Skip to content

BusinessPractitioner10 min read

1. Building the business case

Work out with management control what incidents really cost, derive the return on investment and the payback period, count the full cost and present scenarios.

For: finance, risk and compliance · executives and CIOsPrerequisites: None.

Reading mode

Without figures, observability remains one IT project among many. When budgets get tight, it is the first one to be cut. This lesson gives a method for building a case that a finance department takes seriously.

Every business case starts with what incidents cost over the last twelve months. This figure is almost always missing from existing dashboards. Its absence is already an argument: you cannot steer what you do not measure.

Think of a fire in a warehouse. What drives the bill is the time before someone smells smoke, the time before the fire brigade arrives, the time it takes to put the fire out and what burned in the meantime. An IT incident breaks down the same way, with English acronyms you will see everywhere. MTTD (mean time to detect) is the average time between the start of an anomaly and the moment it is spotted. MTTA (mean time to acknowledge) is the time between the alert and someone taking ownership of it. MTTR (mean time to recovery) is the time between detection and getting back to normal. Then comes the financial impact: margin lost during downtime, penalties written into contracts, goodwill gestures to customers.

Cost per minute: your figure, not a study’s

Section titled “Cost per minute: your figure, not a study’s”

The most quoted figure comes from Gartner: $5,600 per minute of network downtime on average, or about $300,000 per hour. It is an estimate published in 2014 by Andrew Lerner, an analyst at Gartner. It is an old average, in dollars, that mixes companies of all sizes. It shows that an order of magnitude exists. It does not put a figure on your case.

Your figure is calculated with management control, as lost margin, not as IT cost. An hour of ERP downtime does not cost the salaries of the engineers fixing it. It costs the parts that do not leave the factory.

hourly downtime cost = lost hourly margin
+ contractual penalties per hour
+ catch-up costs (overtime, scrap, follow-ups)
annual direct cost = sum of downtime hours x hourly cost of the system concerned

Indirect costs (a customer losing trust, team hours mobilized) are added separately, presented as an estimate along with their assumption. If you mix a validated direct cost with an assumed indirect cost, the second drags down the credibility of the first.

Two formulas are enough. ROI (return on investment) tells you how much each euro invested brings back. The payback period tells you after how many months you have recovered the outlay, the way you work out how many years solar panels take to pay for themselves.

ROI (%) = (gains - costs) / costs x 100
payback period (months) = initial investment / monthly gain

Gains come from four sources, each quantified separately. The first is lower incident costs: fewer hours of downtime because problems are detected earlier and fixed faster. The second is tooling rationalization, when duplicate licenses go away. The third is team time recovered, with fewer pointless alerts and shorter diagnostics. The fourth concerns the observability data itself, whose bill can come down (see lesson 4).

On the cost side, count licenses or infrastructure, storage, people (those who run the platform and those who instrument the applications), training, change management and migration.

An executive committee often looks at the payback period before the ROI. “How many months until we get our money back?” lands better than a five-year percentage. To estimate it with your own assumptions, the MTTR business case calculates the return, the payback period and how sensitive the result is to each assumption.

Order matters. A case that skips a step or swaps two of them convinces poorly.

flowchart LR
  E1["1. Cost of<br/>incidents"] --> E2["2. Expected<br/>gains"] --> E3["3. Full<br/>cost"] --> E4["4. ROI and<br/>payback"] --> E5["5. Three years,<br/>three scenarios"] --> E6["6. Pitch to<br/>committee"]

You start by putting a figure on the cost of incidents over the last twelve months: their number, their duration, the hourly cost of each system affected, the penalties. Then you identify the expected gains, source by source. The downtime reduction assumption must be justified: for example, the list of incidents you would have caught earlier with an alert on what the user sees.

Third, you total the full cost over three years, not just the licenses. Then you calculate the ROI, the payback period and the net present value if your finance department uses it. You project everything over three years with three scenarios (conservative, central, favorable), comparing at least two architectures with the same criteria (see lesson 3). Finally, you present to the committee: five slides, ten minutes, one decision requested (see lesson 7).

A business case that only shows licenses falls apart at the first serious budget review. It is like costing a car without insurance, fuel or servicing. Here are the items most often underestimated.

ItemWhat to countMain lever
Metrics storagevolume of active series, retention, replicationcontrol of cardinality
Log storagevolume ingested and indexed, retention by categoryhot, warm and cold tiers
Tracesspans ingested, retentiontail sampling
Peoplerunning the platform, on-call, instrumentationautomation, versioned configuration
SaaS subscriptionsbilled units (hosts, volume, users, modules)negotiation, caps, usage review
Change managementtraining, support, documentationinternal champions, written procedures

A few words in the table deserve a translation. An active series is a curve the platform tracks continuously; the more of them there are, the bigger the bill. Retention is how long a piece of data is kept. A span is one step in a request’s journey, like a stamp on a parcel’s route.

I do not give a price per gigabyte here, nor a total cost range by company size. They depend too much on architecture and contract to be useful out of context. The Collector cost simulator puts a figure on your own volume with the unit prices you enter.

These levers are decided at design time. Pulled early, they weigh on all three years of the business case.

The first concerns traces, those detailed accounts of a request’s journey. You do not have to keep all of them: well-tuned sampling keeps every trace that failed or ran slow and only a fraction of normal traffic, like a quality check that examines every defective part but only one good part in a hundred.

The second is differentiated retention. Not all data ages the same way: a debug log does not need the lifespan of an authentication log that an auditor may ask for a year later.

The third is cardinality, meaning the number of different combinations a metric can take. You avoid putting a unique identifier in it (a user or request number), which multiplies the curves endlessly.

Then come three steering levers. An observability cost dashboard shows the cost per gigabyte ingested, per service and per team, with an alert when it drifts. Contract negotiation covers commitment, caps and the exit clause, always with comparable offers in hand. Finally, the architecture is chosen on criteria: self-managed open source, SaaS or a mix of the two, compared with the same solution evaluation grid.

The effect of each lever depends on your volume and your contract. Rather than trusting a generic percentage, measure it: the Collector cost simulator shows the bill before and after each lever.

Here is a template to fill in. The “method” column matters as much as the value, because that is what a CFO will ask about.

Cost of incidents (last twelve months)

ItemCalculation methodValue
Major incidentspriority 1 and 2 tickets
Downtime durationsum of downtime durations, per systemhours
Hourly cost of downtimelost hourly margin, validated by management control€/h
Direct costduration x hourly cost, per system€
Estimated indirect costexplicit assumption, validated or not€
Regulatory exposureseparate line, not added to gainsqualitative

Projected gains over three years

YearDowntime reduction assumption (to be justified)Annual gainCumulative
1
2
3

Summary

IndicatorFormulaValue
Three-year ROI(gains - costs) / costs x 100%
Payback periodinitial investment / monthly gainmonths
Gains to costs ratiototal gains / total coststo 1

Allow 45 minutes, alone or in pairs.

  1. Put a figure on the incidents of the last twelve months: number, total duration, cumulative cost.
  2. Obtain or estimate the hourly cost of downtime for your most critical system, with its assumption.
  3. Formulate a downtime reduction assumption and justify it incident by incident.
  4. Calculate the annual gains, the three-year ROI and the payback period in the MTTR business case.
  5. Write down your two biggest uncertainties: they will be the committee’s first questions.

To finish, write down your next action: which one, by what date, with whom?