How do you build a repeatable sales process?
You can't systematize what you haven't done by hand first. A repeatable sales process starts with unscalable, manual, deeply personal selling — then you extract the pattern from what actually works. Most founders try to skip to the system before they understand the signal.
Start by selling one customer obsessively well
The instinct to build a scalable process from day one is exactly backwards. Before you can repeat something, you need a single successful sale you can dissect. That means finding one customer who genuinely needs what you're building, then doing whatever it takes to close them and keep them — even if those actions would never work at scale. Think of it as forensic selling: you're gathering evidence about what actually moves a buyer from skeptical to committed.
Paul Graham's argument in 'Do Things That Don't Scale' is that treating your first customers like consulting clients — giving them attention that a bigger company could never afford to give — is not a mistake to be corrected, it's a deliberate intelligence-gathering operation. The goal of that first sale isn't revenue. It's a complete map of the buying journey: what objection came up first, what language made it dissolve, what moment made the buyer actually trust you, and what final trigger got them to say yes.
Write all of this down immediately after each conversation, not at the end of the week. Memory distorts fast. What you're building in this phase is a qualitative dataset — a raw log of what happened — and that log is the raw material your eventual process gets cut from.
Recruit your first users one at a time, not through broadcasts
Early-stage founders consistently overestimate the value of a launch and underestimate the value of a direct conversation. Sending emails, posting on Product Hunt, or announcing on LinkedIn feels like selling because it's effortful — but it's actually broadcasting, which is a fundamentally different activity with much lower signal-to-close rates at the start.
Paul Graham makes the observation that 'getting users will always be a gradual process' because even great products are novel, and buyers have other things on their minds. This is not a problem to solve with better marketing copy. It's a structural feature of early adoption that you work with by going direct: email specific people, ask for introductions to specific decision-makers, show up where your buyers already are and start specific conversations. The conversion rate on a targeted one-to-one outreach is dramatically higher than broadcast at the early stage, and more importantly, you learn something from every conversation regardless of outcome.
This one-at-a-time approach also forces you to get crisp on your ideal customer profile before you've wasted budget on the wrong audience. Each direct conversation either confirms or challenges your assumption about who has this problem badly enough to pay for a solution. That refinement is the foundation of every good sales process — knowing exactly who you're targeting and why.
Extract the pattern: turn conversations into a documented process
After ten to twenty manual sales conversations — wins and losses — you have enough data to start looking for structure. The process you're building should be empirical, not borrowed from a sales methodology book. What objections actually came up? Not the ones you expected — the real ones. What questions did buyers ask that you weren't prepared for? At what stage did deals stall, and what moved them forward?
Document your pipeline stages based on buyer behavior, not seller activity. Most early-stage CRM setups are organized around what the sales rep does (sent proposal, followed up, scheduled demo) rather than what the buyer does (confirmed budget exists, brought in a second stakeholder, requested a trial). Buyer-behavior stages are more predictive of close probability and give you much better data for forecasting. Define each stage by a specific, observable thing the buyer has done — not something you've done to them.
Once you have those stages defined and populated with real deals, you can start measuring conversion rates between stages. That's where your process improvements should be targeted: if 80% of deals die between demo and proposal, that's a specific problem to solve — maybe your demo isn't addressing the right pain, or you're presenting before confirming budget. This kind of stage-specific diagnosis is only possible if your stages reflect reality rather than wishful thinking.
Avoid the common traps that stall process-building
The most common mistake is expecting a partnership with a larger company to substitute for building a direct sales process. Graham's observation that partnerships 'don't work for startups in general, but they especially don't work as a way to get growth started' is one of the most consistently validated patterns in early-stage sales. Channel deals feel like leverage but they introduce a middleman who has their own customers and quotas and will not prioritize your product unless you've already proven it can sell. Build direct first.
A related trap is what Garry Tan identifies as founders imitating the flaws of big companies — specifically, projecting indifference toward individual users in an attempt to seem established and professional. In sales terms, this shows up as sending templated outreach, routing prospects to a FAQ instead of answering their specific question, or refusing to customize a proposal because it 'doesn't scale.' At the early stage, the inability to scale is actually an advantage: you can do things your larger competitors literally cannot do, and that personal attention is often the deciding factor in a close.
Finally, resist the urge to hire a salesperson before you've personally closed at least ten to twenty deals. If you can't sell it yourself, you can't train someone else to sell it, and you won't be able to tell a bad sales hire from a good one. The founder's first job is to make the process legible enough that it can be transferred — and that requires having lived it.
When to systematize: signals that you're ready to scale
You're ready to build a formal, repeatable process when you can answer three questions with specificity: Who are your best customers (defined by role, company type, and the specific trigger that made them ready to buy)? What is the exact sequence of steps that moves a qualified prospect to a close? And what does a disqualified prospect look like early enough that you can stop spending time on them?
At that point, the systematization work is mostly documentation and tooling. Write a sales playbook that captures your qualification criteria, your discovery questions, your objection responses, and your closing moves — all grounded in what actually worked, not generic best practices. Build your CRM stages to match your documented buyer journey. Record or take detailed notes on successful calls so new reps can hear what good sounds like.
The metric that tells you your process is actually repeatable is your ability to predict close dates and deal values within a reasonable margin. That forecasting accuracy comes entirely from having clean, consistent stage definitions applied to real data. It takes longer to get there than most founders expect — typically six to twelve months of active selling — but the compounding value is enormous: a predictable pipeline is what lets you make confident hiring, marketing, and product investment decisions.
“Getting users will always be a gradual process — partly because great things are usually also novel.”
— Paul Graham, source
The one thing to do
Before you build a sales process, close ten deals manually and write down exactly what happened in each one — that log is your process.
Frequently asked questions
When should a founder stop doing sales personally and hire a rep?
After you've personally closed enough deals to document a repeatable pattern — typically 10–20 — and can explain exactly what a good prospect looks like and how the deal moves. Hiring before that point means you'll hire the wrong person and won't know it until it's expensive.
What CRM should I use early on?
The simplest tool you'll actually keep updated is the right one. A shared spreadsheet with clearly defined stage columns is better than a sophisticated CRM that nobody fills in correctly. Upgrade the tool once your process is stable enough that you know what fields actually matter.
How do I build a sales process for enterprise software with long buying cycles?
Map each stage to a specific buyer action — budget confirmed, champion identified, security review initiated — rather than elapsed time. Long cycles make it especially important to know which stage deals are actually in, because stalled deals hide in vague stage definitions.
How important is a formal sales methodology like MEDDIC or Challenger?
Methodologies are useful frameworks for structuring what you already know, not substitutes for the knowledge itself. Learn them after you've closed your first cohort of deals manually — they'll help you diagnose gaps in your existing process rather than impose a generic one on top.
Sources
- Do Things that Don't Scale — Paul Graham
- gstack: skillify/SKILL.md — Garry Tan