How execution delay impacts payoff functions
[Cost of Delay series — Part V: “Growth”]
“If we don’t change, we don’t grow. If we don’t grow, we aren’t really living.”
(Gail Sheehy)
In Part IV of this series, we introduced Cost of Delay (CoD) as the answer to the question: “How much do we gain or lose if we advance or delay this work item by x amount of time?” Initially we set out to use this metric to help us build budgets, by comparing the Cost of Delay of work units waiting in queue with the Cost of Capacity of hiring more people. But in addition to budgeting, we also found we can use CoD to prioritize, and determine what the most valuable thing is for the team to work on next. We proposed four categories of value: Maintenance, Growth, Cost Savings and Risk Management. We discussed a quantification approach for the first (and arguably easiest) category: Maintenance. This time, we’re going to tackle the most exciting (and perhaps hardest) category: Growth.
We defined Growth as cash income generated from new products. By definition, we’ll deal with many more unknowns to build our CoD model compared to Maintenance. If the team has done similar projects before, we may have a pretty good idea about the required execution time. We’ll have a much harder time modeling the payoff function, given the uncertainty on how fast we can expect business to ramp up, how big the total financial potential will be, and how long we can expect to sustain stable income before it declines and stops. In many cases, we can get the help of a dedicated financial analyst to model the expected payoff function. This usually involves supporting activities like market research, investment analysis, etc. Let’s look at what those payoff functions typically look like, and determine how we can use them to derive CoD estimates.
The left graph shows a typical development of cumulative income over time. The business takes off slowly, then goes through a period of accelerated growth, before growth slows down and stops[1]. In the example model, we use 48 months as “lifetime” and the total cumulative income as 100. The middle graph expresses the same pattern in monthly contributions: while the left shows absolute amounts (“EUR”), the middle one shows a run rate (“EUR / month”). In this model, the monthly run rate peaks in the 18th month. The right again shows the same business in a format we’re more used to seeing: annual income.
To determine Cost of Delay, let’s first look what happens if 1 month of delay simply shifts everything one month to the right.
At first sight, the impact looks modest. In the left graph, the blue curve is the same as the original grey one, shifted by one month. The middle graph plots the delayed amount for every month. Unsurprisingly this is the exact same curve as our original “Monthly Cash Income” graph. “Shifting by one month” is the equivalent of “this month’s cumulative income minus last month’s cumulative income”. This graph shows how the delay damage in the very first month is not so high (about 1 in our model). This happens at the start of the early, low growth phase and the monthly income is still small. The problem is that the shift keeps cascading through all later phases as well, also delaying the high income months (topping out at a bit over 3.5). From this middle graph, we can derive the right graph, which plots the average monthly delayed amount up to that month. And there we observe how the average delay damage tops out at about 2.75 in month 24. This is an appropriate measure we’ll take as our Cost of Delay estimate. For every single one of the first 24 months, the delay in execution has resulted in missed income to the tune of 2.75, on average. After 24 months, the monthly average shrinks again, but this is small consolation as that is mostly because the growth of the business has dramatically slowed down by then. By that time, the damage has been done.
What we would also like to see reflected in our CoD measure is a higher cost if the expected growth rate is higher, and if this aggressive growth rate occurs earlier in lifetime. That is indeed the case, as the following models show.
With a growth rate of 30% (in this figure) instead of 15% in the original model, peak damage is higher (about 7 compared to 3.5) and CoD increases to 3.6 instead of 2.75. We also observe how over the total lifetime, both trajectories still end up at the same average (2).
Adjusting the timing of the fast growth face from 18 (in the original model) to 25 months (in this graph) keeps the peak damage stable at 3.5, but it occurs later. Our CoD metric reflects this reduced urgency and drop from 2.75 to about 2.3. Total lifetime damage is again the same, averaging out at 2 per month.
In practice, we will probably not receive the payoff function in the nice, mathematically elegant form of the lifetime S-curve (top left in the below figure). Most payoff projections we’ll receive will look more like the top left: an annual forecast. We can easily convert this to a lifetime curve though: simply dividing every forecast year in 12 equal portions is often a good enough approximation (top right — blue curve). The monthly rate curve looks crude, with 12 equal monthly contributions for every year (bottom left). But plotting the monthly average produces a reasonable CoD estimate (bottom right): about 2.9 compared to 2.75 in the original example.
A delay in execution can affect the payoff curve in other ways than a mere shift. One such other possible effect is a reduction in the total lifetime value, e.g. because it will be harder to reach the same market share if we’re late. The next figure models the impact of such a “first mover advantage” as a 5% reduction in the lifetime cash income curve. (Admittedly -5% for a one month delay is unrealistically extreme example, but we want to have some visual clarity in our lifetime graph on the left side. Contrary to a mere shift, the monthly delay amount does not peak at some point but keeps growing throughout the entire lifetime. Obviously that is then also the case for our CoD metric, which is simply the average on the very last recorded period. Once again this CoD metric will be higher if the growth rate is steeper and the high growth phase starts earlier. For very steep growth starting almost immediately, the CoD value will get close to the monthly delay amount value of 5% — maximum damage. Cost of Delay is always painful, but especially so if we’re a little late to the party in an explosively unfolding opportunity with a pronounced first mover advantage.
In Part VI of the series, we’ll switch from the Cost of Delay of sources of cash income to estimating the value of reducing cash outflows — cost savings, in other words.
[1] This is mathematically modeled using the so called S-curve.
