Going from a handful of signups to a hundred paying B2B accounts gets treated as a strategy problem, and it isn't one. Do things that don't scale. Talk to your users. Good advice, and inert at 9am on a Tuesday when the product works, eleven people have signed up, and you have no idea what you are meant to do until lunch. What's missing is a loop: four moves you run this week, run again next week, and keep running for most of a year. The moves are dull. Most founders quit at the third one.

Engineers stall on sales because it looks like an unsolved problem

You have spent years in a discipline where the right response to an unfamiliar problem is to model it and then build. So when the problem becomes get a hundred companies to pay us, the same instinct fires. You open a spreadsheet with twelve columns. You write a scraper into a Postgres table and lose a weekend to an internal dashboard you will open twice. It felt like work. It produced zero conversations.

Selling at this stage is not an unsolved problem. It has a known shape:

  • Decide which few hundred companies you are going after this month.
  • Write to a batch of people at those companies.
  • Answer everything that comes back, fast.
  • Get on a call, and then ask for money.

There is no fifth move. Founders take a year instead of a quarter because they run move one obsessively, move two in bursts, and let three and four rot, since those are the parts where a stranger might be rude to you.

Build a list you could defend to a skeptical friend

The usual failure is a target list defined so broadly that no email written to it can say anything specific. "B2B SaaS companies" is not a segment. "US companies with 10 to 50 employees who dispatch their own field technicians" is a segment, because you can picture the first line of the email.

Count before you commit. In Rocketship, searching for people, counting how large that market actually is, and previewing 25 of them by name is free. No card, no credits spent. So you can define five candidate segments on a Tuesday afternoon and find out which are too small to support a business and which are too vague to write to at all. That hour saves a month of sending into the wrong list. Founders skip it because free reconnaissance doesn't feel like effort.

Send fifty a week rather than five thousand once

The blast tempts technical founders in particular, because it presents as a throughput problem and throughput problems are enjoyable. It isn't one. A small weekly batch gives you something you can actually read. A huge one gives you a burned sending domain and noise.

So: a modest batch every week, aimed at a segment narrow enough that the message can reference something true about their situation, sent from a domain you checked first. Rocketship's AI worker writes and sends them, runs the sending-domain check before you start, and suppresses bounces so you aren't hammering dead addresses. You review the batch. You don't hand-type it.

The loop breaks at the reply, not at the send

Sending is the easy half. The hard half arrives four days later at 11:40pm, when someone writes back interesting, how does this handle multi-location? and you are eight tabs into a migration that ships tonight. You'll answer Thursday. By Thursday they have forgotten who you are.

If you fix one thing in this loop, fix how quickly the answer goes out. Your calendar is full of systems that break loudly when ignored, and a warm lead going cold doesn't page anyone. Which is why the AI worker earns its keep before the CRM does: it reads what comes back, classifies it and answers, so an interested reply gets a response in minutes instead of aging out while you're in a debugger.

Right now nobody answers your phone

Founders who dislike calls share a tell: the booking link. Anyone warm gets pushed at Calendly, the meeting lands somewhere in next week, and their curiosity has days to cool off. Meanwhile your own number rings into a voicemail box you check on weekends.

Rocketship answers that line 24 hours a day, handles after-hours, records and transcribes, and transfers to a human mid-call when it should be you talking. On Frontier at $79 the scheduler puts meetings straight onto your calendar. None of that replaces you on a sales call. It means the ten seconds between a stranger deciding to phone you and hearing a voice no longer ends in a voicemail beep.

Five subscriptions, and you are the glue between them

The standard build for this loop is five purchases. Apollo or Lusha for the data. Instantly for the sending. HubSpot or Pipedrive for the pipeline. Calendly for the booking. Ruby or Smith.ai for the phone. Then Zapier to staple them together, and you maintain the staples.

Five logins and five renewal dates is the visible cost. The invisible one is the handoff at every boundary: a CSV exported from one tool, cleaned, re-imported into the next, by the person who also writes the code. That labor never lands on an invoice, so nobody counts it. It is still why the loop stops running in week three.

Run the arithmetic before you run the loop

The acquisition side starts at $24.99 a month on Launch: one AI worker that finds buyers, writes to them and answers what comes back, your inbound line answered around the clock, and 100 credits. Adding a person to your list costs one credit. A verified phone and email cost five, charged only when they're found, so you never pay for a miss. Frontier at $79 adds the scheduler and a second worker with 250 credits. Mission Control at $149 runs three workers with unlimited apps and 500 credits, for people doing this on behalf of clients.

The app builder is separate. Free to build and publish, $12 a month on its own, so the thing you demo lives on your own domain without touching the acquisition budget. Published price, a card, live in minutes, cancel from the dashboard. There is no reason to sit through a discovery call to buy software that costs $24.99.

Doing it again next week is the entire skill

Once conversations start, put them somewhere with stages and custom fields so you can see which ones stalled and why. Send quotes from the same place. Take payment through Stripe instead of inventing a billing flow. When you hire a first salesperson, they get a seat and inherit a pipeline rather than a folder of screenshots.

None of this is clever, which is the honest part. A hundred customers is a long run of small batches, quick replies and a phone that gets answered, in the same order every week whether or not you feel like it. If you do one thing after reading this, define a segment, count it for free, and write to fifty people in it before Friday. Then you'll see what I mean about move three.