A paper for founders
Founder-Led Sales and the Complex Sale
The advice to sell it yourself was written for a sale where one person decides. Many B2B founders are in a harder one, and the bill for the difference arrives in two forms: months of runway, and a confident conclusion about price or product drawn from a deal that was actually lost somewhere else.
-
93%
of startups that scaled ahead of the evidence never passed $100,000 per month in revenue
Startup Genome, n=3,200+
-
20×
faster monthly growth at the scale stage for the companies that sequenced correctly
Startup Genome, n=3,200+
-
18%
more revenue growth at companies with a defined sales process than at those without
Vantage Point / SMA, in HBR
-
28%
more revenue growth at the companies that had mastered three specific pipeline disciplines
Vantage Point / SMA, in HBR
The argument in brief
THE ADVICE: Most founder-led sales advice was written for products one person can buy on their own. A complex B2B sale has three, five, or more people in it, and every one of them is doing a private calculation about what saying yes would mean for them personally.
WHAT IT REALLY COSTS: Selling it yourself looks like the cheap option. What it actually costs is time. Until sales pays for itself, every month spent there burns runway, so the number that matters is how long that stretch lasts.
WHY IT LASTS LONGER: Closing a committee deal means working out which of those people actually carries the consequence. Then, for each one, two questions: what does this decision do to their daily working life, and what does it say about them? Their answers set how much they can tolerate right now, and whether this deal sits inside it. Knowing your product inside out will not tell you. Having sold to a few hundred committees will.
THE PART THAT COMPOUNDS: A lost deal costs you one deal. Deciding it was your price, when it was really something you never saw, costs you every decision you make after that.
WHAT WE RECOMMEND: Stay in the sale. Put someone who has run this play a few hundred times in front of it.
Every figure on this page is traced to its original publication, with its sample and its limits stated at the end. Where we are making an argument rather than citing a measurement, we say so.
THE ADVICE
The advice everyone gives
Every serious investor tells early-stage founders to sell the product themselves; that’s frequently good advice.
The current standard treatment is Pete Kazanjy’s. His Founding Sales handbook, and the playbook GTM fund published from his workshop in March 2026, set the motion out in detail: the founder personally works thirty to fifty qualified prospects before anyone is hired, because “sales is a transfer of enthusiasm” and no employee can transfer what the founder feels. The instruction for when to bring people in is crisp, and we agree with it. “Hire to execute the motion, not to discover it.”
The counsel has a long lineage behind it. Paul Graham put it plainly in Do Things That Don’t Scale: “you can’t avoid doing sales by hiring someone to do it for you. You have to do sales yourself initially.” Martin Casado of Andreessen Horowitz holds that “sales starts with the founders,” on the reasoning that “if the founders can’t sell it (or at least convince a customer they want to buy it), it’s unlikely that anyone can.” Bessemer’s playbook for the road to $1M ARR opens with the same instruction: “You are your company’s first salesperson,” and holds that founders “should be closing deals at the early stage, and at least one deal with a total skeptic.”
“If the founders can't sell it, or at least convince a customer they want to buy it, it's unlikely that anyone can.”
That counsel is presented as universal, and the reasoning beneath it is frequently sound. Nobody knows the product as the founder knows it. Nobody carries the conviction the founder carries. Prospects lower their guard for a founder in a way they never do for a rep, as Bessemer notes explicitly. And the object of the exercise is not the revenue, which at this stage is immaterial, but the learning; the founder who sells learns what the market actually wants, and no secondhand summary of that learning is worth as much as having been in those live conversations.
What follows is an argument about where that advice came from, and about why it doesn’t apply to the complex sale.
The assumption
What the advice assumes
Founder-led sales advice was calibrated for a sale in which one person decides. It is being transplanted into a problem it was never built for.
Read each source for what it is actually describing. Graham’s essay is about recruiting users one at a time, going door to door for the first few dozen; his examples are consumer and prosumer products where the user and the buyer are the same person. Bessemer’s stage table puts founder-led sales at $0 to $100k ARR, landing the first ten to twenty customers by cold outreach. Casado comes closest to our world, and his account of the difficulty is the most instructive. He calls it “a complicated discussion,” and then names what makes it complicated: “the ethos of the company, the technical landscape the product will be entering, the technical underpinnings of the product, and the strategic position and vision of the industry.” Each was describing the sale they had run themselves, which is the only sale any of us describes well.
Every item on that list is a property of the product or the market. Not one of them is a property of the decision.
The newest material has begun to close this gap. Kazanjy’s playbook gives a full section to what it calls the buying group problem, and it is right about the thing it names. Companies do not buy; people do. A single enthusiastic champion is not a decision. Founders are told to map the whole group early and to run a coordinated campaign across all of it rather than routing everything through one contact. That is a real advance on the essays that came before, and a founder who follows it will be better off than one who does not.
Mapping a buying group and reading one are different operations. Coverage tells you who sits in the decision and what each of them does. It leaves three questions untouched: which of those people personally bear the consequence of this decision, up or down; how much of it any one of them can carry this particular quarter; and whether those capacities overlap enough for any single path forward to be survivable for all of them at once. A founder can execute the coverage advice immaculately, reach every one of them, run the coordinated campaign, and still lose to a constraint that no amount of coverage was ever going to surface.
A second problem attaches to the coverage advice, and this one holds even where the advice is exactly right. Mapping a whole buying committee and running a coordinated campaign across all of it is real work, sustained over weeks, and it rewards diligence while punishing every gap. Meanwhile we are shipping product, chasing the round, managing budget and headcount, and doing the dozen things nobody else in the company can do. The follow-up this motion demands sits outside what a founder’s calendar can hold, and most of us know it before anyone says it to us.
Doing this halfway does not half-solve the problem.
The definition
Miss the one person carrying the consequence and the rest of the room does not make up for it. The deal stalls for a reason nobody in it will say out loud, and we are left reading price and product out of a silence that was never about either.
The advice understands technical complexity, and it is correct that the founder is the best-equipped person alive to handle technical complexity. It does not address decision complexity, because in the sales it was drawn from, there was none to address. One person evaluated, one person decided, and the binding constraint really was whether that person could be made to understand the product. In that world, founder-led sales is not merely good advice; it is the only sensible advice.
Our founders are not selling into that world. They are selling into a decision with three, five, or more people in it, several of whom have reasons for their positions that have nothing to do with whether the product is good. The advice arrives unchanged. The problem underneath it has changed entirely.
What makes a sale complex
A sale is complex when more than one person has a say in the decision. The moment a second person can withhold agreement, the seller is no longer persuading a buyer; they are reading a system.
The Reditus B2B Buyer Model (RBBM) exists to describe how that system behaves. Four of its ideas bear directly on the founder’s situation.
Consequence-bearing involvement is not the same as authority
The org chart tells you who signs. It does not tell you which individuals personally bear the consequence of the decision, up or down, and those are the people who govern the outcome. BANT asks after authority and gets a title; the useful question is who personally lives with the outcome, in the credit as much as in the blame.
Selfish utility and identity utility are the two axes that person is judging on
Selfish utility is what the decision does to their daily working life. Identity utility is what the decision says about them, to themselves and to the people whose regard they depend on. Both run alongside the stated business case and frequently override it, and neither is usually said out loud. Often neither is fully conscious.
Together those two set a tolerance, which is individual, situational and time-bound
Where someone sits across those two axes determines how much personal consequence they can carry right now, and that position moves with recent failures, depleted political capital, proximity to a performance review, and circumstances nobody documents and few will articulate. The operative question is never whether the case is good for the company; it is how this particular decision sits relative to what this person can tolerate today. A deal closes only when the tolerance positions of everyone bearing real consequence overlap enough to make one path forward survivable for all of them at once.
The product being good is the price of admission
The deal does not turn on whether the product is good. What determines the outcome is whether the system can sustain the decision, and reading that requires seeing what people are not saying, which is a different skill entirely from explaining what your product does.
The cost
The cost is time, not money
The founder who sells rather than hires is economizing on the wrong line. The expensive thing in early revenue is not what you spend on sales; it is how long you take to get through the phase where sales does not yet pay for itself.
This is the argument of The Sales Learning Curve, published in Harvard Business Review by Mark Leslie and Charles Holloway. Leslie was CEO of Veritas Software for more than a decade, taking it from $0 to $1.5B in revenue; he afterward lectured at Stanford’s Graduate School of Business for twenty-one years on sales organization and scaling. Holloway was a professor at the same school. The paper has been taught for twenty years.
Its claim is that revenue for a new product in a new market follows an S-curve, and that the binding constraint on the curve is organizational learning rather than sales headcount. Leslie’s formulation: the more you talk to customers and the more product and market problems you solve, the more productive your sales staff becomes. The measured variable is sales yield, meaning average annual revenue per fully trained rep. Break-even arrives when sales yield equals the fully loaded cost of a rep.
Three phases follow from that.
| Phase | Runs from | Team | What the team is for |
|---|---|---|---|
| Initiation | Product ready, until break-even sales yield | Three or four senior people | Learning: finding and fixing what the market will not accept |
| Transition | Break-even, until roughly twice fully loaded cost | Add reps of a different profile | Refining the model and the positioning |
| Execution | Once the formula is proven | Hire in earnest | Capitalizing on a motion that works |
Leslie is emphatic that the initiation team should be small and senior, that its people must tolerate ambiguity and carry learning back into the company, and that commission-based compensation is counterproductive at this stage, since the work is learning and not closing.
Hold that against a startup’s actual constraint. Initiation runs below break-even by definition, which is to say every month spent in it consumes runway and returns learning. The sum a founder saves by selling personally is real but small. Those added months of burn typically cost many times what the founder would save.
The question of who sells is not a cost question at all. It is a question about the length of the most expensive phase your company will ever occupy.
The mechanism
What sets the length of initiation
If initiation is the expensive phase, the question that matters is what governs its duration. Leslie tells us what the work requires. RBBM tells us what reading a buying committee requires. Setting the two lists beside one another is instructive.
Leslie: what initiation demands
- Deep interest in the product technology
- Facility communicating across the whole organization
- Tolerance of ambiguity
- Resourcefulness; building what does not yet exist
- Carrying learning back into engineering and marketing
RBBM: what a complex decision demands
- Reading consequence-bearing involvement, not titles
- Sensitivity to timing and context
- Comfort holding several truths at once
- Diagnosing which constraints bind, on partial information
- Reading drivers people cannot or will not name
The left column describes a founder almost exactly. The right column describes something else, and the difference is not intelligence or effort. It is experience. These are skills built through repeated contact with how different people, in different roles, under different pressures, weigh consequence differently.
Experience compresses time in ways process never can.
This is where the two frameworks meet. Leslie shows that the real price of early revenue is the months spent learning. RBBM names what shortens those months: experience reading buying committees. No framework substitutes for it, which is why the foundational paper is blunt on the point: BANT identifies budget authority but cannot distinguish formal authority from actual power; MEDDIC surfaces pain but not personal consequence tolerance; Challenger changes thinking but not the structural constraints determining whether changed thinking can produce action.
And so a founder enters initiation holding one column and not the other. They have conviction and complete product knowledge, and they have not built the pattern library required to read other people’s tolerance curves. RBBM states the consequence without cushioning it: we come to believe that if we can just explain clearly enough, action will follow. We are solving the wrong problem.
So the initiation phase lengthens, because learning by trial is slow when each sale takes months and ends without an explanation.
The compounding
The wrong lesson costs more than the deal
A longer initiation phase is the smaller half of the problem. The larger half is that the learning itself comes back corrupted, and we act on it.
Consider what happens when a deal dies in a complex sale. The champion went quiet. The second meeting never got scheduled. Nobody says why, because the actual reason is that the person who would have borne the consequence could not absorb it this quarter, and that is not a thing people say out loud. So we reason from what we can observe, and we conclude the thing that is observable: the price was too high, or the ICP was wrong, or the product was missing the feature they asked about in week two.
Then we act on it. We discount the next one. We redraw the ICP. We put the feature in the roadmap ahead of work that mattered more. Each of those is a real decision, funded with real runway, made on a reading of a system we did not see clearly.
A lost deal costs one deal. A wrong lesson costs every decision downstream of it, and it compounds, because the next deal is now approached through a model that has been bent away from reality.
The mislearning is more expensive at the early stage than the lost deal ever was. Founder-led sales in a complex environment does not merely produce slower learning; it can produce negative learning, and the founder has no way to tell the difference from the inside.
Nothing about a wrong lesson announces itself. The deal is lost either way, the explanation is plausible either way, and the pipeline report looks the same either way.
This is the work the Reditus Startup Lifecycle (RSL) is built to do. Leslie’s model offers one quantitative gate, sales yield against fully loaded cost, which is sound but arrives late. RSL runs six stages with measurable exit criteria, so that the question of whether a stage has actually been cleared is answered by evidence rather than by the founder’s confidence. Instrumentation is what makes a wrong lesson detectable. Without it, we are grading our own reading of rooms we were never able to see into.
The evidence
What the sequence error costs
WHAT WE WILL NOT SHOW YOU
This is the point at which most firms making our argument reach for numbers that do not exist. There is no study tracking two matched cohorts of B2B startups, one selling by founder and one with seasoned help, and comparing their revenue curves. Any chart purporting to show it was invented. What follows is what genuinely has been measured.
The cost of getting the sequence wrong is documented. Startup Genome studied more than 3,200 high-growth technology startups and found that roughly 70% scaled some part of the business ahead of what the evidence supported. The study’s definition of premature scaling covers spending and hiring broadly rather than sales specifically; we cite it for the magnitude of the penalty attaching to a sequencing error, which is what it measures well.
-
93%
of startups that scaled ahead of the evidence never passed $100,000 per month in revenue
Startup Genome, n=3,200+
-
20×
faster monthly growth at the scale stage for the companies that sequenced correctly
Startup Genome, n=3,200+
-
3×
larger teams among prematurely scaled companies at the equivalent stage
Startup Genome, n=3,200+
The value of a formal sales process is documented separately. Vantage Point Performance and the Sales Management Association surveyed B2B companies on pipeline and process discipline, and Harvard Business Review reported the result. The sample was 62 companies and skewed toward the enterprise, so we present it as evidence that process discipline carries measurable revenue consequence, not as a startup benchmark.
-
18%
difference in revenue growth between companies that had defined a formal sales process and those that had not
Vantage Point / SMA, in HBR
-
28%
difference for those that had mastered three specific pipeline practices
Vantage Point / SMA, in HBR
Neither study proves our thesis on its own. Together they establish the two premises the argument needs: that sequencing errors in early revenue are severely punished, and that disciplined selling is worth a material amount of growth. The bridge between those premises and the founder’s decision is the Sales Learning Curve.
Objection
But product-led growth avoids all of this
It does, exactly as far as one person can buy.
Product-led growth works when an individual can purchase and reach value without committing the organization. Where that holds, there is no committee, no distributed consequence, and no tolerance curve but the user’s own; the product genuinely does the selling, and everything above is beside the point.
The difficulty is that the condition rarely holds for long. The moment a second person with consequence-bearing involvement must agree, whether for the security review, the budget line, the integration, or the seat count, the transaction has become a complex sale by definition. It does not matter that the product has a self-serve tier. What is being transacted is organizational commitment, and organizational commitment is not something a credit card form can carry.
So PLG does not remove the learning curve in these cases. It defers it to the point where the revenue actually lives. The PLG cohort’s own numbers say as much.
-
97%
of product-led companies either have a sales team or plan to add one
Pocus, 200+ respondents
-
7%
permit users to self-serve onto an enterprise plan
Pocus, 200+ respondents
Among the companies that built their entire go-to-market on the premise that the product sells itself, ninety-three times in a hundred the enterprise decision is still transacted by people.
Objection
But AI closes the gap
AI does a great deal of this work, but it does not do this part of it.
AI can analyze patterns, surface correlations, anticipate objections, research an account, and optimize messaging, and it can do all of it faster than any team could unaided. Every one of those is information processing, and information processing is genuinely being transformed.
Reading a tolerance curve is not information processing.
Tolerance is shaped by the lived experience of personal consequence. It moves with recent failures, with depleted political capital, with proximity to a review, with changes in personal circumstance that are rarely documented and almost never articulated. Reading those dynamics requires having experienced what it feels like when a decision you championed fails visibly, when credibility is burned by pushing too hard, when blame lands unevenly for someone else’s risk.
AI has never borne consequence. It has never had its judgment questioned, its reputation damaged, or its position threatened by an outcome it could not control.
The Reditus B2B Buyer Model, foundational paper
That is not a gap in training data, and it cannot be learned from examples; it is a difference in kind. Pattern recognition is part of it, and a library of patterns is exactly what a seasoned operator carries. The difference lies in where those patterns are read from. They come off what people decline to say, and recognizing them takes having once been the person declining to say it, since what gets held back is status, fear, ambition and self-protection.
Complex selling therefore remains human work for as long as buying remains human. Not because people compute better, but because the work requires reading consequence, and only an agent that has borne consequence can recognize it.
Because AI raises the volume of activity a founder can generate, it also raises the rate at which a founder can act on a misread of the committee. Speed applied to a faulty model of the room is not progress along the curve. It is the wrong lesson, arriving faster.
The recommendation
What we actually recommend
We do not want founders to step back from the sale. We want them further into it, just not leading it.
Everything the investment community says about why the founder belongs in the sale is correct, and nothing in this paper disputes it. The founder carries the conviction. The founder knows what the product will and will not do, and can say so without checking. The founder can commit the company to a change in the moment, which no employee can. The founder hears the market firsthand, which is the entire point of the exercise, and that learning cannot be delegated without being degraded.
Our position is narrower, and it concerns who leads. The initiation phase asks for two distinct things at once, and no single person at an early-stage company holds both.
What the founder holds
- Product depth, and the authority to say what it will and will not do
- Conviction that no employee can manufacture
- Organizational reach, and the power to commit the company in the moment
What the seasoned operator holds
- Which of the six people present will actually bear the consequence
- Whether the silence in the room was doubt or arithmetic
- Whether what just happened was a delay or a decision
So expertise leads the sale, and the founder stands beside it.
In practice that means the operator runs the deal, reads the buying system and decides the intervention; the founder holds the product and the vision and is present for both. The founder still learns the market firsthand, which was always the real argument beneath founder-led sales, and it never required the founder to be the one leading. What changes is that the learning is now read correctly, because someone present has seen this system several hundred or thousand times before.
The alternative
Why not simply hire a sales leader
Because the full-time hire asks the company to make its largest bet at the moment it knows least.
Bessemer advises against making a VP of Sales the first hire at minimal revenue. Casado advises against bringing one in before repeatable success has been demonstrated. Both are right, and for Leslie’s reason: during initiation there is no proven motion for a sales leader to run, and asking someone to build one while carrying a number confuses two different jobs.
The tenure data says the same thing.
-
25mo
average tenure of a chief revenue officer, among the shortest in the C-suite
Harvard Business Review, October 2024
-
32%
annual turnover in the CRO and CMO seats, against 4.3 years' average tenure for a chief executive
Pave, 14,000 executives
-
62%
of companies see revenue growth decline or stay flat in the fiscal year after a CRO change
Harvard Business Review, October 2024
Add the months to source and hire, and the months to ramp, and a substantial share of that tenure is consumed before the person is at full effectiveness. Then weigh the third figure, which is the one that ought to give a founder pause. Companies that changed revenue leader saw growth fall from an average of 15.5% the prior year to 11.7% the year following, a drop of nearly four percentage points. Getting the hire wrong does not merely waste the cash and equity committed to it. It takes growth out of the year that follows, at a company that had none to spare and a motion nobody had yet proven exists.
Leslie argues that commission-based compensation is counterproductive during initiation, because it rewards closing when the work is learning. A leader compensated to produce bookings from an unproven motion will produce activity, and activity inside a system whose constraints have not been diagnosed is precisely what RBBM warns against; it raises volume without changing viability, and it generates exactly the lost deals from which the wrong lessons get drawn.
The fractional structure exists to resolve this. It supplies the pattern library during the phase that needs it most, without asking the company to commit cash and equity to a permanent bet before the motion is known. When the motion is proven and the curve has turned, the full-time hire becomes the right decision, and the right fractional executive should help you hire their replacement.
The frameworks
What this paper is built on
Two Reditus papers carry the argument above. The buyer model explains why complex deals behave as they do; the lifecycle is how a company tells where it actually stands.
Sources
Sources and their limits
Every figure in this paper is traced to its original publication, with its sample and its limits stated. Where we are making an argument rather than citing a measurement, we have said so in the text.
Leslie & Holloway, The Sales Learning Curve, Harvard Business Review, July 2006
Used for: The S-curve; sales yield; break-even against fully loaded cost; the three phases; small senior initiation team; commission counterproductive in initiation
Sample: Practitioner model from Leslie's tenure as CEO of Veritas Software, developed with Stanford GSB
Limits: A model, not a measured dataset. Predates product-led growth. Says nothing about buying committees, and does not address who should lead
Startup Genome, Premature Scaling
Used for: 70% scaled prematurely; 93% of those never passed $100k/month revenue; 20× monthly growth gap at scale stage; teams 3× larger at equivalent stage
Sample: 3,200+ high-growth technology startups, classified by machine learning against stage-consistency criteria
Limits: Published 2011. Premature scaling is defined broadly across spending and hiring, not sales specifically. Cited for the magnitude of a sequencing penalty
Vantage Point Performance & Sales Management Association, reported by Jordan & Kelly in Harvard Business Review, January 2015
Used for: 18% revenue growth difference where a formal sales process was defined; 28% for three mastered pipeline practices
Sample: 62 B2B companies surveyed
Limits: Small sample, skewed to enterprise. Correlation, not a controlled comparison. Not a startup benchmark
Pocus, Product-Led Sales Benchmark Report
Used for: 97% of PLG companies have or plan a sales team; only 7% allow self-serve onto enterprise plans
Sample: 200+ respondents at product-led companies
Limits: Self-selected respondents at a vendor's survey. Cited for how the PLG cohort behaves in broad terms rather than for precision
Toman, Kurey & Lingebach, The High Costs of Chief Revenue Officer Turnover, Harvard Business Review, October 2024
Used for: Average CRO tenure of 25 months; 62% of companies see growth decline or stay flat the fiscal year after a CRO change; average growth falling from 15.5% to 11.7%
Sample: Not disclosed in the published article
Limits: States its findings without publishing sample or method. Cited for figures HBR has put its name to, not for a method a reader can inspect. Describes companies past the earliest stage
Pave compensation database, reported by SaaStr
Used for: Average CRO tenure of 1.8 years and 32% annual turnover, against 4.3 years for a chief executive
Sample: 14,000 executives in Pave's real-time compensation database
Limits: Reported by SaaStr rather than published by Pave directly. Skews toward venture-backed technology companies that use Pave
Pete Kazanjy, Founding Sales, and the founder-led sales playbook published by GTMfund from his workshop, March 2026
Used for: The current standard statement of the founder-led motion: thirty to fifty prospects before hiring, “sales is a transfer of enthusiasm,” “hire to execute the motion, not to discover it,” and the buying-group section we engage with directly
Sample: Practitioner guidance drawn from Kazanjy's own operating experience. The stage targets it cites carry no external source
Limits: Quoted to characterize the prevailing advice at its strongest, not as evidence for or against it. Our disagreement is narrow and stated in the text
Graham, Do Things That Don't Scale; Casado, a16z; Bessemer
Used for: The lineage of the founder-led sales counsel, quoted directly
Sample: Practitioner and investor guidance
Limits: Quoted to characterize the prevailing advice accurately, not as evidence for or against it. Each addresses the sale its author was describing, and none claims to cover a committee decision
The Reditus B2B Buyer Model, foundational paper, 2026
Used for: Consequence-bearing involvement; tolerance curves; identity and selfish utility; the limits of BANT, MEDDIC and Challenger; why AI cannot read tolerance
Sample: Reditus Group
Limits: Our own framework. Presented as argument, not as independent evidence
The Reditus Startup Lifecycle
Used for: Six stages with measurable exit gates; the instrumentation that makes a wrong lesson detectable rather than invisible
Sample: Reditus Group
Limits: Our own framework. Presented as argument, not as independent evidence