The question usually arrives as a budget question. Your inbound is piling up, you cannot answer all of it, and you are choosing between paying a person to handle it and paying for software that says it will. Whichever line on the quote is smaller tends to win, which is how a lot of these decisions actually get made.
The shortcut breaks down in a specific way. A subscription and a wage are both inputs, and an input only tells you what you are spending, never what you got for it. Two tools priced within a few dollars of each other can produce very different numbers of booked calls, and so can two setters on the same retainer.
So this is the arithmetic rather than the role comparison. If what you want is the role comparison, meaning what each one is genuinely better at, we wrote that separately in human setter vs AI setter. Here the question is narrower: what does one booked call cost you under each arrangement, and which part of that calculation is usually wrong.
One warning before the numbers. We went looking for neutral data on what a DM setter costs and did not find any, so you will not find invented ranges below.
Two numbers that are not costs
Start with the wage side, because it gets quoted the most confidently and is wrong by the widest margin.
A wage is not an employer's cost. The Bureau of Labor Statistics measures both, and in its June 2026 release on employer costs for employee compensation, private industry employers paid an average of $32.82 an hour in wages and salaries plus another $14.07 an hour in benefits, so benefits came to 30.0 percent of total compensation. Whatever hourly figure you have in your head, the real line in your accounts sits roughly a third above it before anyone has booked anything.
Then there is the item that appears on no quote at all, which is your own time. Someone has to write the scripts, answer the questions the scripts do not cover, read the first weeks of messages and rewrite the ones that missed. That work exists whether you hired a person or bought a tool, and it is heaviest in exactly the weeks when you are deciding whether the choice was good.
The subscription side has a gap of its own. A monthly price buys access to a set of conversations, and the size of that set varies enormously between products whose pricing pages look alike. We come back to it below, because on most accounts it is the figure that decides the comparison.
What a setter actually costs, and why nobody can tell you
The numbers circulating online come almost entirely from the people selling the service: agency pages, outsourcing blogs, job listings. They contradict each other, none of them publishes a method, and some of the per-appointment ranges differ by a factor of five between two pages making the same claim. That is not a benchmark, so none of it is quoted here.
One official figure does exist, for the closest occupation the government actually tracks. O*NET, the US Department of Labor's occupational database, puts the median wage for telemarketers at $17.04 an hour, or $35,450 a year, drawing on 2025 Bureau of Labor Statistics wage data.
Treat that as a floor and not as a quote. The occupation it describes is outbound dialing from a list, which shares the volume of DM setting and little else: its traffic is cold, its medium is the phone, and a clumsy sentence there costs one dial out of hundreds. Holding a written conversation with a warm lead about a program worth thousands is a different trade, and the people who are good at it are paid accordingly.
What follows from the missing benchmark is not that the comparison is impossible. It is that the comparison has to be run on your own account, with your own numbers, over about a month. That version is the only one that will ever be true for you.
Three ways setters get paid, and what each one buys
Pay structure deserves more attention than the headline rate, because it decides what the person optimizes for.
Per booked call. You pay a fixed amount for each appointment that lands on your calendar. It is the easiest model to budget and the easiest to game, since the thing being paid for is a booking and a booking is not a sale. No-shows, soft yeses and leads who were never going to buy all end up on your side of the table. If you use this model, put in writing what counts: scheduled, confirmed, or attended.
Commission on collected cash. Nobody gets paid until money arrives, which lines up the incentives about as well as they can be lined up. It is harder to hire for, because a setter on pure variable pay is lending you a month of their life and will want enough volume to make that bet reasonable. If your inbound cannot feed them, the good ones say no.
A base plus a variable share. The base buys availability while the variable buys effort, which is why most working arrangements end up somewhere in here. Your cost per booked call then has a fixed half and a proportional half, and it tracks the inbound you actually have instead of the inbound you planned for.
Whichever model you pick, a booked call is not a number to take at face value either. How the appointment got made changes how often the person shows up, which is the whole argument in why sending your booking link in DMs kills your calls.
Cost per booked call, with your own numbers
The formula is short. Take everything you paid out in a month, including your own hours at whatever you consider them worth, and divide it by the booked calls that month that would not have happened without the arrangement.
That last clause is where most comparisons quietly fall apart. A lead who messaged you this morning, who you would have answered yourself within the hour, booked a call that was already yours. Crediting it to the tool or the hire flatters both and teaches you nothing. The honest denominator is incremental: appointments that exist because of the arrangement, not appointments that happened during it.
Here is what to write down for a single month, line by line.
| What to measure | Where it comes from | Easy to get wrong |
|---|---|---|
| Everything paid out | Subscription, retainer, per-booking fees, commission, employer costs on top of any wage | Quoting a wage instead of total compensation |
| Your own hours | Setup, scripts, reading messages, answering what the script missed | Treating your time as free because nobody invoices you for it |
| Booked calls | Your calendar, not a dashboard | Counting appointments that would have happened anyway |
| Attended calls | The same calendar, a week later | Reporting bookings while quietly meaning attendance |
| Reachable conversations | How many of your threads the arrangement can write to at all | Assuming the answer is all of them |
Run that for one month before you renew anything. Two arrangements at the same monthly price routinely come out two or three times apart on this measure, and the gap almost always sits in the last row.
The denominator nobody prices
Here is why that last row carries so much weight.
A tool that plugs into the platforms' official business interfaces inherits their rules about who it may write to. Meta's Messenger Platform policy describes a standard messaging window that lasts 24 hours for most user actions, stretching to 7 days after someone clicks a message ad, and outside that window a business is limited to a short list of approved update types and paid formats. In practice the reachable set is the people who wrote to you yesterday.
Your history is not in that set. Neither is the person who asked your price in March, nor the one who promised to get back to you after the holidays, nor the dormant lead who went quiet two summers ago. Those conversations are out of reach for a product built that way, however good its drafting is, and no monthly price makes them cheaper.
This is why price per month is the wrong figure to compare and price per reachable conversation is the right one. What the platform rules actually punish, as opposed to what people assume they punish, is set out in how to automate Instagram DMs without getting banned. If you want the reachable set compared product by product, read the history column in our roundup of AI inboxes for coaches.
The cheapest booked call you will ever get
Run the arithmetic backwards for a moment. The expensive part of any booked call is earning a stranger's attention in the first place, and you already paid for that with every post, ad and referral that produced a first message. A person sitting in your history is attention you bought and never used.
That is the ground Ficiel works on, which is why its cost per booked call behaves oddly compared with the rest. Kord is Ficiel's AI assistant: it rereads your conversations, finds the leads who went quiet, drafts follow-ups in your own voice, and a human always approves before anything is sent. You stay signed in with your own account, the way you are on your phone, which puts a thread from two summers ago within the same reach as this morning's message, across one unified inbox for Instagram, WhatsApp and LinkedIn.
What lands in front of you is a draft, and it sits there until you act on it. For someone who asked about pricing and then disappeared, it reads like this.
Hey Marcus, you were weighing up the 8-week program back in February and then your season started. Is now a better time to look at it?
You read it, change what you want to change, and send it yourself. Ficiel's Essentiel plan is $117/mo after a 10-day free trial, with $0 today, and it includes two seats, so your setter has their own login instead of borrowing yours. Why the last click belongs to a person, and what that protects, is the subject of keeping a human in the loop.
One honest caveat about this math. A backlog is finite, so the first month works through years of accumulated conversations and produces the lowest cost per booked call you are ever likely to see, and it will not repeat at that rate. Judge the arrangement on the second and third month, once the backlog has thinned out and what remains is the steady trickle of people who were left on seen last week.
The hybrid bill, line by line
Work the numbers honestly and the answer that comes out is rarely one side or the other. It is usually a split, because the work hiding inside the word setting breaks into parts that are not priced the same and are rarely best handled by the same thing.
| Line item | What you are paying for | What it should cost you |
|---|---|---|
| Speed on new messages | Answers within minutes, at any hour, in the same tone every time | A flat monthly tool fee, independent of your volume |
| Live conversations and closing | Judgment, tone, and the sale itself | A person, paid in a way that follows revenue |
| The backlog and the follow-ups | Rereading history, drafting, keeping track of who owes you an answer | A flat monthly tool fee, spread across every thread you own |
| Ficiel's part of that bill | The third line, with human approval on every send | Essentiel at $117/mo, or $1,170 a year, after a 10-day free trial |
Before you sign anything, settle five things: what the arrangement can actually reach, what the quote leaves out, what counts as a booking, how many of this month's appointments would have happened anyway, and who writes the last message. The fifth sounds like a detail, and it is the one that decides whether you can defend what went out under your name.
FAQ
What is Ficiel?
Ficiel is an AI inbox for coaches and setters: it surfaces old Instagram, WhatsApp and LinkedIn conversations worth following up, drafts the message, and nothing is sent without your approval. It is not an autonomous AI setter. It is AI setting with human approval, and it covers the line item above that nobody else prices, which is the backlog of conversations you already paid to start.
How much does an AI appointment setter cost?
Published prices run from a free tier up to a few hundred dollars a month, and the unit being billed matters more than the headline figure, because per AI message, per conversation, per seat and per booked call all behave differently once your volume moves. Compare what each product is allowed to reach before you compare the amounts, since a cheaper tool that can only answer yesterday's messages is working with a much smaller denominator. Ficiel's Essentiel plan is $117/mo after a 10-day free trial, with $0 today.
Is an AI appointment setter cheaper than hiring a human?
Per booked call, it depends almost entirely on how much of your inbox the tool can reach. On a large backlog it usually comes out much cheaper, because the monthly fee spreads across thousands of threads you already own. On a thin inbox with a handful of new messages a day it can be the more expensive option, since you pay the same fee against a far smaller denominator.
What is a fair commission for an appointment setter?
There is no survey-grade benchmark to point you at, and whoever quotes one is usually selling setters. What you can judge is the structure. A commission on collected cash ties the setter to the outcome you care about, while a flat fee per booking pays for a calendar entry and leaves the no-shows with you. Decide what you are buying first, then negotiate the rate against your own average deal size.
How do I work out my cost per booked call?
Add up one month of everything you paid out, including employer costs on top of any wage and your own hours at a rate you would accept from a client, then divide by the appointments from that month that would not have happened without the arrangement. Keep the numerator and the denominator inside the same month, count attendance separately from bookings, and run the identical calculation on both options before committing to either.
