The job description says SDR, and the founder writing it usually wants something else entirely: a person who will work out where this company's customers actually come from. That is not a role you can post. It's the thing you need to know before you can post anything, and the distance between the two shows up around week eleven, when the pipeline is empty and neither of you wants to be the one to say so.

A rep is a multiplier. Aim one at a motion that works and you get more of whatever that motion already produces. Aim one at a blank page and you get activity: sequences going out, two meetings on the calendar that don't hold, a long quiet stretch. The rep decides the product is unsellable. The founder decides the hire was wrong. Neither diagnosis is right. There was nothing to repeat, and repetition is the entire skill you were buying.

Eighteen thousand dollars to answer a question you never wrote down

Use a base you'd really offer. I'll take $60,000 so the arithmetic is easy to redo with your own number. That's $5,000 a month in salary, and once employer taxes, benefits and a laptop go on top you're near $6,000 before anyone has been contacted. Now do what most founders do and decide you won't judge results for a full quarter. You've committed something like $18,000 to find out whether a motion exists. The software rides along on top of that: a contact database seat, a sending tool, a CRM license, something to handle booking. None of those bills shrink because you're pre-revenue.

Put the other column next to it. Rocketship Launch is $24.99 a month, and what it buys is the loop rather than the labor: it finds buyers, writes to them, reads and answers what comes back, and your business number is answered 24/7. Spread that $6,000 across a working month and the rep runs somewhere around $35 an hour fully loaded. A month of Launch costs less than an hour of them. The point isn't thrift. It's that one of those is paying to run an experiment and the other is paying to staff one.

Four answers you owe a rep before you hire one

You should be able to hand over all four of these in writing, without hedging:

  • Who. Not "small business owners." A title, a size band, an industry, and something that makes this month the month to reach them.
  • What gets a reply. The opening that works, the ones that got silence, and a reason to believe the difference isn't luck.
  • What comes back. The objections in the order they actually arrive, with a response to each that has worked on more than one person.
  • What a conversation is worth. A number solid enough to tell a rep how many they need and to pay them against it.

All four come out of doing the work yourself. There's no version where a new hire, three weeks in, with no product intuition and no scar tissue from a bad call, derives them for you. They'll do what they were trained to do, which is send more.

Discovery won't delegate, and the volume won't fit in your week

You are the only person who can hear "we already have something for that" and tell whether it means the door is shut or you just found the wedge. You are also a founder. Which means you cannot personally research six hundred companies, write six hundred first lines, chase the bounces, sort replies into interested and irritated, and pick up the phone at 4:10pm while you're mid-deploy.

That gap is the real problem, and hiring a body to stand in it skips the part where the body doesn't know what it's standing in.

The other way to close it is instrumentation. Searching the market, counting it, and seeing twenty-five real people by name are all free, which is usually enough to tell you whether your segment is a segment or a wish. Adding someone to your list costs one credit. A verified phone and email cost five, charged only when they're found. Launch includes a hundred credits a month, so your first hypothesis is cheap enough to throw away, and it will need to be. The worker writes, and more usefully it reads: replies come back classified, bounces get suppressed, and your sending domain gets checked so you don't mistake a deliverability failure for a market verdict. The inbound line is answered, recorded and transcribed, so what a buyer said at 11pm is still there in the morning.

The stack you'd assemble stops at every boundary and waits for you

Apollo or Lusha for contacts. Clay to enrich them. Instantly to send, HubSpot or Pipedrive to hold the record, Calendly for the booking link, Ruby or Smith.ai to answer the phone off a script you wrote before you knew what buyers say, and Zapier stapling it together. Each of those bills separately, and each one hands the work back across a boundary. The founder is the one standing at every boundary, doing CSV surgery and deduplication at midnight.

In a discovery phase that arrangement is worse than expensive. It's unreadable. Your evidence sits in five databases, so you can't follow a single buyer from list to email to reply to phone call and see where it broke. You get five dashboards that agree on nothing and a decision made on feel.

An "AI SDR" subscription is the same hire with a headshot

There's a crop of vendors selling a synthetic rep with a first name and a photo. Same category error, smaller invoice. A rep-shaped subscription still assumes someone already proved the motion, and it still sits alongside your CRM instead of inside it. Rocketship's workers run in the same place the pipeline stages, the call recordings, the quotes and the payments already live, so what you read afterward is one record rather than a reconciliation.

Hire when you're bored of the answers

Hire the rep the week you can already predict how the call goes and you're tired of making it anyway.

By then you know the list definition, you know the line that works, and you know what the third objection is going to be. A person adds leverage at that point, because they're inheriting an answer instead of a question, and you can pay them against a number that exists. If the volume has grown enough that you want meetings landing on a calendar without you touching it, Frontier at $79 adds the scheduler, a second worker and 250 credits.

Until then, keep the payroll. Run the loop yourself with something that handles the mechanical parts, read every reply personally for a couple of months, and write down what you learn while you learn it. The hire will go much better later, and you'll know exactly what to ask in the interview.