Cost of Delay is a pretty simple construct, usually expressed as $/time. There are many different applications of Cost of Delay. Some examples include the timing of market entry, hiring, acquisition (of anything), deployment, receivables/collections, shipping/delivery, and so on. Really, almost any business process with a time element could have a Cost of Delay calculated – but that’s not a habit that most of us have acquired. Instead, we tend to use our intuition and experience (especially with complicated/manufacturing models and education) to substitute for a calculated Cost of Delay. As Don Reinertsen says, the calculated Cost of Delay is the ‘golden key’ to making economically-based decisions about our development effort.  According to Reinertsen, no more than 15% of development teams have made an attempt to calculate even a high-level Cost of Delay (‘If we miss our ship date by one month, what would the Cost of that Delay be?’). Further, the collected sample data showed that within teams, the ‘guess-timated’ Cost of Delay varied by up to 50X! The difference between a $100K and a $5M Cost of Delay is significant, and shows the impact of widely varied intuition and experience, as opposed to actual data. On the other hand, if we have a consensus Cost of Delay metric, we could use it as a critical input for all kinds of rational decision making (rational in the sense of ‘economically based’, as opposed to ‘based on intuition’).

Let’s bring in two other useful pieces of information for our rational decision making. Brooks Law says that ‘adding people to a late development project makes it later.’ This concise rule of thumb covers a multitude of sins, but especially those around the increased burdens of additional communication paths, education/training, and team formation. The second piece is that high-performing teams have been shown to demonstrate vastly greater productivity than ‘normal’ teams – measured at 5X, 10X, even more than 20X. That puts a premium on high-performing teams. How to lay the foundation for the formation and development of high performing teams will have to wait for a future post. For now, let’s postulate that it has value, and so we should generally put a thumb on the scale in favor of methods that we think improve our chances of achieving a high-performing team, and have a skeptical eye about methods that seem to work against such teams.

Next we’ll roll together a mini-case study that includes the Cost of Delay, Brooks Law, and high-performing teams.

In this case, our Product Owner is considering staffing options for her development effort. The Sales, Marketing, and Product Mgmt staff are located in San Francisco. There are three options being considered.

Option A – Development Staff to be hired in India, using a reputable firm with a good track record. Estimated duration of 7-8 months, at a cost of $500K/month.

Option B – Development Staff to be split between San Francisco and India (using the same reputable firm). Estimated duration of 6-7 months, at a cost of $750K/month.

Option C – Development Staff to be hired in San Francisco. Estimated duration of 5-6 months, at a cost of $1.25M/month.

Option A is the least expensive, by far. The cost of Staff is very reasonable, as is the cost of desk/working space. On the other hand, there’s a difference of 11 time zones to the SF staff, so communication will be challenging.

Option B is in the middle, expense wise, but has the benefit of some Dev Staff being co-located in SF with the Sales/Marketing/ProdMgmt staff. Issues of remote LGM and time zones from Option A are still present.

Option C is the most expensive. The cost of Development Staff is extremely high, and desk space is a scarce and expensive resource.

For many of you, this case is probably familiar. Personally, I’ve been involved in decisions like this several times.

The Product Owner for this case, however, has embraced the Cost of Delay metric, and tasked Product Management with calculating a working figure. Using questions like ‘How much less valuable would this product be, if it were not delivered to market for an additional month (or two, or three, or six)?’, they’ve determined that getting the new release to market in 6 months would be worth X $million in profit, but being delayed to 8 months would be worth only X-$4million. That would give us a Cost of $2M, and Delay of 2 months, or a Cost of Delay of $2M/month. Further, delivering the new release in under 6 months would generate approx. $1M/week in additional profit.

Using the ‘high’ estimates for duration, we get the following aggregate figures:

Option A – Eight months duration, development cost of $4M, Cost of Delay (2 months) $4M, total cost of $8M.

Option B – Seven months duration, development cost of $5.25M, Cost of Delay (1 month) $2M, total cost of $7.25M.

Option C – Six months duration, development cost of $7.5M, no Cost of Delay, total cost of $7.5M.

Using the ‘low’ estimates for duration, we get the following aggregate figures:

While one can quibble with the details of these numbers, they’re pretty clear. My personal experience is that Options A & B are far more prone to delay than Option C, with durations at the high end of the estimate, or past the highest estimate. It’s not hard to imagine a scenario where, using Option A, we don’t get to market until 10 or even 12 months have elapsed.

Option A – Ten months duration, development cost of $5M, Cost of Delay (4 months) $8M, total cost of $13M.

Option A – Twelve months duration, development cost of $6M, Cost of Delay (6 months) $12M, total cost of $18M.

   Getting a quality estimate of Cost of Delay is the ‘golden key’ to making good decisions in a complex environment, as it enables us to use economic methods instead of emotion to drive the decision making process. Consider adding this powerful tool to your repertoire, because there’s a good chance your competition isn’t thinking in these terms.

   This is a good example of a complex problem, where easy comparisons of the available options is difficult. In every case, we’re asked to predict the future, with limited data, so we should have inherent mistrust of the projections. Trust can increase as experience confirms (or denies!) the validity of the estimates. While Option A (India) presents a lower initial cost estimate, taking potential Cost of Delay impacts into account changes the equation. If we further expanded this case to include the cost of communication & coordination, the equation would change further (time zones, culture, longer feedback loops).

   The cost graphs for these Options are very different, yet are only estimates. They range from a total of $2.25M (optimistic finish, Option C), to $18M (delayed finish, Option A). If our objective is to deliver the product to market quickly, Option C is our best option (in this case), despite being the highest cost/month. Without factoring in the Cost of Delay and factors relating to high-performing teams, Option A is very appealing. The Cost of Delay is a good example of how bringing in established methods from other fields can enhance our results in a Knowledge Work environment!