Things That Don't Scale: How Startups Win First Customers

TL;DR: Doing things that don't scale—manual onboarding, door-to-door research, founders as support reps—is how most successful startups earn their first customers and find product-market fit. Based on 200+ founder interviews on the PMF Show, founders who did 100–500 manual customer conversations before automating consistently outperformed; Legion surveyed 500 businesses door to door before writing serious code. Do the unscalable work first, then systematize what proves out.

After 200+ Founder Interviews: Why the Unscalable Stuff Wins

After interviewing 200+ founders on the PMF Show, the pattern behind "do things that don't scale" is far more concrete than the Paul Graham essay that named it. The founders who reached product-market fit fastest did shockingly manual work—walking door to door, answering support personally, doing customers' jobs for them by hand—and they did it deliberately, as a learning engine. As Chris Saad, startup advisor and co-host of The Startup Podcast, puts it: startups are learning machines, and unscalable tactics are the fastest way to learn. Here are five founders' playbooks, with the numbers that show why the manual phase pays off.

How Many Customer Conversations Are Enough Before You Build?

The Legion Story: 500 Businesses, Door to Door

Sanish Mondkar, founder and CEO of Legion, wanted to fix hourly-workforce scheduling. Before building, he created a structured questionnaire and hired one person in New York to literally go door to door to local businesses. Over a three-to-four-week period, they collected structured data from over 500 businesses.

"It was a very scrappy day, so I just basically hired a person in New York and she would go... literally door to door and say, can I ask you a few questions? I'm doing a survey. And actually, that worked so well. We had like over 500 businesses that we talked to and very well-organized data over a three, four week period." — Sanish Mondkar, Legion

According to Sanish Mondkar, CEO of Legion, the scale mattered: a manager with 25 hourly employees fields roughly 50 schedule-change requests per week, and it took hundreds of conversations to see that pattern as an industry-wide, software-solvable problem rather than an anecdote. As shared on the PMF Show, founders of very successful businesses repeatedly report 100, 200, even 500 early conversations—an order of magnitude more than intuition suggests.

Key stat: Legion collected structured survey data from 500+ businesses in 3–4 weeks—door to door—before committing to its product thesis.

What Does Unscalable Selling Look Like in Practice?

The VanHack Story: Closing Deals Live From a Slack Group

Ilya Brotzky built VanHack, a marketplace bringing global tech talent to Canada, on pure outbound hustle: cold calls, cold email, LinkedIn, meetups, hiring fairs. His signature move was gloriously unscalable. Mid-conversation with a hiring manager, he'd post the role into VanHack's Slack community—live.

"I would go to the Slack group that we have and say, hey, I'm here with the CEO of X company, and they're looking to hire a Ruby on Rails developer with 10 years' experience... Then maybe 10, 15 minutes later I'd be like, hey, let's check if anyone commented on the post... There'd be 10 people who'd say, hey, me, me, me, me, me." — Ilya Brotzky, VanHack

According to Ilya Brotzky, CEO of VanHack, showing real candidates responding in real time beat any pitch deck. The same bias for revenue over process showed up when fundraising failed after Techstars Berlin: the team simply focused on customers and generated more than €500,000 in revenue in the final three months of that year—unscalable selling literally replaced a failed half-million-euro round.

Key stat: After failing to raise €500K post-Techstars, VanHack closed €500K+ in revenue in three months through founder-led outbound.

Should Founders Personally Do Customer Support?

The Astronomer Story: Four Calls a Day and Used-Bookstore Gifts

Viraj Parekh, co-founder at Astronomer, describes the company's first managed-Airflow customers as "a bit of a knife fight." Their first paying customer—a data scientist at an Irish media company—got founder attention no scaled company could ever offer: "I think I had talked to him no less than four times a day for the first two months."

The unscalable creativity extended to brand-building on a budget. Too broke for company t-shirts, the team researched what each customer majored in and brought them a used book on that topic instead:

"Because we didn't have a bunch of Astronomer t-shirts we could give them... we'd look up who we're visiting, find out what they majored in in college, and we'd go to the used bookstore and get a book about that topic... It showed that we cared more than the other people that we're talking to." — Viraj Parekh, Astronomer

According to Viraj Parekh, that intensity wasn't waste—it was engineering the second customer's experience: "make sure that you are doing your best so the second customer doesn't run those same roadblocks. There's no substitute for elbow grease there."

Key stat: Astronomer's founders spoke with their first customer up to 4 times a day for two months—then converted those learnings into a smoother path for every customer after.

Can Doing the Work Manually Actually Build a Better Product?

The Quanta Story: Engineers as Bookkeepers

Helen Hastings, founder and CEO of Quanta, built an automated accounting product by first doing the accounting by hand—with her engineers. In 2024, the team ran what they called "engineers as bookkeepers": engineers manually reviewed every credit card purchase and reimbursement while simultaneously building the automation.

"All the engineers were going through and looking at all the credit card purchases, all the reimbursements, and kind of figuring out what they were manually as we were building... I really think that is the best way to build an automated product. Because the engineers have to understand the domain very, very deeply." — Helen Hastings, Quanta

According to Helen Hastings, CEO of Quanta, the company deliberately delayed its public launch until the beginning of 2025, when the automation could absorb volume—"we felt ready enough that we could take on a lot of customers and not fall over." Her discovery process was equally manual: a funnel of open-ended conversations starting in 2022 ("what are your day-to-day pain points?") long before code. The manual phase wasn't a compromise; it was the spec.

Key stat: Quanta's engineers manually processed customers' bookkeeping through 2024, converting each manual edge case into automation before scaling customer intake in 2025.

When Do Things That Don't Scale Stop Working?

The Tenex and Solidroad Stories: Systematize What Proves Out

Eric Foster, CEO of Tenex (cybersecurity), shows the transition point. His first deal—a top publicly traded financial institution—closed on founder-led sales and three decades of trust: one early customer had bought from him four times across four different companies. His rule, as shared on the PMF Show:

"I'm a huge proponent of founder-led sales and I think way too many people start a company and they don't do founder-led sales... All of those early adopters, they're buying the founders as much as they are the product." — Eric Foster, Tenex

Mark Hughes of Solidroad shows the same arc on customer success: when a customer, Podium, went quiet, Solidroad doorstepped their Utah office and spent three days meeting every single user. Usage jumped—so they scaled the tactic into a repeatable in-person activation motion that helped drive 186% net revenue retention. The graduation rule from both stories: keep the unscalable tactic exactly until you understand why it works, then build the system that captures 80% of its effect. Founder-led sales becomes an AE team the way Solidroad's doorstepping became structured onboarding—informed by the manual reps, not replaced before them.

Key stat: Solidroad turned a 3-day in-person rescue of one silent customer into a systematized activation motion behind 186% NRR.

Key Takeaways: Things That Don't Scale for Startups

1. Volume of conversations beats quality of guesses. Legion's 500 door-to-door surveys, in 3–4 weeks, produced conviction no focus group could.

2. Sell live, not with decks. VanHack closed hiring managers by summoning 10 real candidates in Slack mid-meeting—proof beats promises.

3. Founder-led support is product research. Astronomer's 4-calls-a-day customer produced the roadmap that smoothed every later onboarding.

4. Do the job by hand before you automate it. Quanta's engineers-as-bookkeepers approach embedded domain depth directly into the product.

5. Founder-led sales is non-negotiable early. Tenex's first enterprise deal closed on decades of trust; early adopters buy the founder as much as the product.

6. Show up physically when digital fails. Solidroad's doorstepping of a silent customer became the seed of a 186% NRR motion.

7. Graduate deliberately. Systematize an unscalable tactic only after you understand its mechanism—then keep the 80% that transfers.

FAQ: Common Questions About Things That Don't Scale

Q: What does "do things that don't scale" mean for startups?

A: It means manually doing work—sales, onboarding, support, even the product's job itself—that can't possibly serve thousands of customers, because it's the fastest way to learn what customers need and to win your first 10–50 customers. Legion, Quanta, and Astronomer all built scaled products from deliberately unscalable beginnings.

Q: How long should a startup do things that don't scale?

A: Until the mechanism behind the tactic is understood and demand is proven—typically through your first dozens of customers. Quanta stayed manual through all of 2024 and launched publicly only when automation could handle volume; Solidroad systematized its in-person motion only after seeing its impact on usage.

Q: What are examples of things that don't scale that worked?

A: Door-to-door surveys of 500 businesses (Legion), founders answering support 4x a day per customer (Astronomer), engineers doing customers' bookkeeping by hand (Quanta), closing sales by pinging a Slack community live (VanHack), and flying to a customer's office uninvited for three days (Solidroad).

Q: Do things that don't scale work for enterprise startups?

A: Yes—arguably better. Tenex closed a top financial institution through founder-led sales, and Solidroad's in-person activation drove 186% NRR with large accounts. Enterprise buyers reward founder attention and trust that no scaled process replicates.

Sources: Listen to the Full Founder Stories

• Sanish Mondkar, Legion — 500 door-to-door business surveys and the scheduling pain that justified the company.
• Ilya Brotzky, VanHack — live Slack demos, pure outbound hustle, and revenue replacing a failed raise.
• Viraj Parekh, Astronomer — knife-fight first customers, 4 calls a day, and used-bookstore brand-building.
• Helen Hastings, Quanta — engineers as bookkeepers and delaying launch until automation was real.
• Eric Foster, Tenex — founder-led sales and early adopters who buy the founder.
• Mark Hughes, Solidroad — doorstepping Podium and turning it into a 186% NRR activation system.

Last updated: July 2026