How to tell if an automation really saves you hours

"We've been so much less swamped since we automated the quotes." Great — but how much is "so much"? Three hours a week? Twenty minutes? Until you know, you can't decide whether that automation stays, gets fixed or gets switched off. And no, measuring it doesn't take dashboards or a consultant: four numbers and one subtraction will do. This guide is for you if you've already automated something (or you're about to) and want to know, with actual data, whether it's giving you hours back.

Why the feeling of saving time lies (both ways)

We're terrible at estimating time. Tedious tasks feel endless, so when they disappear it feels like you've won half your life back — even if it was really twenty minutes a day. And the reverse is just as true: an automation that works well becomes invisible, and three months in nobody remembers how much work it used to swallow.

That double illusion has two very practical consequences:

• You keep automations that actually cost you time — reviewing errors, fixing outputs, chasing edge cases — because the feeling is good.
• You switch off, or quietly let die, valuable automations because nobody could defend them with numbers.

Both problems have the same fix: measure. Not with laboratory precision — with just enough precision to make a decision.

The method: four numbers and one subtraction

Write these four numbers down anywhere; a note on your phone is fine.

1. Baseline. How long the task took by hand, per unit: minutes per quote, per order, per email answered. If you automated without measuring first (we've all done it), reconstruct it: do the task manually three or four times, time yourself, take the average. Don't estimate from memory — memory inflates.
2. Residual time. What you still spend on the task with the automation running: reviewing what the agent proposes, approving, fixing the odd case that came out wrong. This is never zero, and that's fine: you giving the final sign-off is a safeguard, not a system failure.
3. Maintenance. The time spent looking after the automation: updating prices or rules, tuning replies, checking it still works after something changed. Spread it out per week.
4. Frequency. How many times the task happens per week.

Then the subtraction:

Net weekly saving = (baseline − residual time) × frequency − maintenance

That's it. You really don't need a fancier formula.

A worked example: quotes by email

Picture a renovation company. Each quote request that lands by email takes about 25 minutes: read what they're asking for, check rates, build the document, write the reply. Around 12 come in per week — a good five hours.

They set up an agent that drafts both the quote and the reply, and one person reviews and approves. Reviewing takes about 5 minutes per quote: one hour a week. Keeping the agent up to date — rates, materials, the occasional new rule — takes another half hour weekly.

• Baseline: 25 min × 12 = 5 hours
• Residual time: 5 min × 12 = 1 hour
• Maintenance: 0.5 hours
• Net saving: 5 − 1 − 0.5 = 3.5 hours a week

Notice the nuance: enthusiasm would say "we're saving five hours". The honest sum says 3.5. It's less — but it's real, defensible and comparable against what the tool costs.

Then comes a second question, arguably more important: where do those 3.5 hours go? If they evaporate into more email, the saving exists but doesn't pay off. Decide up front what they're for: sales visits, better service, or leaving early on a Friday. All valid — as long as it's a decision, not an evaporation.

The costs almost nobody counts

The subtraction above only works if you don't cheat at solitaire. The usual blind spots:

• The errors you end up fixing. If the agent gets 1 case in 20 wrong and each fix takes half an hour, that's residual time, not a one-off. Count it.
• Keeping one eye on it. Checking the system "just in case" is time too. If you don't trust it and double-check everything, the automation isn't ready yet — and the measurement is telling you exactly that.
• The cost in money. The tool or service has a price. To compare it with hours, translate it: if your hour is worth €30, a €90-a-month subscription is three hours a month the saving has to clear before it starts counting in your favour.
• The ramp-up weeks. The first days always cost more: configuring, correcting, fine-tuning. Don't measure week 1; measure week 4 or 6, once the system runs on its own.

When measuring isn't worth it

Let's be honest: not everything deserves measuring.

If the task took you less than half an hour a week, timing it precisely costs more than it clarifies. A five-minute check once a quarter — "is this still working, and still not creating work for me?" — is plenty.

Also, don't measure only hours when the value lives somewhere else. A lead getting a reply in two minutes instead of six hours barely saves you any time, but it changes the outcome: more conversations that end in a customer. There, the right metric isn't hours recovered — it's response rate or closed sales. Measuring the wrong thing misleads just as much as not measuring at all.

And one very human exception: if an automation saves you little but takes the task you hate most off your plate, keeping it for your own sanity is legitimate. Just be aware that that is why you're keeping it — not because the numbers say so.

Keep, fix or switch off

With the net saving in hand, the decision becomes simple:

• Clear, stable saving: it stays. Write the number down and revisit it quarterly, because processes change and what saves time today can get in the way tomorrow.
• Small or erratic saving: fix before you give up. The culprit is almost always residual time — too many corrections. Improve the rules, narrow the scope (let the agent handle the typical cases and hand you the odd ones) and measure again in two weeks.
• Negative saving after 4–6 weeks: switch it off, no guilt. An automation that costs more than it gives isn't "technology" — it's a process that doesn't work, and shutting it down is good management too.

Measuring once tells you whether the automation worked. Measuring quarterly tells you whether it still works — which is the question that actually matters. And if the task you want off your plate is an operational one — quotes, orders, follow-ups, coordination — take a look at our operations agent: it works by your rules, and you approve before anything goes out the door.