Surviving Twitter's Growing Pains: Ex-CEO Dick Costolo
In a Nutshell
Dick Costolo, Twitter's ex-CEO, shares a masterclass on scaling hypergrowth companies by accelerating decision velocity (e.g., "If it takes six weeks, how to make it two?"), replacing vague principles with DRIs (Directly Responsible Individuals) and bias-to-yes, and using management by walking around late nights to align teams and communicate context. He stresses pushing decisions down without "Dick said" excuses, giving blunt feedback (write it out, deliver unfluffed, sit in silence), strategic forestry management over firefighting, hiring/firing deliberately (slow hire, fast fire), and regrets like missing Instagram acquisition. Core lesson: Focus ruthlessly on few priorities, ensure everyone grasps the "why" and personal stakes, and lead authentically without copying archetypes like Jobs.
These notes were generated by AI and may contain inaccuracies.
Twitter was the drama queen of hyper growth companies. When Dick Costolo took over as CEO, the Twitter bird had hanged itself and exploded on two different magazine covers. He told head of communications Gabriel Stricker, "Great news. Next year's goal, no dead birds on magazine covers."
Dick Costolo was Twitter's third CEO: first Jack Dorsey, then Ev Williams. Costolo joined as COO then became CEO, running it for 5 years. Twitter created a category, grew extremely fast, but had no revenue model and was famously dysfunctional. Costolo acted as the adult in the room, running a scale-up CEO master class. Key topics: increasing velocity of decisions and execution as companies scale; management by walking around; giving tough feedback to tough people; dealing with press; focus in the AI era; managing the forest instead of fighting fires. Regrets include not acquiring Instagram.
Costolo joined in summer 2009 after selling FeedBurner to Google, where he reconnected with Twitter founders from Blogger preparing to start Odeo (which became Twitter). Ev Williams pitched him initially for interim COO during paternity leave, then permanent COO. Twitter had zero revenue model despite hype (Ashton Kutcher and Oprah Winfrey fighting for 1 million followers). Costolo saw it as a rocket ship opportunity.
Twitter was chaotic: decisions made collectively as a group (e.g., debating native retweet vs. manual RT). Things moved methodically and slowly. The process of pushing Ev out and installing Costolo as CEO was awful—happened super quickly yet slowly and painfully. Costolo was informed in driving rain on the road home by Evan and Jason that the board decided he should be CEO. Bill Campbell called advising him to quit. Board reversed, then invested in his options and suggested he leave. Over days and weeks, Bjan Sabay from Spark texted him not to quit. Finally, board insisted he take CEO role.
Priority 1: Velocity. Push for faster cadence: "If this takes six weeks, what would have to be true for it to be two weeks?" No more group voting on decisions. Example: Status line changed from "What's going on" to "What's happening" without asking front-end engineer Bret Taylor, who complained; Costolo said, "We're not doing that anymore."
Priority 2: Replace meaningless operating principles. Removed whiteboard principles like generic "integrity and honesty" that didn't aid decisions and were weaponized against management.
Cultural cleanup: Sent email about dirty kitchen dishes: "This is why the site crashes—everything's messy, someone else will clean it up." Removed poster "We'll make better mistakes next time," indicative of complacency.
Founders didn't push back.
As companies hypergrow, they slow down due to organizational barnacles. Sub-200 people tactics:
- Push decisions down the stack: Leaders tell teams to figure it out; don't come to Costolo unless no idea.
- Avoid "Dick said" as reason: Example, Android app boot time slowed usage; engineer asked if Costolo said "remove JSON file"—managers weaponized CEO's name to override pushback.
Bias to Yes (from improv background, similar to Bezos' bias for action): Only direct manager or legal (for illegal/privacy violations) can block; no other orgs require approval. Enabled experiments to fly: Engineers built wireframes themselves for 1% tests without design. Repeatedly explained at all-hands: Doesn't mean ignore priorities—managers measure and can fire for that.
Don't solve problems with processes: Avoid adding steps (e.g., Google's 17-page launch checklist). Use DRIs (Directly Responsible Individuals), operating control, span of control, decision-making—not processes. Manage to outcomes, not checklists (e.g., TSA: ensure no bomb, not no water bottle).
Implication: Can't punish mistakes. Leadership's job: "Not to prevent mistakes, but to correct them quickly." Tell leadership immediately to fix; no one gets in trouble for screwing up.
Under 50 employees: Everyone knows what's going on. Over 150: Nobody does. Fixes:
- Make sure everybody understands what you understand (from Ben Horowitz's management class): Managers ensure teams know priorities, how they fit, personal/professional implications if achieved.
- Example: Tesla factory worker explained widget role in making EVs popular; bricklayer analogy: "Laying bricks" vs. "Building a cathedral."
Good news travels fast, bad slow—worse as org grows, signal distorts. Communicate problems up constantly. Repeat priorities at all-hands: What, how measured, why important, company implications.
Steve Jobs trick at Pixar (one day/week while fixing Apple, during Toy Story 2): Met illustrators without director John Lasseter/Lacier. Asked: "Tell me something not working well at Pixar" (not movie/job-specific), then "something working well." Summarized: "Here's what I heard, I'll talk to John Monday."
Never use "because so-and-so said"—lacks context. Combination: Push decisions down, ensure understanding/context, communicate issues up for quick resolution.
Leaders take notes during meetings with teams, ensure they hear everything, then follow up with the team leader like Lester. Started doing this with teams to check if they know priorities and are in sync with leadership. Some leaders hated it, raising concerns about why. In meetings, respond with, "Okay, here's what I heard. I'm going to go talk to my management team about it on Monday, and Chris or whoever will follow up with you guys."
CEOs have priority lists, but opportunities arise. Rejigger priorities once or twice a year, not regularly. With four huge priorities for the first four to five months, can't drop two in February. Regular changes lead to hesitation: "We can get to work on the direct messaging architecture but Dick's probably going to change it 50/50." Companies slow because people won't bust their ass if there's a 50% chance it'll be thrown out in a month.
Hypergrowth companies do more in parallel because software deployment is faster; many one-way doors are now two-way doors. Focus is never overrated. With two-way doors, do more quickly since fewer existential decisions exist—only take long time on decisions where wrong choice means company death. Fewer such decisions now; screw up reversible ones right away. Focus still important, but can do more with less.
In 2011, tried to get Jeff Bezos on Twitter board—independent seat open, best possible director. Jack Dorsey and Costolo met Bezos at Allen Company conference. Bezos asked about strategy; Jack cited Steve Jobs: most important CEO job is focus, say no to almost everything, be editor-in-chief. Bezos: "Oh yeah, well I like to do everything," big laugh, "My team has to talk me out of stuff. There's lots of ways to be successful." Today Bezos would say can do even more while focused.
Did MBWA at all hours, mostly late night, old HP concept. Started because Bill Campbell said roll a grenade into Twitter at 5:30 p.m. and only take out cleaning people. Can't just tell people to work harder. Started getting dinner, returning 9-9:30 p.m., talking to whoever's there about their work, bumping it up at next all-hands.
Marcel Molina working late on native retweet; Costolo highlighted his work company-wide. People noticed: "How does he always know what Marcel is doing?" Others realized Dick walks around after dinner late night, slowly changed culture—people around when Dick walks get prioritized. Also walked Friday afternoons, learned a lot, asked engineers what's going on.
Helping a tactical director be more strategic: can't just put out forest fires. Job as director is to map territory and do forestry management—what to do so next year fire here doesn't threaten highway. Can only work so many hours stamping out fires as you scale. Push those down; team fights fires that might burn houses. Focus on whole pie, what has to be true 24 months out for territory in better shape. Metaphor for strategic vs. tactical leadership.
Silicon Valley passive-aggressive; Costolo aggressive due to impatience. People come in sad: "Stop. Seems like you don't like working here. You should quit." Direct and blunt, not a jerk. Manage performance and poor performers out without being a jerk—be direct, forthright, empathetic, no stupid stuff like "no one's sadder than me."
For young CEOs never given feedback: Bill Campbell advice—write down what you want to say night before, say that, nothing else. In bad review, write honest feedback; 99/100 don't say it next day, add fluff like "it's hard time for everybody." Person knows hit coming. Be forthright: "I understand if you can't deal with it anymore, I get it." Sit with the silence—unnatural but sit until they run out of gas.
In management class six-hour course, role-play feedback like: worked six weeks till midnight, project cancelled as no longer priority. Only Elad Gil great at sitting in silence after giving feedback.
Promote from within as much as possible so insiders feel no cap. Kevin Weil promoted from IC in 2009 (40-person company) to running all product after ads product. Noto great at Goldman Sachs tech/media, dealing with investors; brought in to handle Twitter drama (Twitter bird hanged/exploded on magazine covers). Goal: no dead birds on covers next year.
Adam Bain non-consensus hire from Fox Interactive LA to run revenue in SF. Board skeptical, but tireless motor for dogfight vs. Facebook social ad dollars. Met during FeedBurner sale process (News Corp interested before Google buy). Flew to LA five-six months to convince.
"Hire slow, fire fast" adage—said out loud but fired slow, often knew at six weeks. Horrible to fire. Board member David Rosenblatt (ex-DoubleClick CEO) shared firing comms plan for global op; only remembered "6 a.m. wake up in cold sweat."
Jeff Weiner (LinkedIn CEO) analogy: don't be manager having first mound visit with pitcher who cranks homer after; have second conversation. Pedro Martinez/Grady Little example.
Pitfall: raise too much money, take eye off operating leverage/efficiency. Burn skyrockets. Even efficiency-focused stop worrying.
People raise too much (e.g., $100M Series B = new $20M Series B); app companies especially.
Hiring breaks: CEOs say only A-players, but customer success hires twice as many after turning customers due to no check-ins in growth. Realize not only A+; manage out low performers. Inexperienced managers avoid firing, do transfiring. Prevent: below X in reviews (1-4 scale, 4 amazing, 1 go) can't switch teams. If bounce 2-3-2, manager gaming; senior team reviews curves, super hardcore.
Acknowledge competition. Facebook juggernaut, Mark Zuckerberg most competitive; they obsess over losses. Understand what they're doing, out-hustle despite fewer resources.
Adam Bain: at CES customer meetings (e.g., Unilever CMO), ask pre-meeting "How'd Facebook meeting go? What psyched to show?" Learn, then in meeting: spend two hours understanding your business/problems (knowing Facebook pitched products without asking). Smart salesmanship vs. juggernaut's new things/microtargeting—come different angle.
Picked advertising model—good call. ARPU higher vs. alternatives.
Tried to buy Instagram a month and a half before Facebook bought it for a much higher percentage of Twitter. Facebook's offer was famously $1 billion ($700M stock, $300M cash); Instagram had 8-9 employees, Mike Krieger and Kevin Systrom. Recognized Kevin Systrom and Evan Spiegel as two of the top product thinkers in Silicon Valley. Before Mark Zuckerberg's offer, went to dinner with Kevin and Mike, offered well over 10% of Twitter (equivalent to $700M in stock at Twitter's $5B valuation). Would have gone higher and offered to borrow $300M from JP Morgan in cash. If could go back, would tell Kevin: "You can run the company when I leave. Congrats, you'll get the CEO role." Instagram was a game-changer for rich media social media feeds; Twitter had Vine but much smaller, and Instagram later pushed video. That was checkmate.
Content moderation is infinitely hard due to subjectivity; can't have hard and fast rules. Example: ISIS captures three pilots, puts them in cages on a beach, lights them on fire, posts photos/videos to Twitter. Instructed general counsel to suspend terrorist accounts posting that media immediately. An hour later, New York Post posts huge photo of the pilot being lit on fire. Then calls come in: "Why'd you suspend the New York Post?" Response: We didn't, but now question who can post it. Requires judgment calls from people with operating control and situational understanding like the GC; will get some wrong, that's life.
During tenure, moderation team was probably left-of-center but didn't impact abuse decisions. No rules like "not allowed to dead name trans people"—that came post-tenure. Dealt with ridiculous amounts of Russian bots (phishing attacks, hacking accounts for credit card info), groundwork for 2016 election interference.
Got a metric ton of bad press, which was soul-crushing despite showing otherwise. Built resilience from improv comedy background: auditioned for SNL, performed at Adelaide Comedy Festival midnight show at opera house (1,000 drunk people, hour+10 min improvised). Five minutes in, crowd yells "You suck, get off." Had to continue 80+ more minutes.
Daughter texted end of 2014 (post-2013 IPO success, CEO of the Year award): Bad news: Yahoo Finance says you're one of five worst CEOs of 2014. Good news: Nobody reads Yahoo Finance.
Declined Vanity Fair Oscars party invite despite daughter's excitement; didn't want to get hooked on CEO-perks then miss them later. Went to NBA All-Star game (Dick Costolo invite via Adam Silver), not Twitter CEO ones. White House visits (e.g., president on Edward Snowden, platform abuse, terrorists) were job requirements.
Hiring adult external CEOs is rare now (founders not replaced), happens in waves. Something will trigger return: companies making $4B revenue after three years run by 26-year-old founders. Examples: Satya Nadella (rare), MongoDB, Workday (maybe one a year).
Advice to VP wanting to be like Dick Costolo: Do what Anthony Noto (Twitter CFO) did—tell executive recruiters you want CEO roles, get name out there. Never took a risk wasn't super psyched about; SNL failure helped (e.g., comfortable on CNBC live TV after being booed by 1,000 in Australia).
Comedy/improv helps: stand up at all-hands easy after stage experience, even scripted.
SF has lots of passive aggressiveness, but exceptions: Collison brothers (John, Patrick) at Stripe—widely read polymaths with broad perspective on humanity/biology/human condition, forthright communicators (neither passive-aggressive nor hyper-aggressive). Reed Hastings at Netflix: straightforward, "This is just how it is, not trying to be mean."
Those three (John/Patrick Collison, Reed Hastings) extremely self-aware. At dinner, Reed asked table: How much of your success is DNA vs. learning to kick ass? When his turn: 80% DNA (luck vs. skill distinction; encoded traits put you in position, 20% luck for the role).
CEO rulebook changing; Jensen Huang runs NVIDIA with 60 direct reports, no criticism, different playbook—no one copying him. Jeff Bezos to Jack Dorsey: Lots of different ways to be successful. Mistake: CEOs read Walter Isaacson books on Elon Musk/Steve Jobs as archetype (mythology, leaves stuff out). Don't try to copy—nobody fooled, miserable being someone you're not.
Management class anecdote (1990s NBA Finals, Houston Rockets locker room): Reporter to Hakeem Olajuwon on Charles Barkley's Nike ad ("I ain't no role model"): Charles different in public/private, has anxiety. Hakeem: Always the same person, happy to be role model.
Misery trying Steve Jobs playbook (impossible even for him).
Two immensely helpful board members: Peter Chernin (from News Corp, independent; managed tens of thousands, global operating experience)—user-perspective thinker ("If I'm chief user officer..."). Peter Fenton (Benchmark)—helped interview/close execs (third-party perspective on why Twitter would be a monster, convinced Adam Bain, Anthony Noto).
Bring in independents with operating expertise early (not wait for IPO/series B).
Would keep private longer if could redo: Easier running private—no missing revenue by 1%, employees freaking over 8% after-hours stock drop, daily ticker stress. Easier to focus long-term vs. short-term tradeoffs.
Master class on scaling as terrific scaleup CEO. Did management by walking around (7pm halls through dev/support/sales for context, see midnight oil burners). Tough feedback: Write/practice night before in mirror, deliver no-fluff, sit in silence.
Kill committees, use DRI (directly responsible individual)—key for cross-silo priorities. Ben Horowitz quote: "Make sure everybody understands what you understand, not just the what, the why, and what success means for them personally. Priorities without context are useless."
Accuses Silicon Valley of passive aggressiveness (he's aggressive). Most CEOs: Hire fast, fire slow (he didn't, wishes he did).
Keep Sequoia Capital in your library
Save the videos and channels worth coming back to, and find them again in one place.
