What an SDR really costs, and when an AI SDR is cheaper

This guide is for founders and sales leaders deciding whether to hire a sales development rep, add an AI SDR, or do both. It builds the cost of an SDR seat line by line from public data, turns it into a cost per held meeting, and sets it against what an AI SDR costs at the same output. The AI SDR tools comparison uses pay and benefits only; this page adds the manager, tools, recruiting, ramp and turnover, then shows where the comparison actually turns.

The cost of an SDR, line by line

Three notes on the figures:

• The benefits load is a range. The 68.5% figure covers all full-time private-industry workers. BLS's figure for sales and related occupations is 75.3%, which gives a lower load ($106,242 instead of $116,788) but includes part-time retail jobs that look little like an SDR's. Use your finance team's real load if you have it.
• Pay is at target. Only 60% of SDRs were at quota in the Bridge Group's 2025 survey, so actual variable pay is often below $25,000. Output is lower too, which the next section accounts for.
• What's left out: a CRM seat, contact data, equipment and the weeks a seat sits empty while you hire. Each makes the human figure higher, not lower.

What a held meeting costs

Output comes from the same survey: a median monthly quota of 10 held first meetings, a 3.0-month ramp and 40% yearly attrition (21% to 57% between the 25th and 75th percentiles). Attrition includes promotions, which still leave a seat to refill.

One assumption, ours: a new hire produces half a quota during the ramp. With 0.4 backfills a seat each year, that loses 0.4 × 3 months × half = 0.6 months of output, leaving 11.4 productive months.

Your first SDR costs more. In year one there's no turnover to average over: $116,788 in loaded pay, $33,303 of a manager, $1,572 of tools and a full $4,700 to recruit comes to $156,363. With three ramp months at half pace and nine at quota, that's 3 × 5 + 9 × 10 = 105 held meetings, or $1,489 each.

Where the money goes

Two things follow. The manager is a fifth of the cost of every seat, so an AI SDR comparison that leaves out management time on the human side, or the review time on the AI side, is lopsided. And attainment moves the cost per meeting far more than anything else: halving attrition to 20% only takes the cost per held meeting from $1,347 to $1,304, while falling from quota to 60% of it takes it to $2,245. If your SDRs are below quota because the list is weak, fix the list before you compare them with software.

AI SDR vs SDR: the break-even

On the published prices in AI SDR pricing, an illustrative quarter cost $309 to $439 per held meeting, including the time someone spent approving drafts. That looks like a landslide, but AI-booked meetings may convert to qualified opportunities less often. The question is how much less before the human is cheaper.

If a human SDR's held meetings convert at 50%, the AI breaks even when its conversion rate equals 50% × (AI cost per held meeting ÷ human cost per held meeting).

Read it like this: an AI SDR costing $439 per held meeting is cheaper per opportunity than an SDR at quota unless fewer than 16 in 100 of its meetings become qualified opportunities. That's a low bar on cost. The limits that bite first are elsewhere:

• Segment size. An AI SDR sends at a volume that exhausts a niche fast. The worked example below runs out of companies in about 11 months.
• The work it doesn't do. Live phone conversations, qualifying on a call and deciding which accounts deserve effort stay with people. The AI SDR guide splits the job task by task.
• Sending risk. Volume is where spam complaints come from, and a damaged domain costs more than either option.

AI BDR vs SDR: what the labels mean

For people, the usual convention is that SDRs follow up inbound leads and BDRs prospect outbound, though many companies use the titles interchangeably; see SDR vs BDR. For software, the two labels are vendor naming, not different products: Artisan calls Ava "the AI BDR", while AiSDR and others use AI SDR for the same kind of work. The Bridge Group's 2025 survey was the first in which "AI SDRs" appeared as a distinct category, at 1% of respondents. Compare tools by which tasks they do, not by the title.

There's also a middle path: people send and talk, software does the research. Bebop finds matching accounts, explains why each fits and writes a playbook for each, so an SDR's hours go on conversations rather than reading.

Worked example: a third SDR, or an AI SDR?

Illustrative example. The company and its results are invented; the segment counts are from the Census Bureau.

The team: Northfield Dispatch (northfield.example) sells scheduling software to plumbing, heating and air-conditioning contractors. Two SDRs run at 80% of quota. The choice: hire a third SDR, or put an AI SDR on contractors with 20 to 49 employees and keep the people on firms with 50 to 249.

Even with the AI's meetings converting at half the human rate, option 2 costs less than half as much per opportunity. Then the segment check: County Business Patterns counts 8,384 US establishments in NAICS 238220 (plumbing, heating and air-conditioning contractors) with 20 to 49 employees in 2023. Contacting two people at each, 1,500 contacts a month reaches 750 companies a month, so the segment is covered in 8,384 ÷ 750 = about 11 months. CBP counts establishments, not companies, and multi-site firms appear more than once, so the real number of companies is lower and the list runs out sooner.

The decision: Northfield runs the AI SDR on the 20 to 49 segment for a year with a rule against re-contacting anyone within six months, and keeps its two SDRs on the 3,861 establishments with 50 to 249 employees (2,561 with 50 to 99 and 1,300 with 100 to 249), where the deals are larger. It reviews cost per qualified opportunity at 90 days on a randomly split list, as in the AI sales tool test.

Run the numbers for your team

A. On-target earnings per SDR: $[80,000]
B. Loaded pay: A ÷ [0.685, or your real wage share] = $[ ]
C. Manager share: (manager OTE ÷ [0.685]) ÷ [SDRs per manager] = $[ ]
D. Tools per seat a year (CRM, data, sequencer, LinkedIn): $[ ]
E. Recruiting: [cost per hire] × [yearly attrition rate] = $[ ]
F. Seat cost a year: B + C + D + E = $[ ]

G. Held meetings per SDR in the last 12 months (from the CRM, not quota): [ ]
H. Cost per held meeting: F ÷ G = $[ ]
I. Share of held meetings that became qualified opportunities: [ ]%
J. Cost per qualified opportunity: H ÷ I = $[ ]

Compare J with an AI pilot's cost per qualified opportunity, measured on a randomly split list, including review time.

What to ask an AI SDR vendor before you replace a hire

• Which of my SDRs' tasks does the product do, and which still need a person: calls, qualification, hand-off notes?
• What cost per held meeting do customers in my segment see, and how did they measure it?
• How many contacts a month does a customer my size send, and what happens when the segment runs out?
• Who handles replies that need judgment, such as pricing, scope and complaints?
• Does it write a hand-off summary into my CRM that an account executive will actually read?

Next step: price the AI side at your own volume with AI SDR pricing.