Make a Mess, Then Scale: A Life Science Founder Playbook

David Nathan told me on Pathways in Life Science that the best way to grow a services business is to make a mess, then figure out how to clean it up. Most founders try the opposite, the careful path. The data on how that goes is worth a closer look.

Prefer to listen? Catch the episode Apple Podcasts or Spotify.

The “make a mess” idea has a Paul Graham pedigree

The phrase belongs to Paul Graham. His July 2013 essay “Do Things That Don’t Scale,” written from Y Combinator, made the case that the early survival moves in a business are by definition the ones you cannot keep doing forever. The famous application is Brian Chesky’s. After Y Combinator admitted Airbnb in 2009, Graham told him it was better to have 100 customers who love you than a million who only kind of like you. Chesky and Joe Gebbia bought a wide-angle camera and went door to door in New York City, photographing every host’s apartment themselves. Bookings climbed. The unscalable work was the experiment that told them where the real product-market fit was hiding. That move is the original cleanup-later move, and it travels well beyond consumer tech.

The careful path has worse odds than founders assume

CB Insights ran its 2024 postmortem on 431 venture-backed companies that shut down since 2023. The headline: 43 percent failed for poor product-market fit, 29 percent for bad timing, 19 percent for broken unit economics. Running out of cash was reframed as the symptom, not the disease. The deeper number sits at the United States Bureau of Labor Statistics Business Employment Dynamics release: 20.4 percent of new establishments close by year one, 49.4 percent by year five. Most founders never get to the careful-scaling problem, because they never get past the survival problem first. Saying no to a messy deal because you cannot yet deliver it cleanly is often a faster route to closure than saying yes and figuring it out.

Reid Hoffman built a whole framework on the same instinct

The April 2016 Harvard Business Review piece by Reid Hoffman and Chris Yeh, later expanded into the 2018 book Blitzscaling with a foreword by Bill Gates, named the move and gave it math. Their line: deliberately make decisions and commit to them with confidence substantially below 100 percent, and accept the cost of significant operating inefficiencies in exchange for the ability to move faster. Translated into a services founder’s life, that is the moment your pipeline shifts from “I should turn this away” to “I’ll take it and figure out how to deliver.” The cost of saying yes early is mess. The cost of saying no early is irrelevance.

Eric Ries’s warning still applies, just to a different layer

None of this contradicts Eric Ries. The Lean Startup, published September 2011, named premature scaling as a recurring kill pattern. The build-measure-learn loop exists to keep founders honest before they pour capital into infrastructure. The reconciliation is straightforward. Make the mess on demand, not on overhead. Take the client. Do not build the org chart, the new floor of the office, or the senior hires before the client exists. Mess on the delivery side is recoverable. Mess on the cost base is not.

The 2025 services-firm benchmark shows where the cleanup happens

Service Performance Insight’s 18th annual Professional Services Maturity Benchmark, published February 2025, covered 403 firms across information technology consulting, management consulting, software, accounting, marketing, advertising, architecture, and engineering. Billable utilization fell to 66.4 percent in 2025, 3.6 points below the 70 percent threshold the report calls minimum healthy. Operating margins held at 15 to 16 percent from 2021 through 2023, then crashed to 9.8 percent in a single year. Subcontractors rose to 10.9 percent of revenue while permanent headcount grew only 1.9 percent. Translation: the firms protecting margin in 2025 are not the ones that hired ahead in fixed seats. They are the ones that took the mess and cleaned it up with flexible, on-demand labor.

For life science services, the math compounds

A life science buyer pool is small. A scientific deal cycle is long. Word of mouth between principal investigators, biotech operations leads, and procurement carries more weight than any ad budget. That is exactly the environment where Graham’s 100-customers-who-love-you instinct returns the most. Going deep on one customer for a year seeds three referrals. Going wide on cold outreach often seeds none.

At North Star Scientific, we see the same pattern with brand partners like GENinCode, AffinityImmuno, Trialynx, and ProNique. The partners who keep the same rep on the account for three years build referral networks that take five years to dismantle. The ones who optimize for a clean org chart on day one usually pay for it later, and almost always at the worst possible time.

Want the full conversation with David Nathan? Hit play above or grab the episode wherever you listen.