How to be a CEO when AI breaks all the old playbooks | Sequoia CEO Coach Brian Halligan
These notes were generated by AI and may contain inaccuracies.
Founder CEOs face crises with no one to rescue them—neither parents nor VCs. Starting a company has never been easier, but scaling into a durable, high-impact organization has never been harder. The number of companies formed will mushroom over the next 10 years compared to the last 10, making it hard to stand out and accelerate.
Today's CEOs face massive tax and optionality from moving fast and trying many things, pressuring them to be faster, better decision makers. Many want to be founders or CEOs, but few can handle it. Founders work 7 days a week, always on—e.g., texting Sunday nights. It's full contact.
Brian Halligan is co-founder and longtime CEO of HubSpot. Now Sequoia's in-house CEO coach, he coaches top CEOs one-on-one, hosts groups of dozens of top CEOs to learn from each other ("kids table" for companies under 100 employees, "adults table" for over 100), and runs the podcast Long Strange Trip interviewing successful CEOs.
Board members describe Brian as in a perpetual state of constructive dissatisfaction. He sees this as a compliment from sub-chair Lori Norvington. Current fast-growth CEOs share this positive dissatisfaction, rarely taking stock of achievements, always focused on the end state. This generation of CEOs is surprisingly humble, unlike Brian's generation.
CEOs graduating from kids table (<100 employees) to adults table (>100) focus intensely on exec team, direct reports, and next-level org design. They spend half their time recruiting and interviewing—surprising many who realize their job is mostly that. CEOs overrate their interviewing ability and gut feelings, underrate blind references.
Dave from MongoDB had 2 C-level turnovers per year over 10 years. HubSpot and similar startups experience high exec turnover.
Coaching on Interviewing and Hiring
Even after decades, Brian overrates his interviewing ability. Advice:
- Parker Conrad's hack: Before C-level interview (e.g., CFO, CPO), have candidate sign NDA, send recent board deck/memo, chat 30 minutes about it. Complimentary responses are red flags—want challengers, not yes-people.
- Work through problems on whiteboard.
- Skip standard background walkthroughs.
- Do high-quality blind references: Ask "Would you enthusiastically rehire them for that role?" or "On a scale of 1-10, how likely to rehire them back from me?" or "Top 1% of employees? Top 10?" VCs who ask hard questions show seriousness.
- Hire slow, fire fast (opposite of common practice). 50%+ of C-level execs gone within 18 months.
At HubSpot, for head of engineering candidates:
- Candidate A: 4/4 from four interviewers, 2/4 from four others.
- Candidate B: All 3/4 across eight. They hired 3/4s initially (fewest weaknesses), switched to spikier candidates with weaknesses/challenges—improved hit rate. Shrunk interview panel from 8 to 4 (recent head of product hire).
To get honest references: Detect if caller has decided (e.g., superficial strengths/weaknesses questions). Hard questions prompt honesty.
Avoid big-company hires at small scale (50-500 employees)—impedance mismatch on expectations. 100% attrition hiring from Salesforce, Google, Microsoft. Avoid McKinsey hires—they're conservative, clash with founders skeptical of conventional wisdom.
Phases of failed hires: McKinsey cohort, Apple group, Amazon group—none work consistently.
Boston Red Sox won World Series in 2004 (first in 86 years since trading Babe Ruth) with mix of homegrown high-quality inexpensive talent from farm system + key free agents like David Ortiz, Pedro Martinez, Curt Schilling (older, big-company types). Culture worked perfectly. Underrate homegrown talent—aim for this mix, not all veterans or none.
Brian is part owner of Red Sox. At HubSpot, half the management team are long-timers ("150 years"). Same at Apple. Tip: If close between homegrown VP and external, give homegrown shot—externals shine in interviews, hide warts (unless great blind refs). Brian Chesky over-rotates to experienced hires/delegation, but most CEOs still hire externally.
Not anyone can be a successful CEO—requires work, but specific profiles. Brian's LOCK algorithm:
- L: Lovable—inspires followership (e.g., would 28-year-old Brian crawl across broken glass to work for them?). Steve Jobs inspired despite not being lovable.
- O: Obsession—deeply obsessed with problem, founder-market fit, evidence of past rabbit holes. Negative on 6-month-old ideas.
- C: Chip on shoulder (boulder)—nearly all have it, including Brian.
- K: Deeply knowledgeable about domain.
Add S: Student—deep students of the game (e.g., Winston Weinberg of Harvey, James of Profound, Gabe of Rogo). Study peers, history, like LLMs constantly learning.
Common in successful founders (Lenny's research with Terren Rohan): Extreme ambition for wild ideas (e.g., Spotify's global music subscription—"that's crazy").
Learnable CEO Skills
Young CEOs (e.g., Winston, late 20s, no management experience) must learn:
- Constant feedback (positive/negative)—unnatural, but critical.
- BS detection—org always sells to them.
- Inspiration—many never inspired before (e.g., lawyers starting companies). Best learn fast on scale-up path.
Most common to work on: Feedback, especially tricky transitions like:
- Co-founder steps aside from product/eng ops to CTO/thinker/labs, hire operator.
- Layer above early head of sales who built team but can't define process/forecast. Unnatural for 25-year-olds.
Rare five-tool CEOs (baseball analogy: hit, power, run, field, throw—all top-rated):
- Code, taste, vision, sell product, convince employees/investors. Examples: Brett Taylor. New breed post-AI; past icons like Steve Jobs, Jeff Bezos lacked some (e.g., Jobs didn't code).
AI won't easily fill gaps (e.g., genius dev lacking sales/inspiration for investors/employees/Fortune 500).逆に, mere mortals like Brian can now build via AI.
Best CEOs get good at uncomfortable skills by studying them and sucking it up. Misery loves company: Brian runs peer groups like the "kids table" (15 CEOs of companies under 100 employees) and "adults table" (15 CEOs of companies over 100 employees). They discuss challenges in a safe space where peers weigh in more effectively than Brian. Problems at kids table differ from adults table but rhyme. Find peers to talk, share, and be vulnerable.
Brian teaches MIT course on scaling startups, not starting them. Quote from syllabus: "Starting a company has never been easier, scaling one into a durable high impact organization has never been harder." Number of companies formed will mushroom over next 10 years vs. last 10 (which mushroomed vs. previous 10). Example: Kid walks into CVS for toothbrush (4-5 options); 90s/2000s Amazon has 4-5,000 toothbrushes, so 4-5,000 companies. AWS made starting software companies easier (huge jump in 2006). Now even bigger jump with AI. More noise/competition makes standing out and scaling harder. Distribution key to breaking through noise; founders didn't grow up doing it, so fastest learners win.
Tweet: Sales is maybe the last job AI replaces. Inside enterprises like HubSpot, AI excels in software development, customer support, legal starting well. Go-to-market slow: support works, BDR first, but enterprise sales (trust between carbon-based life forms) very late to change.
Traditional funnel: Get found in Google, click blue link, land on website, rabbit hole, contact sales, wait for rep.
Future: Buyers start in Gemini/Anthropic/ChatGPT (knows your site, competitors). Websites less important; homepages get high-quality avatar knowing products, company, pricing/packaging. Conversation stored in CRM, scored, sales rep follows up with avatar + all-knowing SE on calls.
Buyer has agent; seller has avatar. Every knowledge worker wants deli clone (Brian's in footer, link del.i). Connect to email, Google, Notion; invite to meetings as participant (not just notes like Granola). Every website changes to all-knowing homepage avatar; considered purchases hand off to sales rep with SE avatar on Zoom.
Hasn't unlocked yet but will change a lot. Future: Bots chatting, humans podcasting on vacation (send avatars to meetings, e.g., buy HubSpot seats). Like Cloudbot/Moldbot: personal agent doing stuff.
AI-driven companies do same as always: Call CS/SEs system consultants or forward deployed engineers. Hiring same folks, same enterprise sales processes. Enterprise sales unchanged since 1990s (Brian's 10 years at PTC). Forward deployed engineers help implement, integrate, customize—not plug-and-play. Similar to solutions consultant/sales engineer but retrained; term rebrand.
Advice: Lean into employees helping customers onboard/succeed. Top of funnel changes first: AEO (get found in AI vs. Google), optimize websites differently, homepage = avatar conversation vs. pages.
HubSpot built one; working. No one else yet.
Being CEO different: Winston from Harvey: "You can just do a lot more." AI agents boost productivity; devs do year-of-work in 2 months. More projects possible but dangerous—don't hop to second act too fast, lose focus on beachhead (e.g., OpenAI/ChatGPT consumer app core, refocused vs. Gemini). Competition fast; planning cycles now 3 months vs. year. CEOs must be faster/better decision-makers.
At HubSpot, slowdowns/churn from Brian's fault: Delayed one-way door decisions. Making them freed company to move faster. Valued optionality before; now massive tax on it when moving fast. Tech shifts weekly make long planning waste.
Eat a Sandwich, Don't Nibble
Stole from Ruth Porat (Google CFO). Rip band-aid off: Layoffs, bad news—do fully, tell adults. Upcoming retrenchment: Public markets tight (anti-bubble), private bubbly; reckoning, many layoffs. Avoid nibble/grow into it.
Next Play (Mike Krzyzewski)
Duke coach yells "Let's play!" after miss—don't compound error with aggressive defense/foul; forget, run next play. HubSpot used Krzyzewski slide in meetings after unforced errors. Example: March 31, 2019 all-day outage, customers canceled/yelled, Brian cried in company meeting, used "next play" slide. Most mistakes self-inflicted; companies die of overeating vs. indigestion (or suicide vs. homicide).
Never Waste a Good Crisis
Most good HubSpot changes from crises (overcorrect). 2019 outage: Rethought software deployment/development; no serious outage since. Pivoted from marketing software to CRM (Salesforce threat); CRM downtime (esp. quarter-end) kills customer business—mindset shift, major changes. Purposely swing pendulum hard.
If You Want to Kill a Plant, Have Two People Water It
Plant dies: Over/under-watered. CEOs religious about DRIs (directly responsible individual) at scale. Startup mode: Everyone aligned on pilots. Scale: Silos (sales, forward deployed engineers, PMs, devs)—need powerful DRI owning cross-functionally (e.g., salesperson directs others). Committees fail; one ass on the line motivates.
No Silver Bullet—Lead Bullets (One Step Forward, Two Back)
No single hire/investor/event/product = silver bullet. Inside HubSpot (looks smooth up/right): Two steps forward, one back (often crisis). Founders on own—no rescue. Brian talks many out of starting: Painful, sacrifices, obsession (7 days/week, 996+). Jensen Huang: Wouldn't start Nvidia over. Brian would restart HubSpot—proud, good deathbed life (Dalai Lama). Massive platform change now; don't waste opportunity.
People today are much more hardcore than in previous eras. Brian worked 60 to 70 hours a week throughout his time at HubSpot, never really turning it off. Now people are much more focused, inspired by Elon Musk. With his earlier startup, which was nowhere near as successful as HubSpot, Brian's mindset was to give it everything: "This is my chance. Forget balance. Go for it seven days a week for a while, then scale back." It's empowering to try fully, knowing it won't be forever.
Balance for CEOs is not something you should have if you want to be incredibly successful.
None of the founders Brian works with have work-life balance. He didn't have it, his co-founder Dharmesh didn't. The only CEO outlier is Kareem from Clay, who insists on balance and taking weekends. Everyone else is obsessed and doesn't have much of a life. Clay took a long time to find product-market fit, pre-AI, but it worked out.
EV is enterprise value, TV is team's value, Me is your value. As HubSpot scaled, VPs with sizable organizations failed when they solved for TV over EV—optimizing for their own team, like sales pushing bookings regardless of downstream service issues, or marketing-sales friction. Immature managers suboptimized for themselves, peers, or employees.
This showed up in quarterly employee net promoter score (eNPS) surveys, rated by department alongside customer NPS. Example: Sales eNPS drops from 65-68 to 30, comments criticizing the leader for TV focus. Feedback was given, but scores often dropped further (e.g., to -5). Once you lose your team, it's hard to recover. Hence, hire slow, fire fast. This issue emerges after 100-150 employees, when CEOs lose direct touch.
HubSpot put on the wall: Solve for EV > TV > Me, later adding CV (customer value) first: CV > EV > TV > Me.
You don't work for your boss, you work for the company. (Steve Jobs)
You work for HubSpot first, then your boss.
Tactics to enforce:
- Explicitly in performance reviews: score 1-10 on solving for EV.
- Public feedback in large meetings (like Jensen at Nvidia, no 1:1s initially).
- Quarterly company meetings ending with Champagne Awards: signed Veuve Clicquot bottles to EV solvers.
- Posters emphasizing company growth/customers above all.
HubSpot started very employee-centric: #1 on Glassdoor best places to work, Brian #1 CEO. But wanting to be liked isn't ideal for CEOs. Toby from Shopify has lower scores but excels. Overindexing on employees diverted from customers.
Co-founder Dharmesh and HR head Katie Burke drove this. Management meetings spent half the time on employee NPS (e.g., 60) vs. customer NPS (25). Brian pushed to trade 10 employee NPS points for 10 customer points, shifting center of gravity to customers.
Shifts implemented:
- Monthly management meetings start with customer panel Brian runs, extracting bad news.
- Customer panels at board meetings: "What do you love/hate about HubSpot?"
- Customers featured in company meetings.
- Management comp tied to retention and NPS, not revenue.
Companies have one center of gravity: customer, employee, or investor-centric. Changing requires overcorrecting, repetition (say the same thing over and over), and making it obvious as the company grows.
As CEO, be careful: offhand hallway comments (e.g., "It'd be cool if...") get overinterpreted as directives. Dharmesh created flash tags system for emails/Slack:
- "How do you want me to get this done this week?"
- "We should talk about this."
- "FYI, just thinking about it."
- "Pleading you to do this."
Brian and Dharmesh were co-founders for 20 years. Dharmesh never wanted direct reports, even from day one. He was a prior CEO and refused, forcing Brian (an engineer who can code but not exceptionally) to handle engineering. Dharmesh has never had a single direct report at HubSpot, freeing him to tinker, think, and be creative.
As companies scale, CEO job shifts: startup phase is 90% perspiration, 10% inspiration; scale-up is 90% inspiration, 10% perspiration. Must let go to scale, stay attached to customers. Brian had trust issues, small trust surface—a common scaling limit for CEOs.
Brian's Halliganisms post summarizes 15 years of CEO mistakes to help others avoid them. His podcast Long Strange Trip interviews CEOs (listen to Lenny's first).
Grateful Dead as Silicon Valley startup: started 1964 in Palo Alto, early concerts at Stanford/Silicon Valley. First-principles, created new category, disintermediated ticketing. Steve Jobs and Jerry Garcia similar craft people.
Brian's book: Marketing Lessons from the Grateful Dead. He stewards Jerry Garcia's guitar, played by Dead & Co and cover bands.
Lesson: Spiky teams create categories. Grateful Dead: Garcia (bluegrass banjo), Bob Weir (country), Phil Lesh (avant-garde jazz trombonist), Pigpen (harmonica), marching band drummer. Created jam band genre.
Books: Hasn't read a book in a long time—prefers podcasts, X, talking to CEOs. Barbell strategy like Mark Andreesen: Twitter or books 10+ years old.
Recent movie/TV: Ken Burns' Revolutionary War documentary (10+ hours). America as disruptor startup: gutsy, two steps forward/10 back, near-losses, key alliances (French), leadership details on George Washington.
Favorite product: Deli clone for teaching Scaling Entrepreneurial Ventures course (no office hours, uses AI). Voice feature coolest; trained on podcasts, newsletters, lectures, proprietary docs. lennybot.com.
Sonos: Great but annoying (both agree); nothing better.
Life motto: Snowmobile accident 4 years ago—drove off cliff in Vermont, unconscious, thought he'd freeze to death, called 911. 20 broken bones, 33 screws, year out. Realized he didn't like being CEO of 8,000-person company; passed to Yam (still CEO). Life's short, don't waste it. People think they'll live forever; at 58, be intentional about decisions, people, joy.
Red Sox: Part owner. Not as profitable as thought—league economics flawed, no salary cap (Dodgers $400M payroll vs. Marlins $100M). Broken model, may correct soon.
Stay in AI SaaS if you want to make money, particularly vertical SaaS.
Brian, this was incredible, covered almost everything hoped to cover. Two final questions: Where can folks find you online if they want to reach out? Where do they find the bot? How do they work with you if they wanted to, or do they have to be a Sequoia founder? And how can listeners be useful to you?
Two things: Listen to Long Strange Trip, my podcast. Get comments, but not a ton—Lenny, you have more comments than me, I'm jealous. I'd like feedback on how I'm doing. It's very new, started a couple months ago, going pretty well, but currently just family and Sequoia people giving feedback. Would love to hear what all of you think.
Hop on your YouTube and leave some comments about what they think. The real honest feedback.
Brian, thank you so much for doing this. Appreciate you. Thank you. Bye everyone.
Thank you so much for listening. If you found this valuable, subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Please consider giving a rating or leaving a review as that really helps other listeners find the podcast. Find all past episodes or learn more about the show at lennypodcast.com. See you in the next episode.
Keep Lenny's Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
