You can tell a startup has no sales process within about ten minutes of watching a demo call. The founder is brilliant. The product is good. And every conversation takes a different shape, because there's nothing underneath it.
I know because I built one of those. For the first eight months of my last company, every deal was improvised. We closed 11 customers in that stretch, and I could not tell you a single thing that connected them. Then a friend joined as our first real sales hire, asked me to write down how I sold, and I produced two paragraphs. That was the whole "process."
Building a repeatable sales process for startups isn't a paperwork exercise. It's the moment you stop guessing and start learning. Here's what that actually looks like, including the numbered rules people keep asking about.
Key takeaways
- Repeatability means any trained person can run the same steps and get comparable results — not that every deal is identical.
- You need two things before you can document anything: a written ICP and clear exit criteria for each stage.
- The 3-3-3, 30-60-90, 2-2-2 and 70/30 rules are rhythm and calibration tools, not magic formulas. They work best once you have a skeleton to hang them on.
- Test repeatability with numbers, not feelings: conversion per stage, cycle length, and win rate across at least 15–20 deals.
- Founder-led selling has to be documented and handed off, or your first hire will reinvent everything from scratch.
Why most early startups can't repeat a sale
Ask a founder what makes a prospect buy and you'll get a story. Ask two different people at the same company and you'll get two different stories.
That's the core problem. Repeatability isn't about scripts or CRM hygiene. It's about whether your organization knows why deals close. Most early-stage startups don't, and they paper over it by hiring "experienced" salespeople who then build a process around the wrong buyer.
The ICP problem nobody wants to admit
At my last company we spent months selling to anyone with a pulse and a budget. Our biggest customer — roughly 40% of our early revenue — was an accident. We'd never targeted their segment. We just happened to take a call.
When we finally mapped that customer against the deals we'd lost, the profile was obvious in hindsight: mid-size operations teams, a specific compliance burden, one decision maker. Every deal outside that profile either stalled or churned within six months. We'd been counting wins that were really refunds in slow motion.
Before you write a single stage name, do this:
- List every closed deal from the last 12 months, won and lost
- Note the pattern that separates the two groups — company size, trigger event, who signed
- Write a one-paragraph definition of your ideal customer profile you could hand to a stranger
- Delete the deals that don't fit and recalculate your real win rate
That last step hurts. My recalculated win rate dropped from what I thought was a healthy number to something humbling. But it was finally true.
How to build a repeatable sales process for startups
The process itself is smaller than the internet makes it look. Four things, in order.
Step 1: name the stages and define exit criteria
A stage without an exit criterion is just a label. "Demo" means nothing. "Demo completed, technical decision maker present, at least one next step agreed in writing" — that's a stage you can measure.
For most early B2B startups, four to six stages is enough: qualification, discovery, demo or evaluation, business case, negotiation. Each needs a verifiable condition that moves the deal forward. If your rep can't answer "is this deal past stage 2?" in five seconds, the criteria are too vague.
Step 2: instrument the pipeline
Track conversion between stages, median days in each stage, and win rate. That's it for the first quarter. I once added six custom fields to our CRM and used exactly none of them — my own mistake, and a common one. Start with three metrics and add more only when a decision depends on them.
Step 3: write the playbook from real calls
Record your next twenty founder calls. Then read the transcripts and extract what actually happened: the questions that opened people up, the objections that recurred, the language buyers used to describe their problem.
Copy that language verbatim into your scripts. Not your marketing copy — their words. When our prospects started saying "we need to survive an audit" instead of "we need better reporting," our demo close rate moved noticeably. Same product. Different words.
Step 4: run a repeatability test
Here's the check nobody published in the articles I read while building mine. Take 15–20 deals and run them through the documented process with someone other than the founder. Then compare results against the founder-led baseline. If the conversion rates land within a few points, you have a process. If they don't, you have a document.
Sound familiar? Most playbooks fail this test and nobody runs it, so nobody finds out.
The numbered rules: what they actually mean
Four rules circulate constantly, and they're often taught as universal law. They aren't. They're heuristics, and they fit specific situations.
What is the 3-3-3 rule in sales?
The 3-3-3 rule holds that you should aim for three new opportunities, three proposals, and three closes in a defined cycle — usually a month or a quarter, depending on your deal size.
It works as a personal activity target for an individual rep, or as a shape for a founder's selling week. Three new conversations, three active proposals, three signed. It's a pacing device, not a forecast. For longer enterprise cycles the numbers scale up, but the ratio stays useful: you always need more at the top than at the bottom.
What is the 30-60-90 rule in sales?
The 30-60-90 rule is a ramp plan. In the first 30 days a new sales hire learns the product, the ICP and the playbook. In days 31–60 they run real calls with support and shadowing. By day 90 they should be closing independently against a defined quota.
I'll be blunt: if you can't get a new rep to a documented, measurable state within 90 days, the problem is usually the playbook, not the hire. We hired someone who took five months to ramp, and I blamed the person. Looking back, our onboarding consisted of a shared folder and a shrug.
What is the 2-2-2 rule in sales?
The 2-2-2 rule is a follow-up cadence: reach out twice in the first two days, then twice more across the next two weeks (or two weeks and two months, depending on the version you're taught).
The point is front-loading contact while interest is warm, then spacing it out instead of spraying daily emails. It also sets an expectation — if a prospect goes silent after two rounds, that's data, and it belongs in your CRM as a stage exit, not a limbo deal that lives forever in the pipeline.
What is the 70/30 rule in sales?
The 70/30 rule appears in two common forms, and both are useful. In time allocation, it means spending 70% of your selling time on existing pipeline and 30% on prospecting new business. In pipeline coverage, it means keeping roughly 70% of your forecast in late-stage deals and 30% in early-stage, so your number isn't built entirely on hope.
In my experience the time version is the one that actually changes behavior. Founders default to 95/5 — all pipeline, no prospecting — and then wake up to an empty quarter.
| Rule | What it governs | Best applied at |
|---|---|---|
| 3-3-3 | Activity pacing | Individual rep's weekly rhythm |
| 30-60-90 | New hire ramp | Onboarding and first quota |
| 2-2-2 | Follow-up cadence | Outbound and stalled deals |
| 70/30 | Time split or pipeline mix | Ongoing pipeline management |
From founder-led selling to a team that repeats
Documentation is where good intentions die. The founder knows the process, so nobody writes it down, so the first hire improvises, so the second hire learns the improvisation as gospel.
The handoff artifact
You need one living document, not a wiki. Call it the playbook and keep it short: ICP paragraph, stage definitions with exit criteria, the five discovery questions that work, the three objections that always come up with your best answers, and the demo outline. Mine ran to nine pages in the end — and the first version was two.
Update it monthly with what the team learned. A playbook that doesn't change is a playbook nobody reads.
Metrics that prove repeatability
Three numbers tell you whether the process holds:
- Stage-to-stage conversion — if it swings wildly between reps, the exit criteria are subjective
- Median cycle length — a process that works has a predictable rhythm
- Win rate by source — reveals whether your ICP definition is holding up in reality
Watch for the sneaky one: deals that sit in the same stage for more than twice your median. Those are the deals that poison forecasts, and every startup has a handful of them sitting quietly in the pipeline.
What to do this week
You don't need a sales ops person or a new tool. You need Tuesday afternoon and a blank page.
Write down your last ten deals. Note which stage they were in when they closed, who you spoke to, and what they said the problem was. Then look for the pattern — because it's already there, buried under the improvisation. The process is less a thing you invent and more a thing you notice, then write down before it slips away again.
The uncomfortable part is that once you've written it down, you can no longer blame the market for a bad quarter. You can only fix the process. That's the trade, and it's worth making.
