A monthly price for managed outbound is easy to read and hard to judge. The figure takes ten seconds. The thing being sold takes a quarter of data to describe, because what you are buying is a monthly quantity of attempts, a stack of data purchases, a sending surface with its own ceiling, and hours in which a person makes judgment calls.
This piece lays out the pricing structures you will meet in 2026 and puts a measured number against each. The numbers come from our own pipeline between 5 September and 5 October 2026: 17 sending identities, 1,623 invitations, 3,788 people sourced, 487 emails sent, 241 reply drafts released.
The five structures a managed offer uses
A managed outbound fee is built one of five ways, and each one meters something different.
Per sending identity. You pay by the number of profiles or mailboxes under management. This tracks the real constraint, because volume per identity has a ceiling that money leaves alone.
Per contact or per credit. You pay for each person sourced or enriched, which meters the top of the pipeline, where most of the volume lives.
Per meeting or per qualified reply. You pay on the outcome, so the monthly total rides on a conversion rate that takes a quarter of sending to establish.
Flat monthly, operated. One figure covers infrastructure, accounts, sourcing, copy, replies and the weekly read, and the provider absorbs the variance.
Usage pass-through. Transport such as email sends, prewarmed domains and data credits rides on the invoice at its own rate.
Most real offers mix two or three. For each line, ask what quantity it meters and what that quantity measured out to last month.
The unit every structure prices: one identity for a month
Everything downstream multiplies off one number: how many new people one sending identity can approach in a month. Between 5 September and 5 October our 17 active identities sent 1,623 invitations. The median identity sent 101, the mean was 95.5, and the band ran from 4 to 180, because new identities climb a ramp and one was pushed to its ceiling. Measured by the week since 6 August, we counted 65 identity-weeks carrying 20 or more invitations: median 45, mean 42.7, band 21 to 80. The pacing that holds an identity in that band is covered in why LinkedIn limits an account and the pacing that holds it steady.
Take the invitations that have had time to land, those sent between 5 and 28 September: 1,151 invitations, 300 accepts at 26.1%, and 52 first replies at 4.5% of invitations.
So one identity run steadily for a month produces roughly 100 invitations, 26 accepts and 4 to 5 first replies. Two identities double it. That line is the quantity under all five structures, priced five different ways.
The end of the funnel takes longer to read. In the 30-day window 125 people replied to us for the first time, and 4 of them already sit at a qualified conversation or a booked call; across the whole history of the pipeline that stage holds 18 people. A reply arriving this week gets its verdict in three to six weeks, and that lag is what a per-meeting price is quoted on top of.
Data is bought per person, and the queue is a fraction of it
A per-contact line meters the top of the pipeline, so its size depends on how many sourced people reach a first touch. In the 30-day window we sourced 3,788 people and sent invitations to 1,288 of them, which is 34%. Another 761 candidates were rejected at review, and 194 went onto exclusion lists. Our own store of purchased records grew by 7,232 people in the same month. Which sources answer which question, and what each yields per 100 companies, sits in the piece on Sales Navigator, Apollo and job boards. That ratio decides what a per-contact price costs you: a dollar per enriched contact prices the whole 3,788, and the 1,288 who got a message carry the bill.
We paid for that lesson in September. Over five days at the end of the month, 614 new people cost about 5,000 provider credits, eight a head, because enrichment ran ahead of selection and company cards were bought for people later discarded. The rule we hold now is one credit per person who reaches the sending queue: selection happens on fields that cost nothing, and the paid lookup fires when a message is queued. So ask where in the sequence the paid call sits. "At the moment of the touch" and "on everything we pull" differ by a factor of eight on the same list.
Pass-through lines belong on the invoice by themselves
Our published rate for the email channel is 6 cents per message sent. In the 30-day window we sent 487 emails, which comes to about $29, and 93 email replies came back. That figure moves with the campaign and sits apart from the monthly fee by design. The same logic covers prewarmed domains and data credits: a fee with a data allowance folded into it is a fee with a volume cap inside it, so ask for the allowance in people per month, in the units your campaign is planned in.
The part of the fee that pays for operating hours
The flat operated fee covers the work between a list and a booked call, and that work is measurable. Text checking ran 57,262 times across 13 rules in the 30-day window and raised 2,318 findings, each one a draft rewritten before a person saw it: repeated phrasing inside a batch, a missing checkable fact, a paragraph that reads as a wall, a day-of-week matching the date beside it.
Reply handling is the other half. The agent wrote and released 241 reply drafts: 163 went out when their 48-hour window closed, 59 went out because a person pressed send, and 93 of the 241 were edited first. Another 469 were dismissed in review and 49 are open. The classes a person still owns, and the ones an agent closes the same day, are set out in reply handling: classes, SLA, and what a human still owns.
Two numbers matter for the first invoice. Median time from connecting a sending identity to its first invitation was 1.9 days, the longest 18.9 days. And the first 14 days of an identity average 42.4 invitations, about one steady week, because the ramp climbs from one a day. Month one buys the setup, the warm-up and the first replies in weeks three and four.
A managed engagement also runs on items owned by both sides. In the 30-day window the client workspaces carried 56 open items: 28 on us and 28 on the client, the client side being approvals, exclusions, facts about their own market and a decision on price. Half of the open list belongs to the buyer.
Our own ladder, and the arithmetic to run on it
We publish four steps and the quantity each one meters.
- MCP, from $199 a month, on a ladder by connected identity: one $199, two $299, three $399, four $499. Your own model drives our stack and you run the campaign.
- Operated agent, $2,000 a month. The agent does the work, a person tunes it for your market and reads it daily.
- Team, $5,000 a month. Several channels, more volume, a named manager, new tests weekly.
- Fractional CRO, $10,000 a month. All of that plus the calls, the deal reviews and the strategy.
All monthly, email at 6 cents per message sent, everything built stays with you. The route-by-route arithmetic on a hire and an agency retainer is in AI SDR, agency or a hire: the cost per meeting, and the loaded seat it prices against is broken down line by line in what an SDR actually costs in 2026.
Run the numbers on the operated step. Two identities at our measured median produce about 200 invitations a month, which at 26.1% gives 52 accepts and at 4.5% gives 9 first replies. At $2,000 that is roughly $220 per first reply, before the qualified rate takes its cut. A loaded seat in the US mid-market band costs $8,300 to $9,300 a month and meets the same per-identity ceiling.
What to ask before you sign
Four questions, answerable in one call, each tied to a number above.
- How many sending identities does the fee cover, and what weekly volume do you hold them to? A band of 20 to 80 invitations per identity-week is the operating reality; a promise of 300 is a promise about someone else's account health.
- Where in your sequence does the paid data lookup fire? At the moment of the touch, or on everything pulled. The gap is eight times on the same list.
- Which reply classes go out without a person, and what happens after 48 hours of silence? Our answer: slots, links and factual questions go the same day, price and negotiation wait for a person, and a draft left alone for 48 hours sends itself.
- What is on my side of the task list in month one? An answer with no items in it means the setup will stall on an approval nobody assigned.
The takeaway is one calculation. Take the monthly figure, divide by the identities it covers, divide by 100 invitations per identity per month, and you have the price per attempt. Multiply by your own accept and reply rates, or by ours at 26.1% and 4.5%, and the quote turns into a cost per conversation you can argue with. What an agent closes inside an outbound week sits on our AI SDR page.
Questions buyers ask
How much does a managed AI SDR cost per month in 2026?
Our own ladder runs from $199 a month for one connected identity on the MCP step, through $2,000 for an operated agent, $5,000 for a team and $10,000 with a fractional CRO attached. The email channel is billed at 6 cents per message sent, which came to about $29 for the 487 emails we sent in a 30-day window.
What should a managed outbound fee include?
Infrastructure, sending identities, sourcing, copy, pacing, reply handling and the weekly read. In a 30-day window that work measured out as 1,623 invitations from 17 identities, 3,788 people sourced, 57,262 text checks raising 2,318 findings, and 241 reply drafts released. Transport such as email sends and data credits reads clearly as a separate line.
How many meetings should I expect in the first month?
Month one mostly buys setup and warm-up. Median time from connecting an identity to its first invitation was 1.9 days, and the first 14 days average 42.4 invitations against 42.7 in a steady week. Replies usually start in weeks three and four, and of 125 people who first replied inside our 30-day window, 4 had reached a qualified conversation or booked call by the end of it.
Is per-contact pricing cheaper than a flat monthly fee?
It depends on how many sourced people reach a message. In our 30-day window, 1,288 of 3,788 sourced people got an invitation, which is 34%, so a per-contact rate bills roughly three records for every one that carries a message. One run in September cost eight provider credits per new person because enrichment fired before selection, against the one credit per queued person we hold to now.