CHAPTERS
- 0:00 – 0:26
Defining “terrorists” in startups: not difficult people, but value-destroyers
Dalton and Michael set the boundary for the term “terrorist”: someone (or something) that hijacks attention, creates a cloud of problems, and produces negative value. They emphasize they are not talking about borderline cases or merely demanding stakeholders.
- •“Terrorist” is reserved for extreme, obviously harmful situations
- •Not the same as hard-to-work-with or high-standards people
- •Core trait: hijacks bandwidth and produces no value
- •Founders usually “know who they are” when it’s this bad
- 0:26 – 2:13
Origin story: the “terrorist customer” who was bombing the business
Michael explains how the metaphor emerged from advising a founder emotionally attached to an unhelpable customer. The customer required bespoke work that would never pay off, pushing the startup into self-destructive behavior.
- •Some customers demand work that will never be profitable or repeatable
- •Founders may serve out of pity or emotional attachment
- •Reframing as “this customer is actively harming you” helps founders act
- •Sometimes the right move is simply to fire the customer
- 2:13 – 3:37
How terrorist customers happen: leverage, hostage dynamics, and roadmap capture
Dalton lays out the typical setup: a young startup lands a shiny big logo (or only customer) and becomes dependent. The customer then dominates priorities, often dictating roadmap and consuming support capacity.
- •Common pattern: early-stage startup + one large/only customer
- •Customer leverage can come from revenue concentration
- •Symptoms: roadmap dictated by one account; costly servicing; constant crises
- •Customer becomes the founder’s day-to-day existence
- 3:37 – 4:27
Sometimes you hand them leverage: desperation and the illusion of “good revenue”
Michael notes that founders can grant leverage even when the customer isn’t truly powerful—simply because any engagement feels precious. Dalton adds that founders are often shocked when advisors recommend walking away from revenue.
- •Desperation for replies/calls can distort judgment
- •Logo/revenue may not be as meaningful as it feels
- •Founders often resist: “But you told me to grow”
- •Revenue quality matters, not just revenue quantity
- 4:27 – 5:40
Terrorist requests: when a decent customer asks for the wrong thing
They distinguish a “terrorist customer” from a “terrorist request.” Even good customers may request money-losing projects, irrelevant bespoke work, or features that won’t actually improve outcomes.
- •Requests can be harmful even if the customer relationship isn’t
- •Examples: unprofitable projects, endless custom software, misguided features
- •Founders must judge whether work improves customer outcomes
- •Key question: what to do when a good customer asks for a bad thing
- 5:40 – 6:17
Don’t negotiate: hold the line, because ‘yes’ invites more demands
Dalton argues the founder must find a way to say no, with nuance based on the relationship. They stress that conceding once rarely ends the problem; it reinforces the dynamic and increases demands.
- •“Don’t negotiate with terrorists” = don’t default to appeasement
- •Founders must protect roadmap and priorities
- •Saying yes doesn’t end issues; it escalates them
- •The hard part is holding ground without oversimplifying advice
- 6:17 – 7:10
Customers don’t always know what they need: synthesize, don’t follow feature lists
Michael explains a common failure mode: taking customer requests literally instead of optimizing for what improves the customer’s business or life. Dalton reinforces that blindly following feature requests creates a ‘leaky bucket.’
- •Wrong model: “customer asks, I build, we win”
- •Right model: do what drives real customer outcomes long-term
- •Customers may not articulate the most important value driver
- •Founders must synthesize signals rather than obey requests verbatim
- 7:10 – 9:13
Case studies: Facebook ‘growth as a feature’ and Twitch monetization clues
Dalton recalls Zuckerberg rejecting user-research feature lists to focus on growth/network effects. Michael shares how Twitch streamers asked for features but not payment—forcing the team to infer the real need from context.
- •Facebook: the best ‘feature’ was having friends on the platform
- •User research may miss non-obvious, high-leverage priorities
- •Twitch: creators didn’t ask to get paid because it seemed impossible
- •Use context clues to uncover true needs behind requests
- 9:13 – 10:11
Survey trap: why literal feedback (like YC food requests) misleads
They joke about how surveys can overemphasize superficial wants, like better food, if interpreted via word clouds. The real insight comes from reading feedback like ‘hunting for clues’ rather than tallying popular phrases.
- •Most frequent feedback isn’t always most important
- •Word-cloud thinking can lead to optimizing the wrong thing
- •Look for subtle, high-signal comments over obvious recurring complaints
- •Founder insight requires interpretation, not simple aggregation
- 10:11 – 13:02
Terrorists from inside: recruits and employees who come with endless demands
They pivot from external to internal ‘terrorism’: candidates who look amazing but require an escalating list of conditions, and employees who consume energy while producing negative value. The warning sign is ignoring early clues due to upside fantasies.
- •Hiring desperation makes founders rationalize red flags
- •Terrorist recruits: complicated constraints and unclear commitment escalate
- •Terrorist employees: hijack everything and still create no value
- •Contrast: difficult-but-valuable people have positive net impact
- 13:02 – 14:18
The meta-pattern: you can’t finesse it—cut bait and accept sunk costs
They emphasize that negotiation rarely works; giving in just invites more asks. Founders often delay action due to sunk-cost fallacy, then later realize they should have exited sooner.
- •Appeasement reinforces the harmful dynamic
- •Advisors often identify it quickly: “cut bait”
- •Founders commonly wait months, absorb damage, then finally act
- •Key step: emotionally accepting it as a sunk cost
- 14:18 – 15:58
Terrorist investors: time sinks and faith-destroying pessimism
Dalton describes investors who consume founder time with endless questions and emotional management. Michael adds a worse variant: investors who communicate loss of faith, which can be uniquely demoralizing and destabilizing for founders.
- •Time-sink investors generate infinite questions and overhead
- •More responsiveness often leads to more demands, not less
- •Investor mood tracks company results more than founder explanations
- •Communicating pessimism can be especially damaging to founders
- 15:58 – 18:17
Belief as leadership: giving hard feedback while still expressing faith
They discuss the tension in office hours: addressing the biggest problems without crushing the founder. They cite Paul Graham’s ability to critique while genuinely believing founders could still succeed—and argue belief can’t be faked.
- •Making founders feel good isn’t the goal; truth is
- •Great advice targets the hardest problems while communicating belief
- •PG’s effectiveness came from authentic conviction, not performance
- •If you can’t believe in someone, it may be better not to advise
- 18:17 – 19:02
Wrap-up: avoid the potholes—don’t give an inch, and keep building
They summarize the categories—customers, requests, recruits, employees, investors—and repeat the core directive: don’t negotiate. Terrorists aren’t what make or break your startup; they’re avoidable potholes that waste time and momentum.
- •Terrorists show up across the startup ecosystem
- •“Don’t negotiate” and don’t expect appeasement to fix anything
- •They will take a mile if given an inch
- •These situations rarely determine success—avoid and move forward
