---
title: "AI Appointment Setter ROI: How to Run the Numbers"
slug: ai-appointment-setter-roi
author: "Leonardo Maldonado"
category: "AI Setter Operations"
articleType: data-research
tags: ["AI setter ROI","AI appointment setter payback","ROI of an AI setter"]
publishedAt: 2026-09-23T09:00:00.000Z
updatedAt: 2026-09-23T09:00:00.000Z
canonical: https://setluca.com/blog/ai-appointment-setter-roi
---# AI Appointment Setter ROI: How to Run the Numbers

> AI appointment setter ROI is the value of the extra booked calls it recovers, minus its cost, divided by that cost. The cost is a flat subscription, not a salary. The return comes from replying fast enough to catch leads a slow inbox loses. An AI setter like Luca drafts your first reply in seconds so fewer leads go cold.

## Key takeaways

- ROI = (revenue from added booked calls − annual cost) / annual cost. The cost is fixed on day one; the return scales with your inbox.
- The model has eight inputs: DM volume, reply rate, qualification rate, booking rate, show rate, close rate, average deal value, and tool cost.
- A human setter's pay runs $28,000-$47,000/year (PayScale, 2026); an AI setter like Luca is a flat $99-$499/month.
- Break-even for a $1,500 offer lands near 30 inbound DMs a month on example numbers. Below that, fix lead flow first.
- Average deal value and reply-rate lift move ROI most. Not because the math weights them, but because they vary most between coaches.

---

Most coaches ask what an AI setter costs before they ask what it returns. That's backwards. AI appointment setter ROI depends on whether the cost buys back more than it takes, and that's a model you can run yourself: eight inputs, the arithmetic between them, three worked business sizes, a break-even volume, and the cases where the answer is negative. Every benchmark here comes from someone else's data, and every worked figure is an example, not a promise.

## How do you calculate the ROI of an AI appointment setter?

You run your leads through a chain of six rates, multiply the survivors by what a client is worth, and subtract the subscription. The chain matters more than the formula, because that's where a setter actually does its work.

**Leads -> replied -> qualified -> booked -> showed -> closed x average deal value.**

Each arrow is a percentage. Multiply them and you get clients per month. Do it twice, once at your current rates and once at the rates you'd expect with every message answered. The difference is the only revenue you may credit to the tool.

**ROI = (added revenue − annual cost) / annual cost.**

### The eight inputs, and where to get each one

<table>

  <thead>

    <tr><th>Input</th><th>What it means</th><th>Where a number comes from</th></tr>

  </thead>

  <tbody>

    <tr><td>DM volume</td><td>Inbound messages a month, all channels</td><td>Count last month's inbox. Don't estimate.</td></tr>

    <tr><td>Reply rate</td><td>Share you actually answer</td><td>Your own logs. Most coaches overstate this badly.</td></tr>

    <tr><td>Qualification rate</td><td>Share of answered DMs that fit the offer</td><td>Chili Piper's 2025 benchmark found 14.1% of form submissions were unqualified; DM traffic runs far dirtier.</td></tr>

    <tr><td>Booking rate</td><td>Share of qualified leads who take a slot</td><td>Chili Piper, 2025: 30% baseline, 66.7% when leads book instantly.</td></tr>

    <tr><td>Show rate</td><td>Share of booked calls that happen</td><td><em>The American Journal of Medicine</em> (2010): 23.1% no-show with no reminder, 17.3% with an automated one.</td></tr>

    <tr><td>Close rate</td><td>Share of held calls that buy</td><td>Yours only. Borrowed close rates wreck the model.</td></tr>

    <tr><td>Average deal value</td><td>Revenue you actually collect per client</td><td>Your payment processor, net of refunds.</td></tr>

    <tr><td>Tool cost</td><td>Subscription plus overage top-ups</td><td>Published pricing. See <a href="/blog/how-much-does-an-ai-appointment-setter-cost">how much an AI appointment setter costs</a>.</td></tr>

  </tbody>

</table>

A setter moves three of these eight: reply rate, booking rate, and show rate. It does not move DM volume, qualification, close rate, or deal value. Any model that credits it with those is selling you something.

## What does the ROI model look like at three business sizes?

The same chain gives very different answers by volume and offer price. Qualification holds at 12%, booking at 30%, close at 25%. Show rate moves from 76.9% to 82.7% on the reminder finding above. Reply rate climbs furthest for the busiest coach, because busy inboxes drop more.

<table>

  <thead>

    <tr><th>Input</th><th>Solo coach</th><th>Growing</th><th>Busy</th></tr>

  </thead>

  <tbody>

    <tr><td>Inbound DMs / month</td><td>60</td><td>250</td><td>900</td></tr>

    <tr><td>Reply rate, before -> after</td><td>70% -> 100%</td><td>55% -> 95%</td><td>35% -> 90%</td></tr>

    <tr><td>Average deal value</td><td>$1,500</td><td>$3,000</td><td>$5,000</td></tr>

    <tr><td>Plan (annualized)</td><td>Starter, $1,188</td><td>Pro, $2,988</td><td>Scale, $5,988</td></tr>

    <tr><td>Clients / year before</td><td>3.5</td><td>11.4</td><td>26.2</td></tr>

    <tr><td>Clients / year after</td><td>5.4</td><td>21.2</td><td>72.3</td></tr>

    <tr><td><strong>Added revenue / year</strong></td><td>$2,805</td><td>$29,370</td><td>$230,500</td></tr>

    <tr><td><strong>Annual ROI</strong></td><td>~1.4x</td><td>~8.8x</td><td>~37.5x</td></tr>

    <tr><td>Cost per added client</td><td>$635</td><td>$305</td><td>$130</td></tr>

  </tbody>

</table>

_Example numbers. Reply, qualification, and close rates are made-up inputs, not Luca results. Show and booking rates are anchored to the studies cited above._

The 37.5x should make you suspicious. A flat subscription sitting in front of a big pipeline throws off multiples that look fake, because the cost you divide by is tiny, not because the return is huge. Read the bottom row instead: cost per added client falls from $635 to $130 while the tool price rises.

## Where's the break-even, and which input moves it most?

Break-even is the monthly DM volume where added revenue equals the subscription: **monthly DMs = annual cost / (12 x reply-rate lift x qualification x booking x show x close x deal value)**. At Starter pricing with a 30-point reply lift and the rates above, it lands here.

<table>

  <thead>

    <tr><th>Average deal value</th><th>Inbound DMs / month to break even</th></tr>

  </thead>

  <tbody>

    <tr><td>$500</td><td>~89</td></tr>

    <tr><td>$1,000</td><td>~44</td></tr>

    <tr><td>$2,000</td><td>~22</td></tr>

    <tr><td>$5,000</td><td>~9</td></tr>

  </tbody>

</table>

The higher your ticket, the less volume the math needs, which is why an [AI setter for high-ticket coaching](/blog/ai-setter-for-high-ticket-coaching) clears its cost fastest.

Now, which input matters most. In a chain of multiplied rates, a 10% change in any input moves the answer by exactly 10%, so the math weights them the same. What separates them is how far each realistically travels. Flex the Growing column, base ~8.8x: halve deal value, ~3.9x; halve the reply-rate lift, ~4.4x; drop close rate from 25% to 15%, ~4.9x; drop qualification from 12% to 8%, ~5.6x.

Deal value tops the list because it has the widest spread between coaches, from a $300 program to a $20,000 mastermind. Reply-rate lift is second, moving 40 points for a coach in client sessions all day and roughly zero for one who already answers everything. Get those two right and the rest can be rough. The same chain runs outside coaching, and [getting real estate clients on Instagram](/blog/ai-setter-for-real-estate-agents) sits at the high end of that deal-value spread.

## What goes into the cost side of the ROI calculation?

The cost side is the easy half, because both options publish a price. A full-time human setter's pay ran **$28,000 to $47,000 a year** in 2026, per PayScale, blending base, commission, and bonus. Base alone averaged **$41,415**, about $20 an hour, per Salary.com, with commission adding roughly **$9,000 to $15,000** on top. We break that down in [appointment setter commission rate](/blog/appointment-setter-commission-rate).

An AI setter like Luca costs a flat **$99, $249, or $499 a month** across Starter, Pro, and Scale, or about **$1,188 to $5,988** a year. See [Luca's pricing](/pricing).

### Annual cost inputs: human setter vs AI setter

**Chart:** A human appointment setter's total pay runs $28,000 to $47,000 a year, with base alone averaging $41,415. An AI setter like Luca costs $1,188 to $5,988 a year depending on tier. (Source: PayScale and Salary.com (2026); Luca pricing. Setter figures are pay only and exclude ramp, management, and overhead.)

## Which costs do coaches forget to count?

Sticker price is not the whole cost. Four things get left out, and together they can halve the ROI you'd report.

**Setup time.** Voice training, connecting channels, wiring your calendar and CRM. Budget a few hours in week one, then a lighter pass weekly for a month, priced at whatever your hour is worth.

**Review-queue time.** Auto-send stays off by default, so you approve every draft. At 250 DMs a month and fifteen seconds a draft, that's about an hour a month. Small, real, worth counting.

**Hooking things up.** Tools you've already connected cost nothing. The calendar sync you've been meaning to set up costs an afternoon.

**Going over the plan.** Conversations past your plan cap bill as top-ups, and dirty traffic from a viral reel eats them fast.

## Where does the return on an AI setter actually come from?

Speed decides who books. The odds of _qualifying_ a lead ran **21 times higher** at five minutes than at 30, across more than 15,000 leads in the MIT/InsideSales Lead Response Management study. Most of that five-minute window falls outside your working hours, which is the whole problem.

Follow-up carries the second half. It takes an average of **eight touchpoints** to land a first meeting, and even top performers need about five ([RAIN Group, 2024](https://www.rainsalestraining.com/blog/how-many-touchpoints-does-it-take-to-make-a-sale)). We build a five-touch rescue cadence because most lost coaching deals die from silence rather than a hard no. Mechanics in [DM follow-up sequence](/blog/dm-follow-up-sequence) and [speed to lead](/blog/dm-response-time-speed-to-lead).

## What does ROI look like in month 1, month 3, and month 6?

Negative, then flat, then real. The ramp is just timing: a lead landing in week one needs roughly eight touches to book, then a call, then a decision. One "I need to think about it" stretches that cycle by an average of **173%** ([Gong](https://www.gong.io/blog/sales-stats)). The people who message you in month one close in month two or three.

<table>

  <thead>

    <tr><th>Window</th><th>What's actually happening</th><th>What to measure</th></tr>

  </thead>

  <tbody>

    <tr><td>Month 1</td><td>Setup, voice tuning, heavy review. Subscription paid, almost nothing closed.</td><td>Reply rate and how long drafts wait for approval. Not revenue.</td></tr>

    <tr><td>Month 3</td><td>The five-touch cadence has run end to end. First recovered closes land.</td><td>Booked calls and show rate against your pre-launch baseline.</td></tr>

    <tr><td>Month 6</td><td>Rates settle down. You have real before-and-after data instead of guesses.</td><td>Added revenue and cost per added client.</td></tr>

  </tbody>

</table>

Time savings show up first. Revenue shows up later, and only if the queue gets worked.

## Dani's numbers: a worked example

Dani coaches strength clients, runs sessions from 9 to 6, and gets about **180 inbound DMs a month** across Instagram and WhatsApp. She answers roughly 60%, usually the next morning. Her program is **$2,400** and she closes **22%** of held calls. She's on Pro at $2,988 a year.

At 60% reply the chain gives her about **7.9 clients a year** from DMs. At 95% reply with reminders on booked calls, it gives **13.4**. That difference is 5.5 clients, or **$13,200** against $2,988, roughly **3.4x**.

Then she counts the queue. About 171 drafts a month at fifteen seconds each is 43 minutes, which at her own $150 an hour runs **$1,284 a year**. Fully loaded, her cost is $4,272 and her ROI is **2.1x**. Still worth it, and a very different number from the one on a sales page. One of those recovered threads:

> **DM example (illustrative):Lead (11:48pm):** hey are you taking 1:1 clients rn? saw your reel on consistency**Luca draft (11:48pm):** I am, a couple spots this month. Quick one so I point you right -- are you training on your own now, or coming back after a break?**Lead:** back after a break, keep starting and stopping**Luca draft:** That's the exact pattern we fix first. Want to grab 20 min this week so I can map your restart? I've got Wed 7pm or Thu 12pm.

No salary sent that message. A $249 subscription drafted it at midnight, in Dani's words, and waited for approval before anything went out.

_All figures are examples. Dani isn't a real person, and the rates are made-up inputs, not Luca results._

## When does AI setter ROI turn negative?

Often enough that you should check before you buy. Across 876 companies asked, only **46** could trace more than 10% of their operating profit to gen AI at all ([McKinsey, 2024](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2024)). Most rollouts never reach the profit line. Five ways yours won't either:

- **Thin volume.** Under about 30 inbound DMs a month at a $1,500 offer, there's nothing to recover. Spend the money on [coaching lead generation](/blog/coaching-lead-generation).
- **You're already fast.** If you answer in minutes and book everything worth booking, your reply-rate lift is near zero, and zero times anything is zero.
- **Low ticket, low volume.** A $200 offer with 40 DMs a month can't clear a subscription however the chain runs.
- **A close-rate problem in a lead-flow costume.** If people book and don't buy, a setter multiplies a leaking funnel. Fix the call first.
- **Nobody works the queue.** Drafts sitting unapproved for six hours are slow replies with a monthly bill attached.

That last one is where a human still wins outright. If your offer needs a live conversation to work out what's wrong before anyone commits, a human setter closes what a queue can't, and we'd rather you know before you pay us. [Hire a setter vs AI](/blog/hire-a-setter-vs-ai) walks the tradeoff.

## Which edge cases change the math?

**Launch-based businesses.** Selling in windows makes annual ROI meaningless. Model one launch: DMs in the window, added closes, and the quiet months you carried the subscription.

**Capped groups.** Twelve seats means no extra value from booked call thirteen. Cap the added-clients row at remaining capacity and value the overflow as waitlist, at a lower rate.

**Recurring coaching with churn.** Contract value is not collected revenue. If a $500/month program averages five months, use $2,500, not $6,000. Refunds come out too.

**A low-ticket front end.** If the DM leads to a $47 workshop that upsells later, value the recovered lead at lifetime value, or the model will tell you to switch it off.

**One channel doing the work.** If 80% of qualified DMs come from Instagram, run the model one channel at a time; an average across all of them talks you into coverage you don't need. See [Instagram DM response rate benchmark](/blog/instagram-dm-response-rate-benchmark).

## Troubleshooting: why isn't the ROI showing up?

<table>

  <thead>

    <tr><th>Symptom</th><th>Likely cause</th><th>Fix</th></tr>

  </thead>

  <tbody>

    <tr><td>Reply rate hit 95%, bookings flat</td><td>Drafts answer the question and never ask for the call</td><td>Require a booking ask by the second touch, with two named time slots</td></tr>

    <tr><td>Bookings up, revenue flat</td><td>Show rate collapsed under the extra volume</td><td>Add automated reminders; the AJM data puts no-shows at 17.3% with one versus 23.1% without. See <a href="/blog/how-to-reduce-no-shows-on-sales-calls">reducing no-shows</a>.</td></tr>

    <tr><td>ROI looks huge on paper, bank account disagrees</td><td>You're crediting total revenue instead of the lift</td><td>Compare against the 90 days before you started and count only the difference</td></tr>

    <tr><td>Recovered leads close far below your usual rate</td><td>You're now talking to colder leads you used to ignore</td><td>Tighten qualification questions; a lower booking rate with a higher close rate usually wins</td></tr>

    <tr><td>Cost per added client climbing</td><td>Top-ups burned on unqualified conversation volume</td><td>Screen out obvious non-buyers before they use up a conversation, or move up a tier</td></tr>

    <tr><td>Month three, still nothing</td><td>Drafts sitting too long, or drafts that don't sound like you</td><td>Check median time from draft to approval; retrain voice on your last 50 sent DMs</td></tr>

  </tbody>

</table>

## Common mistakes in AI setter ROI math

1. **No baseline.** Counting every booked call as recovered. You booked calls before. Only the difference belongs on the revenue side.
2. **Judging it in month one.** The cadence hasn't finished running. You're measuring setup cost against a cohort that hasn't decided.
3. **Contract value instead of money actually collected.** Refunds, churn, and broken payment plans all shrink what a client is worth.
4. **Leaving review time out.** Dani's ROI moved from 3.4x to 2.1x on 43 minutes a month.
5. **Borrowing a close rate.** Published averages come from other people's offers, and close rate is the input you already have the best data on.
6. **Expecting it to fix demand.** A setter lowers the cost of replying. It doesn't create anyone to reply to.

## Should you buy, hire, or wait?

Run the chain, then take a real branch instead of a feeling.

- **Buy an AI setter if** you get 30 or more inbound DMs a month, answer under 80% of them within an hour, and your offer clears $1,000. Flat cost means every extra DM improves the number.
- **Hire a human setter if** your offer needs a live diagnosis call before anyone buys, or you run enough qualified volume that commission costs less than the closes a queue loses you.
- **Do both if** you're past roughly 500 DMs a month on a $5,000-plus offer. Coverage from AI, closing from a person.
- **Do neither yet if** you're under 30 DMs a month. Buy lead flow first; the setter will still be here when the inbox is full.

If you can't tell which side you're on, the guide to AI setters for coaches covers the mechanics, then run your last 30 days through the chain above.

Want to run your own numbers against a live inbox? [Start free with Luca](/pricing), draft replies in your own voice, and approve every one before it sends.


## FAQ

### How do you calculate the ROI of an AI appointment setter?

Run your leads through six rates (replied, qualified, booked, showed, closed), multiply by average deal value, then do it again at the rates you'd expect with every message answered. ROI = (added revenue − annual cost) / annual cost. Only the difference counts, never total revenue.

### What inputs go into AI appointment setter ROI?

Eight: DM volume, reply rate, qualification rate, booking rate, show rate, close rate, average deal value, and tool cost. A setter moves only three of them: reply, booking, and show. Chili Piper's 2025 benchmark of nearly four million submissions puts instant booking at 66.7% versus a 30% baseline.

### How many leads do you need for an AI setter to pay for itself?

Around 30 inbound DMs a month at a $1,500 offer, on illustrative rates and Starter pricing. At a $5,000 offer it drops near nine; at $500 it rises to about 89. Break-even = annual cost / (12 x reply-rate lift x qualification x booking x show x close x deal value).

### Which input moves AI setter ROI the most?

Average deal value, then reply-rate lift. Mathematically every rate carries equal weight in a multiplied chain, so a 10% change anywhere moves the answer 10%. Those two win because they vary most between coaches: from a $300 program to a $20,000 mastermind, and from a 40-point reply gain to none.

### How long before an AI setter shows positive ROI?

Plan on three months, not one. Leads need roughly eight touchpoints to book (RAIN Group, 2024), and Gong found a single "I need to think about it" stretches the sales cycle by an average of 173%. Month one measures setup and reply rate. Month six is your first honest revenue read.

### When is AI setter ROI negative?

When volume is thin, when you already answer fast, when the offer is low-ticket, or when nobody approves the queue. McKinsey's 2024 State of AI survey found only 46 of 876 respondents attributed more than 10% of EBIT to gen AI. Most deployments never reach the P&L.

### Is an AI setter a better ROI than a human appointment setter?

Often, for lower-volume coaches. A human setter's pay runs $28,000-$47,000 a year (PayScale, 2026) and rises with volume. An AI setter is a flat $99-$499 a month, so cost per added client falls as bookings grow. Offers that need a live diagnostic call still favor a person.

### Does follow-up change AI setter ROI?

Follow-up is where much of the return hides. RAIN Group found it takes about eight touchpoints to land an initial meeting, and even top performers need five. A solo coach rarely runs that many by hand, so a setter's steady multi-day cadence recovers booked calls that silence would otherwise cost you.


## Sources

1. [PayScale -- Appointment Setter Hourly Rate (2026)](https://www.payscale.com/research/US/Job=Appointment_Setter/Hourly_Rate)
2. [Salary.com -- Appointment Setter Salary, United States (2026)](https://www.salary.com/research/salary/hiring/appointment-setter-salary)
3. [Lead Response Management Study (Dr. James Oldroyd, MIT + InsideSales.com)](https://www.leadresponsemanagement.org/lrm_study/)
4. [McKinsey -- The State of AI in Early 2024](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2024)
5. [RAIN Group -- How Many Touchpoints Does It Take to Make a Sale (2024)](https://www.rainsalestraining.com/blog/how-many-touchpoints-does-it-take-to-make-a-sale)
6. [Chili Piper -- Demo Form Conversion Rate Benchmark Report (2025)](https://www.chilipiper.com/post/form-conversion-rate-benchmark-report)
7. [The American Journal of Medicine -- The Effectiveness of Outpatient Appointment Reminder Systems in Reducing No-Show Rates (2010)](https://www.amjmed.com/article/S0002-9343(10)
8. [Gong -- Sales Statistics](https://www.gong.io/blog/sales-stats)

---

Published by SetLuca, the company behind Luca.