Quick summary · Chapter 1
What Computing Costs
First editionThe essential ideas, decisions, and diagrams from this chapter. Follow the section links for the examples, calculations, and fuller argument.
First-edition visual companion · Selected from the current chapter · 8x10-inch PDF
The main idea
Go back to the case from the introduction: a demonstration has turned into a launch, the service can absorb the new work only while nothing fails, and the additional capacity won’t arrive in time. One of Finance’s first questions is why the resources we already pay for can’t do more of the work. It’s a fair question. Before we can answer it, or ask anyone to approve a purchase, we need to say what the organization pays for today, what it has already promised to keep paying, and what would actually change if we did something different.
Knowing what you consume before you change it
Sometimes an application will stop using resources the organization has already committed to paying for. Releasing them may let another team use them or avoid a purchase later, but it does not cancel the payment. Every proposed change has to be followed through to the obligation before anyone calls it a reduction.
Several dimensions of the same dollar
Fixed, variable, committed, metered: these are useful words, but trouble starts when we treat them as four separate boxes and assume that placing an expense in one tells us how to manage it. A single cloud product might charge one component as fixed overhead and another by usage. Consumption can be metered whether it is covered by a commitment or billed at an on-demand rate. A service whose usage varies can still carry a minimum payment. The label alone does not tell us how the charge will respond to a change in use.
Then there is attribution, which connects an expense to the workload or decision that caused it. The easy version is to tag resources to the teams that use them and charge those teams, but many costs are shared. Some organizations spread them evenly across the company and call it a peanut butter spread; management accountants know it as peanut-butter costing and warn that it overcharges light users and undercharges heavy ones.1 Other shared costs go to the groups that use a particular service, and some are simply overhead. Attribution also gets political, because who pays is a question about who is answerable. It tells us who is consuming resources and where to start a conversation, not what the enterprise would save if that consumer went away, and shifting a cost between teams is surprisingly easy to mistake for reducing it.
A small number of teams dominate cost
How the charges behave
Follow a change across the boundary
The same bill, five ways
The same $448,000 has now appeared as five tables, each because a different conversation needs it. The work view says what could change and what the change would disturb; the attribution view, whom to talk to; the location view, which agreement governs the change; the charge view, how much of the bill is floor; and the funding view, whose budget moves and whether its owner has agreed.
Enough information to have the next conversation
I would begin the cost record with the work being supported, its owner, the period, the amount, the decision under consideration, and whether each charge is direct, shared, or still unresolved, then follow the change through every dimension. For this chapter’s two changes, it would start like this:
| Dimension | Eight borrowed batch workers | Sixteen new workers |
|---|---|---|
| Amount | $9,600 per thirty days moves to the web service; total unchanged | $19,200 per thirty days added, before any commercial offset |
| Cost behavior | Usage-sensitive; the charge continues because the workers stay in use | Usage-sensitive unless bought under a commitment |
| Obligation | None new; the agreement with the batch owner sets the return date | None until bought; a one- or three-year commitment would outlast the launch |
| Metering | Worker-days, unchanged | Worker-days from provisioning, which precedes qualified service |
| Attribution and funding | Web service; funding moves only if both budget owners agree | Web service; website product line, if it agrees |
| Cash | No additional payment | Set by the agreement, effective date, and invoice terms |
| Modeled labor | Qualifying the moved workers; the batch catch-up | Qualifying the new supply |
| Still open | The cost of the batch delay: unpriced, not zero | Lead time to qualified service |
Keep the underlying records available, because the first version will contain questions that somebody else can answer better than we can.
Charles T. Horngren, Srikant M. Datar, and Madhav V. Rajan, Cost Accounting: A Managerial Emphasis, 15th ed. (Pearson, 2015), chapter 5, “Activity-Based Costing and Activity-Based Management”: broad averaging, or peanut-butter costing, spreads resource costs uniformly across products or services that use those resources non-uniformly, overcosting some and undercosting others. The application to shared cloud costs is the author’s.↩︎