In January 2013, Jason Lemkin, the founder of the SaaS community SaaStr, wrote a short post with a long title: When You Hire Your First Sales Rep, Just Make Sure You Hire Two. His argument is simple and a little uncomfortable.
Say you hire one salesperson and they struggle. Is it the rep? The product? The price? The market? Lemkin writes: “If your first rep does poorly, you’ll have no idea why.” Then he adds the twist. If the rep does well, you still won’t know why. Was it their skill, the product, or a lucky kind of customer?
The post is about hiring two reps instead of one, so you can compare them. But underneath sits a bigger point. A salesperson can’t tell you how your product sells. Only selling it can. And at the start, the only person who can do that properly is you.
Seven years earlier, Mark Leslie, the founder and former CEO of Veritas Software, made a related case with Stanford professor Charles Holloway in Harvard Business Review. In The Sales Learning Curve (July 2006), they argue that companies launching a new product often expand the sales force too fast. They burn cash and miss their revenue targets, because the company hasn’t yet learned how customers buy and use the product.
This post is the “how” behind that. In Your First 10 Customers I covered where first customers come from and how to run a learning conversation. Here we go one level deeper: the actual sales process a developer-founder can run, the questions to ask on a sales call, what to do with “no” and “maybe,” the math of a pipeline, and when it’s finally time to hire someone else.
Why founders must sell first
Most developers treat sales as a chore to hand off quickly. But early sales is not mainly about revenue. It’s about learning. Every sales conversation answers questions you can’t answer from your desk:
- Who actually buys? Not who you imagined. The person with the problem is often not the person with the budget.
- What words do they use? Their description of the pain becomes your homepage, your positioning and your onboarding.
- What do they compare you to? Often it’s not a competitor. It’s a spreadsheet, a chat group, or “we just live with it.” (See Your Real Competitor Is a Spreadsheet.)
- What does the price feel like? You learn whether $79 a month is a “sure” or a “let me ask my boss.” That feeds straight into pricing.
If a hired salesperson collects these lessons, they mostly stay in that person’s head. If you collect them, they change the product the same week.
There’s a second reason. A first sales hire needs something to copy. You can’t train someone on a process that doesn’t exist. Founder-led sales is how the process gets written.
So the goal of founder-led sales is two things at once: close customers, and write down how you closed them.
A simple sales process for developers
A sales process is just the list of steps between “doesn’t know I exist” and “uses my product and pays for it.” Developers can think of it as a state machine. Each lead (a person or company that might buy) is in one state. It moves to the next state only when something specific happens.
Here are six steps that work for most small B2B products (business-to-business: you sell to companies, not consumers).
| Step | What you do | It moves forward when… |
|---|---|---|
| 1. Find | Build a list of people who match your ideal customer | You have a name, a role and a reason to contact them |
| 2. Reach out | Send a short, personal message | They reply with interest |
| 3. Discovery call | Learn their situation; check if it’s a real fit | They confirm a real, current problem and agree to a next step |
| 4. Demo or trial | Show the product on their problem, or let them try it | They hit the success test you agreed on |
| 5. Close | Ask for the purchase; handle questions | They pay or sign |
| 6. Onboard | Set them up yourself until they get value | They use it on their own |
Two rules make this table useful instead of decorative.
Rule 1: a stage moves only on a buyer action. Not on your action. Sending a proposal doesn’t move a deal to “close.” The buyer agreeing to review it with their partner on Thursday might. This keeps you honest about where deals really are.
Rule 2: every open lead has a next step and a date. “Follow up sometime” isn’t a next step. “Call Tuesday at 10 to review the trial results” is.
Step 1: Find
Write one sentence describing who you sell to, narrow enough to find ten of them in an hour. “Small businesses” is too wide. “Software agencies with 5 to 30 people that bill by the hour” is searchable. Where the names come from is covered in Your First 10 Customers and Distribution Beats Product.
Step 2: Reach out
A good first message is short, about them, and asks for something small.
Hi Linh, I saw your agency ships client apps on a weekly
release cycle. I'm building a tool that turns merged pull
requests into client-ready release notes. I'm talking to
agency leads about how they write these today.
Would you have 20 minutes next week? Happy to share what
other agencies told me, either way.
No feature list, no attachment. If there’s no reply, follow up once or twice, a few days apart, with something new each time. Then mark the lead closed. Chasing silence forever just fills your list with noise.
Step 4: Demo or trial
A demo is not a tour of every feature. It’s a replay of their problem, solved. Open with what they told you on the discovery call: “You said release notes take your lead dev two hours every Friday. Let me show you that Friday with this tool.” Then show only that.
If you offer a trial (free use for a limited time), agree on a success test before it starts. Ask: “At the end of two weeks, what would need to be true for this to be worth paying for?” Write their answer down and send it back to them. A trial without a success test tends to drift: they sign up, get busy, and forget.
Step 5: Close
Closing scares developers most, because a direct question can get a direct “no.” Ask anyway. If the trial passed the success test, just say it: “The test was release notes in under ten minutes, and you’ve been at about five. Do you want to start the paid plan this month?” Then stop talking and let them answer.
Step 6: Onboard
The deal is done when they get value without you, not when they pay. Early on, do the setup yourself, the “Collison installation” from Your First 10 Customers. More in Activation: The First Five Minutes.
Discovery questions that don’t lead
A discovery call has two jobs. You learn about their world, and you qualify them: decide whether they’re likely to buy, so you don’t spend weeks on someone who never will.
The danger is leading questions: questions that point to the answer you want. “Wouldn’t it be great if release notes wrote themselves?” Everyone says yes. You’ve learned nothing.
Rob Fitzpatrick’s book The Mom Test is the best short guide to avoiding this. As summarised in Michael Lynch’s review, its three core rules are: talk about their life instead of your idea; ask about specifics in the past instead of opinions about the future; and talk less, listen more. The book also warns that once you start describing your idea, people stop describing their problems.
The Mom Test is written mainly for idea validation. On a sales call you’ll show the product eventually. But the first half of the call should follow the same rules. Here’s a set of questions that does.
Their current situation
- “Walk me through the last time you did this. What happened, step by step?”
- “Who does it today? How long did it take last time?”
The cost of the problem
- “When did it last go wrong? What happened next?”
- “What did that cost you? Time, a client complaint, a late night?”
How buying works for them
- “If you decided to fix this, who else would be involved?”
- “What do you pay for tools in this area today?”
Timing
- “Is this something you’re solving this quarter, or someday?”
And the swaps worth memorising:
| Leading question | Why it fails | Ask instead |
|---|---|---|
| “Wouldn’t it help if this were automatic?” | Contains the answer you want | “What’s the slowest part of this today?” |
| “Would you pay $79 a month for this?” | Hypothetical; easy to say yes | “What do you spend on tools for this now?” |
| “Is this a big problem for you?” | Invites politeness | “When did it last happen? What did it cost?” |
| “Do you like this feature?” | Fishing for a compliment | “When would you use this? Show me where.” |
A useful habit: say less than 30% of the words in the first half of the call. You can’t learn while you’re talking.
End every discovery call with a clear ask for a next step. Fitzpatrick’s terms are commitment (they give up something they value: time, reputation or money) and advancement (they move to the next step of your process). A booked demo with their technical lead is both. “Sounds interesting, send me some info” is neither.
Handling “no” and “maybe”
Most of your pipeline will end in “no.” That’s normal and fine. The one to fear is “maybe.”
“No” is useful
A clear no saves you weeks. Treat it as data:
- Thank them, and ask why. “That’s totally fine. Can I ask what made it a no? It helps me a lot.” People usually answer honestly once the pressure is off.
- Log the reason in a short, fixed list: no real problem, wrong person, price, timing, missing feature, went with an alternative. After 20 no’s, count them. That count is a product and pricing roadmap.
- Ask for a referral. “Is there anyone you know who deals with this more than you do?” A no for them is often a yes for someone they know.
- If it’s “not now,” set a date in the sheet, and keep it.
“Maybe” is where pipelines rot
A “maybe” feels like progress. It usually isn’t. Summaries of The Mom Test describe what happens when you don’t push for commitment: you end up with prospects who keep taking meetings and saying nice things, but never buy. One summary calls them “zombie customers.”
Turn every maybe into a yes, a no, or a dated next step:
- Name the gap. “It sounds like you’re not sure yet. What would need to be true for this to be a clear yes?”
- Shrink the ask. If a paid plan is too big, offer a smaller step: a two-week trial with a success test, or a 30-minute call with the person who decides.
- Find the real blocker. “Maybe” often means “I don’t decide” or “not a priority.” Both are fine to hear.
- Close the loop. If a lead goes quiet after two follow-ups, send a polite last message: “I’ll assume the timing isn’t right and stop following up. If that changes, just reply here.” Then mark it closed. Some people reply to this message who ignored the others. Either way, your list is honest again.
A short, honest pipeline beats a long, hopeful one. The long one lies to you about how much revenue is coming.
Pipeline math: a worked example
A pipeline is the set of all open deals, grouped by stage. Pipeline math works backwards from a goal to tell you how much activity you need at the top.
Fictional example: ReleaseNote. A solo developer-founder sells a tool that turns merged pull requests into client-ready release notes for small software agencies. It costs $79 a month. Her goal: 8 new paying customers next month. ReleaseNote and every number below are made up for illustration. They’re not benchmarks and not data from any real company, including mine.
She has run her process for two months and tracked each step in a spreadsheet. Her rough conversion rates (the share of leads that make it from one stage to the next):
| From → to | Her rate |
|---|---|
| Reached out → replied with interest | 15% |
| Replied → discovery call held | 60% |
| Discovery call → trial started | 40% |
| Trial → paid | 35% |
| Paid → still active after 30 days | 90% |
Working backwards
To get 8 paying customers, she divides by each rate and rounds up:
| Step | Calculation | Needed |
|---|---|---|
| Paying customers | goal | 8 |
| Trials | 8 ÷ 0.35 = 22.9 | 23 |
| Discovery calls | 23 ÷ 0.40 = 57.5 | 58 |
| Interested replies | 58 ÷ 0.60 = 96.7 | 97 |
| People reached out to | 97 ÷ 0.15 = 646.7 | 647 |
Check it forwards: 647 × 0.15 = 97 replies, × 0.60 = 58 calls, × 0.40 = 23 trials, × 0.35 = 8 paying. And at 90% still active after a month, about 7 of them stick.
The overall rate from first message to paid is 0.15 × 0.60 × 0.40 × 0.35 = 1.26%, or roughly 1 customer per 79 people contacted.
What the math tells her
1. The top of the funnel is the real job. 647 messages a month is about 32 every working day (647 ÷ 20). That’s a lot of personal messages for one person. It’s the honest cost of her current rates, and it’s why the next point matters.
2. Every stage is a lever, and they’re equal on paper. If she improves any single rate by half (say, replies from 15% to 22.5%, or trial-to-paid from 35% to 52.5%), the outreach she needs drops from 647 to about 423. The math doesn’t care which stage improves. Her effort does. So she asks: which rate is cheapest for me to improve?
- Low reply rate? Fix the target list and the message. Warm introductions usually beat cold messages.
- Calls that don’t lead to trials? She may be talking to the wrong people, or not asking for a next step.
- Trials that don’t convert? Weak onboarding, or no success test agreed upfront.
- Customers who leave after 30 days? That’s a product problem, not a sales one. Look at cohort retention.
3. Weight the open pipeline by its odds. Mid-month, she has 15 leads who replied but haven’t booked a call, 12 who had a discovery call and are deciding on a trial, and 9 in a trial. A lead’s chance of paying is the product of the rates still ahead of it:
| Stage now | Leads | Chance of paying | Expected customers |
|---|---|---|---|
| Replied, no call yet | 15 | 0.60 × 0.40 × 0.35 = 8.4% | 1.26 |
| Discovery done | 12 | 0.40 × 0.35 = 14% | 1.68 |
| Trial running | 9 | 35% | 3.15 |
| Total | 36 | 6.09 |
Thirty-six open leads feels like plenty. Weighted, it’s about 6 customers, roughly $481 a month in new revenue at $79 each, against a goal of 8 ($632). She’s short by about two, so she needs more outreach this week, not next month. Deals take weeks to move through the stages. Today’s outreach is next month’s revenue.
Your real rates will differ. Measure your own after the first 30 to 50 leads, and let them tell you where to work.
CRM in a spreadsheet
A CRM (customer relationship management tool) is a database of leads, contacts and deals. You don’t need to buy one yet. For your first 50 to 100 customers, a spreadsheet is enough, and it has a big advantage: you’ll design the columns from what you actually need, not from what a vendor assumed.
One row per lead. These columns cover almost everything:
| Column | Example | Why it matters |
|---|---|---|
| Company | Kite Studio | Who |
| Contact, role | Linh, tech lead | The person you talk to |
| Decision maker | Founder (Bao) | The person who says yes |
| Source | Intro from Minh | Which channels work |
| Stage | Trial | Where the deal is (one of your six steps) |
| Stage date | 2026-11-12 | How long it has been stuck |
| Next step | Review trial results | What happens next |
| Next step date | 2026-11-26 | When |
| Pain, in their words | “Friday release notes eat 2 hours” | Your future marketing copy |
| Success test | Notes in under 10 min | What the trial must prove |
| Price / month | 79 | Pipeline value |
| Lost reason | (empty, or one fixed label) | Your roadmap, once counted |
(Kite Studio and the names are fictional.)
A few simple formulas turn it into a dashboard:
Leads per stage: =COUNTIF(E:E, "Trial")
Days since stage: =TODAY() - F2
Overdue next steps: =COUNTIFS(H:H, "<"&TODAY(), E:E, "<>Lost")
Lost reasons: =COUNTIF(L:L, "Price")
And two habits keep it honest:
- Every Monday, sort by “next step date.” Anything in the past is a promise you broke. Fix those first.
- Every lead without a next step gets one, or gets closed. No exceptions. This is what stops zombie leads from piling up.
Move to a real CRM when the sheet starts failing you: several people editing, too many rows, or a need for email tracking. By then you’ll know which features you need.
When to hire your first salesperson
The common advice, and the one I agree with, is simple: close the early deals yourself before you hire anyone to sell. In a reply under his 2013 post, Lemkin says it “would be best if you closed a handful of customers yourself first.” Leslie and Holloway’s HBR article makes the cost of ignoring this concrete: a sales force hired before the company has learned how customers buy burns cash and misses its numbers.
From those sources and my own view, here are the signals that you’re ready:
- You’ve closed customers who aren’t your friends. Deals from people who already trusted you prove less than deals from strangers.
- You can write the process down. Who to target, what to say first, the discovery questions, the common objections and your answers, how a trial runs. If it only lives in your head, a new hire can’t use it.
- The objections repeat. When you can predict the next three questions, the sale has become a pattern someone else can learn.
- The numbers pay for a salesperson. A rep costs a salary plus their ramp-up time. If each customer is worth $79 a month and stays a year, that’s under $1,000 per customer. A rep would need to close a lot of them just to cover their own pay. Lemkin’s post notes that a rep only makes sense once deals are large enough to justify one. For low-priced self-serve products, the better “first sales hire” is often better onboarding and better distribution.
- Your time is the bottleneck. There are more good leads than you can handle.
And when you do hire, Lemkin’s original advice: hire two, so you can compare them and learn which part of the result came from the person and which came from the process. That’s a funded-startup move, and not every small company can afford it. If you can only hire one, keep selling alongside them for a while, so you can still tell the difference.
One more thing I believe strongly: even after you hire, stay on some calls. The learning from sales is the reason you started doing it. Don’t let it go to zero.
Key ideas
- Founder-led sales is for learning first, revenue second. You find out who buys, what words they use, and what the price feels like, and you write the process a future hire will copy.
- Run a simple six-step process: find, reach out, discovery call, demo or trial, close, onboard. A deal moves only when the buyer does something, and every open lead has a dated next step.
- Discovery questions ask about the past, not the future. Their last time, the cost, who decides, what they pay today. No leading questions.
- “No” is data; “maybe” is the danger. Log every lost reason, and turn every maybe into a yes, a no, or a dated next step.
- Pipeline math works backwards from the goal. In the fictional example, 8 customers needed 647 messages. Improve whichever stage is cheapest to fix, and weight open deals by their odds.
- A spreadsheet is a fine CRM for the first 50 to 100 customers, if every row has a next step and a date.
- Hire a salesperson only after you’ve closed the early deals yourself and can write the process down. If you can afford it, Lemkin’s advice is to hire two.
Sources: Jason Lemkin, When You Hire Your First Sales Rep, Just Make Sure You Hire Two (SaaStr, January 2013, including the author’s replies in comments); Mark Leslie and Charles A. Holloway, The Sales Learning Curve (Harvard Business Review, July 2006); Rob Fitzpatrick, The Mom Test; summaries of The Mom Test by Michael Lynch, Yevgeniy Brikman and LessWrong.
Related: Your First 10 Customers Won’t Come From Marketing, Activation: The First Five Minutes, and Distribution Beats Product.



