Learn by clicking · ~4 minutes · Day 28 of 30

Is this AI tool actually paying off?

ROI (return on investment) just means: for every dollar you put in, how much did you get back? Measuring AI ROI is the same as weighing a new hire: did the help you gained beat what it cost you? Here is the whole idea in one picture:

Before

your baseline: hours, response time, no-shows, leads won

→turn it
on

After

the same numbers, measured again

−minus

Cost

subscription + token usage + cleanup time

📏 Start with a baseline you can count ⏱️ Before vs after on the same task 💵 Subtract the real cost ✅ What is left is your return

A baseline is simply the number you live with today before any AI touches it: hours per week on a task, average response time, your no-show rate, leads converted. Measure it now, turn the tool on, measure again. The gap is the value. You will run that exact math on your own numbers in a minute.

02 / 05 · The myth

"It is worth it because everyone says so."

This is the costliest myth in AI right now. It comes in two flavors, and both are wrong. Let us call them out and correct them.

Myth A: "AI is obviously worth it, everyone is using it." Everyone using something is not evidence it pays off for you. Plenty of paid seats sit unused. Hype is not a baseline.

Myth B: "You cannot really measure AI, the value is too fuzzy." You can. The trick is to measure the task, not the vibe. Pick one number you already track, and watch it move.

🚫 The fuzzy way

How most people "decide."

– "It feels faster." (Feelings are not numbers.)
– "A competitor uses it." (Their task is not yours.)
– "Look how many people it touched." (A vanity win.)
– No before, no after, no cost on the page.

📐 The honest way

How you will do it after this lesson.

✓ One baseline number you can count today
✓ The same number after, on the same task
✓ Full cost: subscription plus usage plus cleanup
✓ A plain dollar figure you can defend

The key: the simplest honest ROI is hours saved times your hourly value, minus what the tool costs. That is it. If a tool saves you 4 hours a week and your time is worth $60 an hour, that is roughly $960 a month of saved time to weigh against the bill. You do not need a spreadsheet army, you need one number you trust.

03 / 05 · Watch it work

Your ROI, in real time.

Move the sliders. The math updates instantly. Try a starting point, then nudge it to match your own situation.

Hours per week the tool saves you 4

What an hour of that work is worth to you

$

Tool cost per month (subscription + usage)

$

Hours per week YOU spend reviewing or fixing its work 1

Net value per month

$0

Time saved value$0

Minus your cleanup time$0

Minus tool cost$0

Illustrative only. Figures are approximate, and prices change. This is a thinking tool, not an invoice.

🔒 This runs entirely in your browser. Nothing you type here is sent anywhere.

💡 Notice the cleanup slider. Review and fix time is the hidden cost that quietly eats a tool's value. A tool that saves 5 hours but needs 4 hours of correction is barely worth it. Token usage is a real cost too: see /learn/tokens.

04 / 05 · Measure it honestly

5 traps that fake a good ROI.

A number is only as honest as what you left out of it. Before you trust your figure, check it against these.

1. Hidden cleanup cost. Count the time you spend reviewing, correcting, and re-prompting. It is real, and it is usually missing. (That is why the calculator has a cleanup slider.)
2. Vanity wins. "It wrote 200 emails" is a count, not a result. Did replies, bookings, or revenue actually move? Measure the outcome, not the activity.
3. Token usage you forgot. The subscription is rarely the whole bill. Heavy use adds per-token cost on top. Pull the real number, do not guess.
4. The wrong baseline. Compare to how the task really ran before, not to a perfect imaginary version of it. An honest before makes an honest after.
5. Quality drift. Faster is not better if the work got worse. If output quality dropped, that is a cost even when the clock says you saved time.

Keep your volatile inputs (the price, your usage) in one place and revisit them, as of writing these numbers move month to month. The durable idea does not: before, after, minus cost.

05 / 05 · Done

You now measure AI ROI better than most people who buy it.

You can pick a baseline, compare before vs after on the same task, weigh it against the full cost, and spot the traps (cleanup time, vanity wins, forgotten tokens) that fake a good number. That is the whole discipline.

The honest version of this is just: hours saved times your hourly value, minus cost. The hard part is wiring AI to a real outcome you can actually measure, then reporting on it month after month. That is what we build, end to end.

We build AI tied to outcomes you can measure, and we report on it. Have Rabbithole build it →

Day 28 of 30 free, working AI lessons for small business.

See this idea doing real work in the kits:

What tokens cost → Lead catcher (leads converted) → AI secretary (no-shows, hours saved) → Quote bot (response time) → Which model (cost vs quality) →