The number in the job posting is the smallest number in this decision. Here is the rest of it, line by line, with every multiplier named, plus the funnel figures we measure on a live pipeline so you can turn a monthly cost into a cost per meeting. Bring your own salary band and you can fill this in over one coffee.
Line 1: the compensation everyone quotes
A sales development rep is paid a base plus a variable, and the variable is quoted as if it were earned only on success. In practice most of it is paid, because most reps land somewhere near their number and the ones who miss badly leave before the year closes.
Public compensation work puts US SDR on-target earnings in a wide band. The Bridge Group's SDR Metrics and Compensation reports have tracked this role for years, and posted ranges on Glassdoor sit in the same neighbourhood: roughly $70,000 to $95,000 on target for a mid-market seat, with base at about two thirds of it. Pull the current edition for your own market before you commit to a figure, because the band moves every year. In Western Europe it runs lower and the employer side runs higher, which mostly cancels out.
Take $80,000 on target as the working figure. That is line one of six, and it is the line that gets quoted in the meeting where the decision is made.
Line 2: the employer load on top of it
The seat costs more than the salary before anyone opens a laptop.
In the US, employer-side FICA is 7.65% of wages up to the wage base, published by the IRS. On top of that sit unemployment insurance, workers compensation, and health cover, which is the big one. Add equipment, software seats that are shared across the company, and the share of an office if you have one.
Finance teams model this with a burden multiplier, and the common planning range is 1.25 to 1.4 times compensation. At the low end, an $80,000 seat is $100,000. At the high end it is $112,000. Pick the multiplier your own finance person uses and write the result down, because every later number in this model hangs off it.
Line 3: the tools the seat needs to work
The seat needs data and sending infrastructure to produce anything at all. The standing per-seat stack in 2026 is short and its prices are public:
- a LinkedIn Sales Navigator seat, list price around $99 a month on LinkedIn's own pricing page;
- a contact data provider with export or enrichment credits;
- a sequencing tool for email, plus domains and mailboxes that have been warmed;
- a place for the replies to land where they can be searched and counted later.
Check the current list prices yourself, they move every year. Budgeting $250 to $500 a month per seat is honest, which is $3,000 to $6,000 a year sitting on top of line two.
Line 4: ramp, and you pay every week of it
Ramp is the line most cost models leave out, and in year one it is the largest single number on the page. A new rep spends the first stretch learning the product, the list and the objections, and payroll runs the whole time. Published benchmarks for this role have put ramp at about three months for years, and the hiring managers we talk to plan for three to six.
We can put our own number next to that, because the channel has a ramp of its own, independent of who does the sending. We measured five sending profiles by invitation sequence number. The first twenty invitations from a profile returned 11 accepts out of 100, or 11.0%. Invitations twenty-one and later returned 48 out of 256, or 18.8%. Same signals, same copy, same week in some cases.
The honest reading of that gap is narrow and useful: the first twenty invitations from any new sender are excluded from every verdict we make, about the copy, about the signal, and about the sender itself. The rule earned its keep on a profile we came close to switching off in week one, which went on to produce 10 accepts on its next 37 invitations.
Read that across to the hire. Three months of ramp on a $100,000 loaded seat is $25,000 spent before the first honest data point, and the first data you get is the data least worth trusting.
Line 5: the management time that skips the budget
Somebody writes the sequences, checks the list, reviews the replies, and reads the weekly numbers. Early on that somebody is usually the founder. Two to four hours a week from a founder is the most expensive time in the company, and it lands outside every budget line.
Here is what that time buys when it goes wrong. On 6 August our own weekly review reported that accept rate had fallen by two thirds, from 37.5% in June to 12.5% in August. The recommendation that came out of it, move the sending window to the American morning, went to the top of the work list. On 17 August the same August cohort, now matured, read 15.5%. There had been no fall. We had compared June invitations that had a month to be accepted with August invitations that had a few hours.
The correction is cheap and worth stealing. Acceptance in this channel arrives with a long tail: across 136 accepts, the median came in 3.6 hours, the 90th percentile at 2.7 days, the 95th at 4.7 days. So a five day window catches roughly 95% of them, and any invitation younger than five days is excluded from every comparison. Compare cohorts of equal maturity, always by send period rather than by measurement date.
That single mistake cost eleven days of attention and the top recommendation of a review. Measurement error gets paid twice: once for the work it sends you off to do, and once for the real gap that sits still while you do it. Ours was sitting at number three on the same list, the delay between a connection being accepted and the first message going out, and it moved by zero hours in those eleven days.
Line 6: tenure, and why the model resets
The role has the shortest tenure in the revenue org. Benchmark reports have put average SDR tenure well under two years for most of the last decade, and the reasons are structural: the job is a stepping stone by design, and the good ones get promoted out of it.
Treat the model as a per cycle cost. Ramp is paid again on the next hire. The recruiter fee, if you use one, adds a further 15 to 25% of first-year salary. And the accumulated knowledge, which lists worked, which opener got answered, which objection killed three deals, walks out with the person on their last day.
Keeping that knowledge in a place the company owns is the cheapest line in this whole model, and it is the one most often skipped.
Adding it up, and turning it into cost per meeting
One US mid-market SDR seat, loaded, with tools and a normal share of management time, lands between $110,000 and $145,000 a year. Call it $9,000 to $12,000 a month, with the first three months producing ramp rather than pipeline.
A cost per year is easy to argue with. A cost per meeting is harder, so build that one. You need three conversion rates and one volume number.
From our own engine over the last 21 days, across 15 sending profiles: 1,550 outbound actions, made up of 929 invitations, 395 direct messages, 188 likes on posts and 38 InMails. On the invitations old enough to judge, 780 of them, 135 turned into connections, which is 17.3%. Of 159 accepted connections, 22 wrote back, which is 13.8%. Counting every channel together, 34 first replies have come back, so roughly 46 outbound actions per reply.
From reply to booked call our own sample is still too small to publish as a rate, so use yours. If one reply in three becomes a call, 46 actions per reply means about 140 outbound actions per meeting. Divide your monthly seat cost by the meetings that volume produces, and you have the only number that belongs on a slide.
Two warnings on doing this yourself. Every rate needs its denominator written next to it, in the dashboard and in the sentence you say out loud, because "17.3%" and "17.3% on 780 mature invitations" are two different claims and only one of them can be checked. And under 30 sends or under 3 replies in a slice, the gap between two variants is noise, so wait for the sample before you act on it.
What to do with this tomorrow
Three things, and each one holds whoever ends up doing the sending.
- Build the model with your own numbers. Salary band, your finance team's burden multiplier, your tool list prices, your ramp assumption in months. Twenty minutes, one sheet.
- Fix your measurement window before you hire. Decide now that cohorts get compared at equal maturity and that a profile's first twenty touches are excluded. Otherwise your first review of the new hire will be a coin flip presented as a verdict.
- Pick where the knowledge lives. Lists, signals, copy variants and reply classes belong in a place the company keeps. People move on by design, and the system they built can stay where it is.
The work itself is mostly sourcing lists that deserve to be written to, checking one fact per person, watching sender health, and answering replies within a day. We run exactly that loop for clients at $2,000 a month, all in, and the same numbers you just read are the ones we report on. The levels and what each one covers are laid out on the pricing page, and if you would rather query the pipeline yourself, MCP access puts it inside the assistant you already use for $500 a month.