Twilio on Shipping Fast

How Jeff Leads Twilio 

Twilio is 17 years old, has scaled to over 5,000 employees, and serves hundreds of thousands of businesses worldwide. The company generates over $4B in revenue and operates across four core communication pillars: (1) a CPaaS platform supporting SMS, voice, and messaging APIs, (2) customer engagement tools like Twilio Segment for data-driven marketing and personalized outreach, (3) contact center solutions through Twilio Flex, enabling customizable cloud-based customer support, and (4) developer infrastructure supporting scalable and secure global communications.

1. Operating Cadence

The Alignment You Create Becomes Your Team’s Operating System – Team Structures Are Less Important 

• There’s no universally “right” way to structure teams — functional, structural, or hybrid models can all work. What matters more than structure is how clearly your team understands what matters and moves toward it together.
• For companies with 100 or less employees, teams are usually functionally arranged under a domain leader. Around the 50 person mark is when decision making slows down as these decisions are made functionally instead of as a business. 
  
  • After 50 employees, it’s hard to keep everyone wanting and prioritizing the same things. 
• A company’s structure reinforces its operating system – creating clarity and alignment around what you value and reward is the most important thing.
  ‍

Leadership is Creating Clarity 

• At 10-20 people, it’s easy to keep everyone on the same page – everyone’s in the same rooms and organically has access to the same information. 
• At the growth stage, teams become siloed, and it becomes the CEOs job to create clarity by often and loudly communicating what’s important. 
• Outside the company, the CEO’s job can include raising money and external marketing but inside the company, the sole job is to create clarity so employees can cut through any noise. 
• To create this clarity, Jeff initially experimented with his own system but ultimately found that there was value to everyone already being familiar with OKRs. 
  
  • Twilio held off on introducing a formal system until the company reached ~1000 people. In hindsight, Jeff wishes they did this much earlier (<100 people) – once they did, they went from not getting stuff done to delivering as a company.

Tactic: How Jeff Uses OKRs to Create Clarity 

• OKRs are the operating system to move the company forward
• Jeff starts with the big picture – this is one or two sentences that really describe the why. 
  
  • Why are we all here? What are we trying to accomplish?
  • The big picture should be correlated to your mission, but captured as a three year horizon
  • The point is for your employees to understand what the point of all their work is over a horizon of several years. 
  • You should be repeating the big picture a lot – to the point where any employee should be able to cite it if asked. 
    
    • An example of a big picture Jeff used at Twilio was “Evolve from a communications company to a CRM company”
• The objectives are the how. These are the major pillars of what you’re going to do. 
  
  • There should be three or four major objectives fulfilling how the big picture gets done
    
    • People should be able to remember all the objectives
    • After three or four, it becomes difficult for employees to remember
  • These objectives should generally have an 18 month lifespan 
    
    • This also gives the CEO time to communicate them in the company
  • Jeff orders these objectives to communicate the order of each one’s importance
• The key metrics are the guideposts along the way to tell you if you’re making progress
  
  • These key metrics should be measurable and quantifiable
    
    • A key issue people make is starting with the measures instead of the goal 
  • Jeff would order the measures to create clarity and signal to the company what is more important when resources are contested  
  • Jeff always used the wording “key measures” instead of “key results”
    
    • People will not remember all the measures – that's okay. They only need to remember the ones that are important to their job function.
• Example: 
  
  • Big picture: Build an enterprise software company
  • Objective: Deepen our relationships with enterprises 
  • Measure: Increased engagement by 14%

Example: Teams, Particularly Sales, Can Optimize for the Wrong Metrics – Assign Incentives Carefully  

• Twilio wanted to diversify their business away from messaging – messaging was their core offering, but the aspiration was always to layer in more products on the side. 
• The problem was, their sales compensation model never properly evolved far enough to encourage salespeople to sell products other than messaging. 
• Even when they tried tweaking the compensation model, selling messaging products was still easier, so employees continued to prioritize it. 
  
  • This is quite common in sales and other well-operated functions – when people optimize for the wrong thing, they can very efficiently get that thing, even if it’s wrong for the company
  • This often happens in the absence of employees knowing why they’re doing something, because there’s no alignment among the company. The goal of the OKR system is to align what everyone at the company is doing. 
    
    • Otherwise everyone thinks working on their siloed function equals success 

Example: The Toilet Paper Problem

• Jeff believes that the OKR objectives don’t need to encapsulate every aspect of what a company does – the focus should only be on what the CEO wants to instill or change.
  
  • Some tasks do not need to be explicitly listed to be done, like putting toilet paper in the bathrooms. 
    
    • If everything is listed as a goal, no one will remember the goals
• Each function should monitor their relevant “toilet paper” tasks, and only highlight them if anything abnormal comes up. The way you can encourage this is by highlighting these “toilet paper” metrics when they expand outside the bounds of what is expected. This way teams know you are looking. 

OKRs Should be Communicated at least Quarterly

• Jeff believes quarterly all-hands meetings are the best time to communicate and refresh OKRs
  
  • Annually slowed down the cadence of the company while weekly only created more confusion
  • He pushed Twilio to adopt a faster operating cadence, ideally syncing the company on two-week sprints. They eventually landed on monthly rhythms instead.
• The goal with communicating OKRs should be to create a storyline that makes sense over the course of people’s employment. 
• At Twilio, Jeff would make sure to communicate OKRs visually.
  
  • During the all-hands, he reviews last quarter’s metrics and assesses each one as red, yellow, or green
• Your big picture and objectives likely won’t be changing quarter-to-quarter – that’s a good thing.
  
  • Jeff explicitly communicates that the company continues to be focused on the same big picture and objectives 
• Before setting up the new measures, always go back to the big picture. 
  
  • At Twilio’s all-hands, Jeff used a slide that has only the big picture written on it to build alignment and really drive the point home. 
• List the same ordered list of objectives on the next slide. 
  
  • If there’s ever a change, call it out – remind people what you were aiming for last quarter, and explicitly say that instead the company’s focusing on a new objective this quarter.  
• Clearly communicate how the goalposts are moving by listing the new measures for the quarter.
  
  • Jeff would dim out everything on the slide other than the current objective being discussed. Then, he’d list the new measures out one-by-one. 
    
    • Every measure isn’t relevant to every employee of the company – sometimes measures wouldn’t even be discussed at the all-hands, and instead function leaders would receive an email with the measures relevant to their roles. 
      ‍

Goal Setting Should be Directional 

• Jeff emphasized that companies—especially early-stage ones—need clarity in what they’re trying to achieve, but should avoid rigid planning
• He suggests matching your planning precision with how mature your company is:
  
  • Early stage: directional ambition (e.g. grow 100% YoY)
  • Growth stage: introduce a model to make the business profitable
  • Public stage: forecast with predictability and precision
    ‍

Your OKRs Don’t Need to Go Down to the Individual 

• Twilio never got OKRs to trickle down to the individual level. 
• As CEO, the most important thing is to communicate the company’s goals really well, and then hold the company accountable for achieving it. 
  
  • This usually means company leaders should set similar OKRs with their teams, but Jeff believes that enforcement becomes difficult and performative down to individual contributors (although some companies like Salesforce and Google do require individual OKRs). 
  • Example: for companies with 50 people, CEOs should expect that each function of ~10 people should set OKRs with their leader. 
    ‍

Renaming Well-Known Conventions Helps Employees Unlearn 

• At Twilio, Jeff renamed OKRs to BPMs (Big Picture Metrics) to prevent incoming employees from copy-pasting the OKR processes used at other companies. He believed other companies did OKRs poorly, and wanted to retrain employees to use Twilio’s OKRs. 
• Generally, Jeff advises either adopting standard processes, or erasing employees’ slates:
  
  • Adopting the standard is easier and more familiar to everyone
  • If you feel strongly about innovating on a process – call it something different. This will erase the slate for employees and push them to familiarize themselves with the new process. 
• These innovations don’t always work – eventually, Twilio switched from BPMs to OKRs because their internal system didn’t work with any off-the-shelf tracking software. 
  
  • Even after switching, Jeff kept the big picture as the overall arc guiding Twilio’s OKRs
    ‍

Without Prioritizing, You’re Not Creating Clarity

• Early on, the company’s annual planning document was a six-pager the team would dump everything into. 
  
  • Exec teams are usually full of smart people with different ideas, and conflict happens when trying to arrive at the best idea. Instead, Jeff noticed that while compiling the six-pager Twilio’s executive team never fought. 
• As Twilio began to grow, this process broke because they weren’t making prioritization decisions
  
  • When Twilio brought in a new exec, he pointed out that after reading the six-pager he still didn’t know which ideas were a priority. He also pointed out that Twilio may not have been accomplishing any of these ideas because none had been prioritized. 
    
    • Jeff found that saying yes to everything wasn’t leading – Twilio was just documenting a lot of ideas without making decisions. Once you’re making decisions, that’s when conflict arises, but that’s also when you’re creating clarity and leading the company. 
• As Jeff forced decisions and explicit prioritization, tension came up. But this helped them prioritize what they needed to do – when teams debate what’s important, it forces clarity:
  
  • Either their work isn’t genuinely important, or
  • It is important, and your objective/metric system isn’t capturing it
    
    • Resource-constrained environments will make these tensions worse – when layoffs are looming, employees may conflate “this isn’t a priority” with “I’m at risk” causing anxiety and unproductive behavior. 
• Avoid tying compensation to performance metrics. Incentives make everything more rigid and political. 
  
  • If you’re not excellent at goal-setting, tying comp to goals will backfire. Employees may chase the wrong outcomes, just because of the incentive structure. 

Example: How Jeff Aligned the Onion’s Leadership Team

• After acquiring the Onion, Jeff saw the leadership team scatter in different directions
  
  • Everyone was chasing disparate, external ideas and taking unrelated meetings 
  • The team had no shared perspective on what was important – during weekly syncs, each person figured the most important priority was the last call they got off of
• Jeff led an offsite with the 10-person leadership team where they aligned on a big-picture vision and defined 3–4 clear objectives to anchor execution
  
  • Every item was assigned an owner—if was on the list, someone in the room was accountable
  • The team initially struggled, constantly trying to sneak in extra ideas, before they eventually stuck with the fact that if it wasn’t on the list, they weren’t going to do it
• Their final plan was:
  
  • Big Picture: Create an Onion we’re proud of—putting our audience first and monetizing primarily through memberships, starting with print.
  • Objectives:
    
    • 1) Put our audience at the center of the onion, even at the expense of revenue when needed
    • 2) Build a membership program that gives our most engaged users a physical paper
    • 3) Grow and engage our audience on Tiktok, Instagram reels and YouTube
    • 4) Finish making the onion our independent company
• The plan was simple and posted on every employee’s desk. Everyone in the org could recite the plan—and understand what they were doing and why. 

Tactic: How Jeff Introduced OKRs at Twilio

• Jeff started by getting buy-in from his executive team. While it’s the CEO’s right to implement a system, getting buy-in is still necessary. 
  
  • Start small: approach your executive team with 4 objectives that are the most important for the company and a big picture for why.  
  • The focus should be on getting buy-in that a system is needed, not mandating a system from above.
• Jeff framed the change as a response to Twilio scaling. He emphasized:
  
  • His draft wasn’t final
  • He wanted feedback
  • His team’s reactions would shape what came next

Objectives Should Cut Across Functions 

• Your functional background will shape how you see priorities. Product-oriented CEOs tend to define objectives around product outcomes, while GTM-oriented CEOs will lean towards hiring salespeople or increasing distribution. Some of this is valid, some is bias to overcome. 
• Companies have a tendency of having objectives mirror their org chart (a sales objective, a product objective, etc.). Instead, Jeff believes the strongest objectives aren’t functional – they’re cross-functional outcomes. 
  
  • Having cross-functional objectives improves storytelling, alignment, and provides clarity 
  • Example objective: Penetrate the enterprise market
    
    • This requires product, sales, marketing, hiring – every team sees their role in achieving it

2. Value Systems

Every Company Has a Reward Function 

• Every company has an unspoken, undocumented reward function. 
  
  • At Amazon, this was invention – they would award someone for the most novel invention of the year and their conference is named Re:Invent
  • At Apple, the reward function is design – you’re more valued for designing a beautiful interface or a typography. 
• At Twilio, Jeff was trying to build a culture of innovation, while the finance team kept killing early-stage ideas that weren’t producing enough ROI
  
  • This revealed a lack of alignment at the company-level on what deserved to be rewarded 
  • Providing clarity on what gets rewarded lets employees know what is and isn’t worth extra time and effort 
• Your company values aren’t what’s written on the wall – it’s shown through who you hire and fire, who you promote and reward, and who you do and don’t publicly praise. These signals shape company culture.
• As Twilio scaled past 800-1000 employees, teams received mixed messages around innovation 
  
  • The IoT team was encouraged to experiment and take risks, but the finance team grilled them for burning $10M to produce $30M. This is because teams lost sight of the reward function at Twilio.
• With no shared understanding of what’s rewarded and what success looks like within that company’s reward function, teams won’t know what to prioritize and strive for. Figure out your team's reward function (i.e., values) and reward these behaviors.

Candidates Should Align With Your Company’s Reward Function

• When interviewing potential candidates, screen for whether they align with your company’s reward function. 
  
  • The goal should be to get in the candidate’s head and see if when rubber meets the road, they use your company’s values and reward function to find a solution. 
• At Twilio, Jeff would ask candidates what they hated about their current job and what ideas they had to fix it. Candidates with innovative solutions aligned well with Twilio’s culture.

Being Too Prescriptive Can Stifle Good Ideas

• OKRs are top-down by design – it’s how leadership communicates clarity, focus, and alignment. For a long time though, Twilio tried to run with a bottoms-up approach where employees on the front lines were empowered to go do what they thought they should be doing. 
  
  • That worked for a while, but eventually things got too chaotic
• To remedy this, Jeff implemented a negotiated OKR approach:
  
  • Jeff would communicate company-level top-down OKRs 
  • Teams were then encouraged to write their own OKRs from scratch instead of just inheriting their manager’s 
  • Teams and managers would then negotiate down to alignment 
• In practice, this was a tough ask – teams didn’t want to challenge their managers. Instead, a more productive framework was to set a clear outcome goal at the top (e.g. Increase conversion rate by 30%) and let teams innovate on how to achieve them. 
  
  • This provided a structure for innovation without dictating the methods 

“Agility is having the same destination—but changing how you get there.”

Innovation Means Trying a Lot and Cleaning Up After Yourself 

• Twilio’s focus on innovation made it a more chaotic company than most. Even though employees wanted innovation, they were frustrated by the clutter from loose ends and messy experiments scattered around the company. 
• Instead of rejecting innovation, Jeff identified the underlying problem – incomplete follow-through. Trying and failing was fine; leaving half-built features for others to clean up was not. 
  
  • If something didn’t work, you had to shut it down cleanly. This way Twilio could make space for experimentation without letting failed ideas become a permanent tax on the company. 
    ‍

One Company, One Set of Values

• Each team may have different objectives, but the company’s values must be universal. If the core value is innovation, everyone—from HR to finance—must operate in a way that supports it.
• Without that alignment, product might ship new ideas that finance shuts down and sales ignores – blocking any innovation within the company
  
  • Values should be reinforced cross-functionally – if you say innovation is a value, then finance should help evaluate smart risk, not just cut losses, sales should showcase new ideas with pride, and HR should recruit and promote people who take smart risks. 

Example: How Amazon Values Invention, Even Within HR

• Jeff’s board member Rick Dalzell (former CIO of Amazon) shared how Amazon’s HR team has 10,000 software engineers
  
  • That’s one of the world’s largest engineering organizations, and it sits within HR
• Amazon employs millions of people, many for just days or weeks (like drivers and fulfillment workers). No off-the-shelf software could handle their needs at that scale or complexity, so they built their own, treating HR as a strategic, productized function.
• Every function should align on the company’s overall values – if the company values invention, these teams should invent too.

3. Organizational Design

Keep Teams Small Enough to Feed With Two Pizzas 

• At Amazon, teams were no larger than what you could feed with two pizzas (typically 8-10 people)
• As teams grew, they were split in half to create smaller, focused units with less coordination overhead 
  
  • These teams require fewer meetings, make faster decisions, and are easier to align 
• Small teams also push the company to formalize interfaces between teams—through APIs, documentation, or processes—rather than relying on messy coordination. 
• These splits, however, are time-consuming to set up and have their own tradeoffs:
  
  • Splitting a team often requires refactoring code, designing new APIs, and clarifying ownership boundaries.This refactoring can take months, but amounts to big gains in productivity over time. 
• For these splits to work, organizational buy-in is required 
  
  • At Amazon, everyone was deeply committed to the idea of small teams moving quickly. At Twilio, splits often failed when teams didn’t fully adopt the supporting practices or lacked a management team to coordinate between teams. 
• The two pizza model also creates leadership opportunities. Every time a team splits, there’s room for internal leadership to scale with it. 
  ‍

The Ideal Team Structure is Small and Aligned

• The ideal team structure is when a single team owns a product area, a customer problem, a mission, and a codebase
• This alignment means teams know who their customer is, own the full stack they’re building on, and can ship without heavy cross-team coordination. 
• But it’s an ongoing process – as teams and systems grow, codebases get messy, ownership lines blur, and tooling breaks 
  
  • Every few months, you need to refactor infrastructure and team charters to keep scaling – this is just a natural tax on growth 
  • Reclarifying who owns what and how things get done is key
    ‍

Middle Management Matters – Even If It Slows You Down

• Google tried eliminating middle management entirely, leaving VPs with up to 75 direct reports. It didn’t last – the lack of structure created chaos and made it impossible to scale leadership.
• Middle management often gets blamed for slowing things down—and in many cases, that’s true. But that inertia also creates stability – functions can keep running even if people leave, mistakes are made, or priorities shift.
  
  • 40% of Twilio’s engineering team turned over in one year during the pandemic. That moment made Jeff appreciate the value of strong middle layers that quietly keep systems running. 
• As companies grow, not every aspect of the business should be in function – some areas should just work and be “battened down”. This allows senior leadership to focus on newer areas. 
• Jeff Bezos is a great example of how delegating stability allows for investing in innovation:
  
  • Once e-commerce stabilized, he focused his time on birthing AWS. 
  • Later he repeated the pattern for Alexa, and less successfully, the Fire Phone 
    ‍

Not All Complaints Are Real Problems 

• As CEO, you’ll constantly hear concerns and requests—some are noise while others are early warnings
• Jeff recommends speaking with enough people to spot trends. Recurring themes across functions are usually real problems. 
  ‍

4. Go-To-Market

Avoid Needing Multiple GTM Motions If You Can

• Jeff initially ignored the advice to focus on a core product offering. Twilio launched products constantly with the goal to become the “everything platform” for communication. Instead, many products ended up subscale and under distributed. 
• A recurring problem was the fact that many of their products lacked a GTM plan. They’d build the product, announce it on stage, publish a landing page, with no real strategy for GTM and teams ill-equipped to support the product's launch. 
  
  • Example: Twilio split their voice product to launch programmable SIP Trunking, but they didn’t account for the fact that IT buyers are inherently different from developers, and their sales team would have to sell the product differently. 
    
    • The product struggled until sales reps figured out how to sell it manually 
  • Without a GTM strategy, Twilio wasted R&D dollars, created internal misalignment, and confused both the market and their own teams
• When you want to build a product, it has to have a GTM attached to it. Else, you shouldn’t be building it. 
• For every product you build, you have to ask yourself if it’s a product with an independent GTM motion, or is it a feature attached to an existing GTM motion
  
  • There’s power in having things that attach to an existing motion – it makes sales, retention, and renewals easier  
  • Do not underestimate how expensive, slow, confusing and failure-prone spinning up a new GTM is for the company
    ‍

One Department Does Not Mean One Sales Team

• You should treat every new buyer as its own motion unless proven otherwise
  
  • A common trap for multi-product companies is thinking, “We sell to finance—our new product will too.” But “finance” isn’t one buyer. It’s made up of FP&A, Tax, Procurement, Payroll, etc. 
• When AWS launched Amazon Connect, they ran it with its own dedicated sales team because the buyer was fundamentally different than the typical AWS buyer
  
  • This separate sales team was also then incentivized to bring leads in, which boded well for Amazon Connect
• At one point, Twilio had 4 different sales teams – for a company of that size, this was too complex. Sales teams were duplicating efforts and buyers were confused who to speak to.
    

Align Sales Teams With Customer Opportunities, Not Org Structure

• Sales teams are economic heat-seekers. Their job is to maximize revenue, and they’re naturally drawn to what sells best. 
• Twilio made the mistake of over-segmenting their sales org – customers were confused why reps could only help with specific Twilio products, and reps wasted time and opportunity being limited by what they could sell. 
  
  • Sales teams should be aligned with the biggest customer opportunities, not limited by rigid org boundaries
• Only create a new sales team when it’s absolutely necessary. This makes sense when the product is strategically critical, and won’t succeed unless reps are solely focused on it. 
  
  • A new sales team is helpful for when a product is serving a completely different buyer or a new product needs to be forced into existence
     

Ship Instantly Usable Features to Accelerate Adoption

• Jeff believes distribution is the most important bottleneck for startups. 
• The best way to aid distribution is to add features in your product that developers don’t have to implement. 
• Most devtools products require engineering time to adopt new features, which kills momentum. Instead, teams should ship features that can be turned on instantly, by anyone. 
  
  • Example: once Cloudfare is installed, features can be activated via UI with no engineering support. This means anyone can test, iterate, and adopt—fast.

Tactic: How Twilio’s Rules Engine Locked-In Customers 

• International messaging introduced complicated rules (e.g., sending hours in France, carrier requirements in Vietnam).
• Twilio decided to build a rules engine into the platform – detecting contextual norms to reformat and route messages automatically, and allowing customers to create custom rules
  
  • Customers could self-serve, adapt rules, and manage complexity without rewriting code
• Over time, these configurations became sticky as rules got layered in and tracking them all became near-impossible. This made switching to a competitor extremely difficult. 
  
  • Jeff thinks about developers to get you in the food, but product managers and users within the company will change how the product works over time. 

‍