Run the calculator
Six numbers. Every one of them has a note explaining what to count and where most people get it wrong — open them if you want the estimate to be defensible rather than fast. Nothing is sent anywhere; the arithmetic runs in your browser.
What to count
Everyone whose working day is affected — support agents, the operations people supporting them, and the handful of others who get pulled in most weeks, usually someone in engineering and someone in finance. Not your whole company.
Where people get it wrong: counting only the support headcount. The engineering time absorbed by escalations is normally the most expensive hour in the whole calculation, and it is the one nobody books against support.
How to work this out
This is loaded cost, not salary — pay plus employer taxes, benefits, equipment, software and a share of overhead. Loaded cost is usually 1.25 to 1.4 times base pay. A support role on $62,000 works out around $40 an hour loaded, across roughly 2,080 working hours a year.
$60 is a fair blended default for a mixed group that includes senior and engineering time. Your finance team will have the exact figure, and they will be pleased you asked for it.
What counts as repeated manual work
Anything a system should have done: re-typing the same details into a second tool, copying ticket data into a spreadsheet, tagging or re-routing by hand, rebuilding the same report every week. Count recurring steps only — not one-off projects.
Where people get it wrong: guessing low. This is the number teams underestimate most. If you can, ask two people to note it for a week before you trust your own figure.
What counts as hunting for answers
Time spent finding something that already exists somewhere — asking in Slack, searching two or three knowledge bases, reading old tickets to work out what was decided last time, or waiting on the one person who knows.
Where people get it wrong: counting only the person asking. The person who stops work to answer loses the time too, and they are usually more senior.
What counts as handoff loss
Time lost at the seams between people and teams: chasing something that went to engineering and stalled, re-explaining context the next person never received, or redoing work that came back incomplete.
Where people get it wrong: counting the handoff itself. The handoff is cheap. The waiting and the rework around it are what cost money.
How onboarding is counted
Roughly how many people join the teams above in a year, including replacements for leavers.
The calculator assumes each new person takes about four extra weeks to reach full productivity because the process lives in people's heads rather than in writing, and counts that at half effectiveness — 20 lost hours a week. That is deliberately cautious. Teams with no written runbooks routinely lose more, and so do the experienced colleagues answering their questions.
Why this number is probably too low
Three things the calculator does not attempt to count, because they cannot be estimated honestly from six inputs:
- What it costs your customers. Slower resolutions, repeat contacts, and the churn that follows them. Real, and not guessable from the outside.
- Software you are paying for twice. Overlapping tools and licences nobody has cancelled. This needs your actual spend data — and it is usually the fastest money to get back.
- Senior attention. The judgment calls only one or two people can make, and the queue that forms behind them.
An assessment counts all three. This calculator exists to tell you whether it is worth finding out.
Now — is that number a problem?
One of the three below will be true. Each leads somewhere useful, none of them needs your email address, and only the third involves us at all.
It's smaller than you feared
Then that is a real result and you should take it. Your process is holding up better than most at your size. Re-run this in a quarter — the number tends to climb quietly with headcount — and take the free patterns from the blog in the meantime.
It's plausible, but you'd want to check the working
Sensible — it is an estimate built on your own guesses, and you should not act on it as it stands. The next thing to read is a complete assessment we built on invented data, with every calculation exposed, so you can judge the method rather than take our word for it.
It's real money, and you want evidence instead of an estimate
That is what the assessment is for. It replaces every guess above with your actual ticket data, tooling and workflows, prices each pattern with the arithmetic shown, and tells you which parts are worth fixing — and which are cheaper to live with.
A 30-minute call. No deck, no sequence. If an assessment is not worth your money we will say so on that call.
What it measures
Four kinds of waste, all of which behave the same way: small enough to ignore in any given week, expensive enough to matter over a year.
- Repeated manual work — steps people do by hand because two tools don't talk to each other, or because nobody ever automated the process.
- Hunting for answers — time spent finding information that should have been written down: policies, past decisions, "how do we handle this one."
- Handoff loss — work that stalls or gets redone at the seams between people and teams.
- Onboarding drag — the extra weeks new people need because the process lives in someone's head rather than in a document.
Every assumption is deliberately cautious, and the working is printed with your result. A number your finance team can pull apart is worth ten numbers that look impressive on a slide.
Common questions
Do I have to give you my email? No. Your number appears on screen, and nothing is sent anywhere — the arithmetic runs entirely inside your own browser, so we never see your figures at all.
How accurate is it? It's an estimate built on your own inputs. It's accurate enough to decide whether the problem deserves attention, and nowhere near accurate enough to put in a budget submission. Every assumption is visible so you can argue with it.
Can I share the result? Yes, and it's designed for that — the working is printed alongside the number precisely so it survives being screenshotted into a leadership deck and questioned.
What's the single biggest source of error? People guess low on repeated manual work, and they leave engineering out of the headcount. Both push the figure down. If you want a sharper estimate, ask two people to note their time for a week before you fill it in.