – prioritize by the Weighted Value given above.

Remember, regardless of our choice of priority, all of the features are done by the 15th sprint.

In the chart below, we show how much value is being received while the features are being developed.

For every week features are not completed and making us money, they are costing us money.
Let’s do some math!

Step 3 – Priority Impact on Cost of Delay

Using the four features we can look at the financial impact of the four alternatives.

All at the same time (No Priority)

If we start all of the features at the same time, it slows down development of all the features.  Even if we deliver each when it is done, the first feature (likely Feature A) will be delivered around the 9thsprint, as can be seen in the first line above. When the first feature is delivered in week 9, we receive an ROI of $8000 each week, which increases until all features are delivered after week 15.

Since our Cost of Delay is the amount of value we have not realized due to features not delivered yet, the Cost of Delay is 47 for the first 8 sprints, 39 (47-8) for the next 3, 34 for the next 2, and 21 for the last 2, for a total Cost of Delay of (47*8) + (39*3) + (34*2) + (21*2) or 603.  If each sprint is 2 weeks and each point of value is $1,000, this is $1,206,000 in total Cost of Delay.

Do the Shortest Job First

When we prioritize based on shortest to longest length of time to complete a feature, it would only take us until our 3rdsprint to get our initial ROI ($8,000), until the 5th sprint we get our next ROI (+$5,000 for a total of $13,000), and so on.  For the 2 sprints we are working on Feature A we incur the Cost of Delay of all four features: $8000 + $5000 + $13,000 + $21,000 per week. This adds up to $47,000 per week times 4 weeks (2 sprints) giving us a total Delay Cost incurred so far of $188,000.

We then move on to developing Feature B. For the 3 sprints this takes us to deliver we incur the Cost of Delay of Features B, C and D: $5,000 + $13,000 + $21,000 per week = $39,000 per week. So the Delay Cost is an additional $234,000, bringing us to a total of $422,000 worth of Delay Cost incurred so far.   Feature C adds another $34,000 each week for 4 sprints, or an additional for $272,000 Delay Cost, and Feature D adds $21,000 each week for 6 sprints, or $252,000.  Thus the total delay cost is $188,000 + $234,000 + $272,000 + $252,000 or $946,000 Total Delay Cost. 

This is almost $300,000 in additional value received compared with all at once, but we’re not done yet.

Do Most Valuable First

If we prioritized based on most to least valuable feature, it would take us until our 7th sprint to get our initial ROI ($21,000), until the 11thsprint until we get our next ROI (+$13,000), and so on until we have all ROI after sprint 15.  For the 6 sprints we are working on Feature D we incur the Cost of Delay of all features, again $47,000. This adds up to $47,000 per week multiplied by 12 weeks giving us a total Delay Cost incurred so far of $564,000.

We then move on to developing Feature C. For the 4 sprints this takes us to deliver we incur the Cost of Delay of Feature C, as well as A and B which are also not yet delivered: $13,000 + $8,000 + $5000 per week = $26,000 per week multiplied by 8 weeks giving us a total Delay Cost of an additional $208,000, bringing us to a total of $772,000 worth of Delay Cost incurred so far.  Feature A adds 4 weeks (2 sprints) at $8,000 + $5,000, for a total of $52,000, and finally Feature B, our lowest value feature, adds 6 weeks at $5,000 for an additional $30,000.  Adding all this delay cost gives us a total of $564,000 + $208,000 + $52,000 + $30,000 for a total of $854,000 Delay Cost incurred.

Weighted Shorted Job First (WSJF): Use Cost of Delay Divided by Duration

If we develop the features based on whichever has the highest ratio of value to duration,we would do Feature A first; followed by Feature D, then Feature C; and finally Feature B.  For the 2 sprints we are working on Feature A we incur Cost of Delay of $47,000 per week. Delay Cost = 4 weeks * $47,000 = $188,000.  For the next 6 sprints we are working on Feature D we incur Cost of Delay of $39,000 per week or 12 weeks * $39,000 so  Delay Cost = $468,000.  For the 4 sprints we are working on Feature C we incur Cost of Delay of $18,000 per week.  Delay Cost = $144,000.  Finally Feature B is developed over 3 sprints at a Cost of Delay of $5,000 per week.  Delay Cost =  $30,000.  Total Delay Cost is $830,000.

Summary

How do I get started with actually using Cost of Delay? Surprisingly, doing the most valuable feature first is not the best economic decision.  Next time you prioritize your portfolio, don’t just try to maximize value delivered. Limit your Cost of Delay with a little quick estimation and some simple math.

Priority Method Cost of Delay
All at the same time $1206000
Shortest first $946000
Most valuable first $854000
WSJF $830000

 

Some other sources on the topic:

Wikipedia: Cost of Delay
The Principles of Product Development Flow: Second Generation Lean Product Development

 

15 comments on “Prioritizing to Minimize Cost of Delay”

  1. Kunal deep

    This observation looks very informative. Great take aways about time management. But I would say a little mention of the methodology and referneces would add more credibility to the argument.

    thanks,
    Kunal

    Reply
    • Derek Huether

      Kunal, thanks for the feedback. I based my blog post off a presentation about economic prioritization by our COO, Dennis Stevens. His presentation was based on the works of Don Reinertsen.

      Regards,
      Derek

      Reply
  2. Matt

    Hey Derek,

    Great post. Recently had a conversation with a group that was struggling with the fact that there was one team and multiple product owners. And the challenge is that work got done based on strength of personality or internal political collateral.

    The advice provided was to get the product owners to establish a common metric to compare the requirements being provided — that common metric is currency and leveraging cost of delay to help vet. It’s not the final point of ranking, but helps set the starting point and in multiple instances take the irrational or emotional aspects of ranking out.

    Thanks for writing this blog, I’ll definitely refer folks over to it!

    Later,
    matt

    Reply
    • Derek Huether

      Matt, thanks for adding your comment. I’ve also seen what you describe. Too often, prioritization becomes an emotional exercise. When people use terms like “it’s nothing personal”, when talking business and prioritizing their project over yours, perhaps they should be the first we put in line to use a mathematical calculation to rank our portfolio. Thanks again!

      Reply
  3. Fredrik

    Derek,

    Isn’t the real problem that you really have no idea how valuable a feature really is before it’s put into production and available to customers, which unfortunately equals already developed? Everything up until that point is guessing. Sure you can do business analysis, customer surveys, marketing research etc. but doing that for all the features you are considering implementing is not feasible.

    Reply
    • Derek Huether

      Fredrik, I think you make a valid point. When I use of the term “Cost” of Delay, I personally use it figuratively and not literally. You’re right, we don’t know exactly the actual worth. But, I do believe we should have an expectation of worth or value of items in our investment portfolio. That said, let’s shift away from the value calculations and instead use some other indicator like risk to our business. We could use indicators like a one for low risk, two for medium, and three for high risk. If we then use those values in combination with time, we’ll still get a similar outcome. With that, we could describe that as “Risk” of Delay. In this case, my goal is to generate a score that would allow us to prioritize our portfolio to mitigate risk.

      Reply
      • David Owen

        Derek, this is spot on! You can also consider your CD3 to be “return” in terms of dollars returned over time expended, then doing the work with the highest returns.

        If you also consider risk or uncertainty, you can calculate the Sharpe Ratio (SR = Return / Volatility), and use that as the criteria instead.

        The Sharpe Ratio also demonstrates that it’s worthwhile to minimize uncertainty, just as it is to maximize return.

        Reply
        • Derek Huether

          David, thank you for the validation. I greatly appreciate it! I’ll admit, I was unfamiliar with the Sharpe Ratio. Looks like I’m off to do some more reading. Thanks for sharing. Awesome stuff.

          Reply
    • Charles Betz

      Feature value can be estimated. See Doug Hubbard, “How to Measure Anything,” or the Maersk case study at blackswanfarming.com.

      Reply
  4. Sundar

    Hi
    Most Valueable calculations should be:
    C first: 6*19000 = 114000
    B next: =4*(7000+3000) = 40000
    A next: 3*3000 = 9000
    Total : 163000
    You considered C again in step-2
    FYI
    –Sundar With best regards

    Reply
    • Derek Huether

      Sundar, thank you for checking my math.
      I’ll run through it again.

      C = 6wks*$9+7+3= $114k check!
      B = 4wks*$7+3=$40k check!
      A = 3wks*$3k=$9k check!
      Total: $163k check!

      I went back to the paragraph I wrote and I see what you mean.
      Thank you! I will fix it right now.

      Reply
  5. Laurent Broyer

    Very Nice Post!

    This présentation is very simple but very powerfull !! Thank you, it makes me understand the full notion of Weighted Shortest Job First developped in the Scaled Agile Frmaework.

    I only have one question : I have made the exercise, and with the “most valuable first” scénario, i found $9000*6 + $7000*10 + $3000*13 = $ 163 000, did i miss something ?

    Best Regards.

    Laurent.

    Reply
    • Derek Huether

      Laurent, looks like Sundar found the same issue in my paragraph. You didn’t miss anything. Actually, you are one of only two people who found the issue! This blog post was RT’d on Twitter quite a few times. Thank you for reading this in detail.

      Reply
  6. Vishwanath

    In the most valuable section you have striked out 64K and added 40K as indicated by two of your readers below but you’ve forgotten to change the features that are contributing to the 40K. The features have to be B & A and not B& C. Thank you for the wonderful post though.

    Reply
    • Derek Huether

      I read your comment several times, checked my math, and shook my head. Yes, you did indeed find an error. I didn’t delete the error entirely. I lined out the original data and added the correct values. I’m not sure why the most valuable, which we all most commonly use, is the one that tripped me up. Still, I appreciate that you read my post in detail and were compelled to leave a comment. Thank you!

      Reply