Your win rate is 35%. Your new AEs are closing at 12%. The board is asking questions. Your instinct is to hire better salespeople.
Wrong diagnosis. Wrong fix.
What’s Actually Happening
When founder-led sales stalls, the conventional playbook is to bring in a VP of Sales with “experience in the space.” The logic feels airtight: if the founder’s edge is contextual knowledge, hire someone who already has it.
But there are two kinds of knowledge here, and they’re not interchangeable.
Market knowledge (who the buyers are, how the category works, what the competition looks like) is transferable. A good VP hire brings that.
Situational knowledge (which objection means “I’m about to close” versus “I’m never buying,” which title is a real champion versus political cover, which deal signals will stall at legal) is yours alone. It lives in your head. It took hundreds of calls to build. And you never wrote it down.
That’s not a people problem. That’s a data problem.
The Framework Nobody Uses
In 2006, Stanford professor and former Veritas CEO Mark Leslie published a piece in the Harvard Business Review that most founders have never read: “The Sales Learning Curve.”
His argument: hiring more sales reps before your organization has learned how to sell efficiently doesn’t accelerate revenue, it burns cash. The learning can’t be parallelized. It unfolds iteratively, through customer contact, pattern recognition, and feedback loops. And critically: the metric that matters in the early phase isn’t revenue per rep, it’s learning yield.
You’re not trying to maximize deals. You’re trying to maximize what each deal teaches you.
Most founders do this intuitively during the zero-to-one phase. The problem is they never document it. And the moment they step back, the learning walks out with them.
What Stripe Got Right
When Patrick and John Collison were signing up Stripe’s early customers, they didn’t send a link and wait. When anyone agreed to try Stripe, they’d say “Right then, give me your laptop” and set them up on the spot.
Paul Graham later named this the “Collison Installation.” What it really was: deliberate signal generation. Every friction point observed, every question asked, every moment of hesitation was data. Data that eventually became the product intelligence behind Stripe’s famously simple developer experience. And data that, over time, powered a sales motion that scaled without the founders on every call.
Olivier Pomel at Datadog took the same approach even further. For the first six months of the company, he and his co-founder wrote zero lines of code. “Our biggest fear,” he explained, “was not to manage to build software. It was to build the wrong thing, to solve a problem that didn’t exist.” Every customer conversation was an experiment. By the time Datadog started scaling its GTM, the founders didn’t just have a product, they had a validated map of exactly what moved buyers.
Both companies scaled GTM without losing the thread of why customers bought. Most companies lose that thread the moment the founder steps back.
The Extraction Problem
If you’re still leading most of your deals at $3M to $10M ARR, you’re sitting on a body of proprietary GTM intelligence that will evaporate when you hand off. Here’s what needs to come out of your head before that happens:
Your real ICP. Not the firmographic version in your positioning doc. The trigger-based version: what was happening at the company when they first reached out? What made them willing to act now? Your best customers have more in common than job title and headcount.
Your objection map. Every objection you’ve heard lives on a spectrum from “real blocker” to “buying signal in disguise.” You know which is which. Your AEs don’t. The gap between your read and theirs is your training curriculum.
Your qualification heuristics. The signals that actually predict close are almost never the CRM fields. They’re softer: how fast the champion replies, whether legal showed up early, how the prospect described the problem in their own words. Surface these explicitly.
Your loss patterns. Ebsta’s research across 4.2 million opportunities found that 63% of deal losses happen before needs assessment. You’re probably losing deals earlier than you think, and for reasons that would show up instantly in a structured loss review.
Running GTM Experiments, Not GTM Intuitions
Once you’ve extracted the knowledge, you have to test whether it’s actually replicable. Treat each element of your GTM motion as a hypothesis, not a best practice.
Aaron Ross, who built Salesforce’s outbound engine and wrote Predictable Revenue (the book people call “the Sales Bible of Silicon Valley”), frames this cleanly: “without consistency, you have no predictability.” And you can’t build consistency from gut feel. You build it from documented experiments with clear inputs and measured outputs.
Geoffrey Moore’s Crossing the Chasm makes the same point from a different angle. The early adopters who bought from you were buying on vision. The pragmatist majority, who represent most of your TAM, buy on proof. Moving them requires a system of evidence: validated use cases, reference customers, repeatable process. Not a founder who can read the room.
This is the chasm most startups fall into. Not the product chasm, the GTM chasm. And they fall into it because they try to cross it by hiring rather than by learning.
The Measurement Infrastructure You Actually Need
None of this works without data plumbing. When you’re in every deal, you have direct observation. When you’re not, you need dashboards.
Ebsta’s data makes this concrete: companies with strong revenue operations see 87% higher win rates and 21% shorter sales cycles. The mechanism isn’t magic. RevOps creates the visibility layer that tells everyone whether the system is working. Founders stop being the monitoring system.
At minimum:
- Multi-touch attribution: which channels produce deals that close, not just leads that enter the funnel
- Pipeline stage exit criteria: what it actually means to move from discovery to proposal to commit. Without these, your pipeline is optimism in a spreadsheet
- Win/loss segmentation: by source, ICP segment, deal size, rep. The patterns here are your next quarter’s experiments
- Call recording with structured review: Gong, Chorus, whatever. Mintel used structured call review to increase win rates by 34%, by building a searchable library of top-performer calls that anyone on the team could learn from
HubSpot spent its first decade as a deeply sales-led company. Not because Halligan and Shah didn’t care about product, they did. But they understood that a scalable GTM required accumulating the same kind of market intelligence that Halligan brought to early deals, systematizing it, and only then shifting the motion. The transition to product-led growth came later, and it was staged.
Most companies try to do this in reverse.
What a Good Handoff Looks Like
The goal isn’t a 200-page playbook that lives in a Notion doc nobody reads. It’s a decision framework that tells an AE: given what you know about this account right now, here’s how to think about it.
Stage it in three phases:
- Document: the founder keeps running deals but narrates their thinking out loud. On call debriefs, in pipeline reviews, on objection-handling calls. This is your knowledge capture phase.
- Co-pilot: AEs lead deals. The founder observes and coaches, not closes. Keep this going until at least two reps can consistently hit 70%+ of the founder’s historical win rate without founder involvement on the call.
- Release: the founder steps back to strategic accounts and quarterly playbook reviews. The system runs without them.
If AEs don’t hit the milestone in phase two, that’s signal. Either the playbook has gaps, the training isn’t landing, or the hire isn’t right. You find out in weeks, not quarters.
The Bottom Line
The win rate gap is real. But it’s not evidence your AEs aren’t good enough. It’s evidence you haven’t finished building your GTM system.
Mark Leslie’s sales learning curve can’t be compressed by throwing headcount at it. Geoffrey Moore’s chasm can’t be crossed on vision alone. Aaron Ross’s insight, that Salesforce’s growth was a system not heroic selling, applies directly here.
The founder’s knowledge is a proprietary asset. Right now it’s locked in tacit form. Your job is to convert it to explicit through extraction, documentation, and deliberate experimentation before you hand off.
That’s the work. And it’s engineering, not hiring.
Build Your GTM Playbook Checklist
Before you make your first proper sales hire, ask:
- Trigger ICP: Do you know what event makes a company ready to buy, not just what kind of company?
- Objection map: Are your top 10 objections documented with the correct interpretation for each?
- Qualification signals: Are your informal closing predictors written down somewhere beyond your head?
- Loss pattern review: Have you done at least one structured analysis of your last 10 lost deals?
- Stage hygiene: Does your CRM have real exit criteria, or just stage names?
- Attribution: Do you know which channels produce deals that actually close?
- Call library: Are your best calls recorded, tagged, and accessible to the team?
If you can’t answer yes to most of these, you’re not ready to hire. You’re ready to document.
Figuring out how to make this transition without losing the momentum you’ve built? Drop me a message at angkan.mukherjee@gmail.com and happy to think through it.
References
- Mark Leslie & Charles A. Holloway, “The Sales Learning Curve,” Harvard Business Review, July 2006
- Geoffrey A. Moore, Crossing the Chasm (3rd ed., HarperCollins, 2014)
- Aaron Ross & Marylou Tyler, Predictable Revenue (PebbleStorm, 2011)
- Ebsta x Pavilion, B2B Sales Benchmarks 2024 (4.2M opportunities analyzed)
- Gong / Mintel case study: 34% win rate increase through structured call review
- Olivier Pomel, Datadog at SaaStr Europa (SaaStr, 2019)
- Paul Graham, “Do Things That Don’t Scale” (essay, 2013)
- HubSpot co-founders at SaaStr Annual (SaaStr, 2022)
