Dev Ittycheria - The Database Evolution - [Invest Like the Best, EP.373]
What this covers
The conversation is with Dev Ittycheria, who became MongoDB's CEO in 2014 and took it public in 2017. It starts with the company's own history. Revenue grew from roughly $30 million to a near $2 billion run rate, and cloud went from 2% of revenue at the IPO to 68%. Ittycheria credits the document model for developer adoption, since it matched how developers think. He explains why databases are so sticky, using Oracle as the example of a product that persists without being liked. He also explains why MongoDB must win workload by workload, which makes selling expensive and developer virality essential.
The middle of the episode covers enterprise sales and people management. Sales material includes the Three Questions of Sales, the distinction between a champion and a coach, and a rule never to answer an RFP for a deal you did not shape. On management, he describes a Skill-Will Matrix, a three-step approach to accountability, and the idea that A-players hire A-players while weaker leaders hire worse. He also covers why leaders should assume bad news is worse than it sounds, because it is dulled on the way up.
The last stretch looks forward and is more contested. Ittycheria applies S-curve technology adoption to AI, arguing it will rise while blockchain, crypto, AR and VR remain unproven. He expects roughly tenfold developer productivity gains to grow demand for software, and expects AI to give small businesses access to legal and financial expertise. He also argues that raising too much capital breeds bad behavior, citing BladeLogic, and ends on mentorship, his deepest worry for MongoDB, and his personal investing approach.
Ittycheria argues that durable software winners come from pairing a great product with a ferocious go-to-market engine, and that leaders must be relentlessly judgmental, inspecting performance and confronting bad news early, to sustain excellence through technology transitions.
- Value in each technology transition accrues first at the infrastructure layer, while application-layer returns arrive later, so AI's payoff will follow the same pattern as the internet, cloud, and mobile.
- Strong sales organizations are built through recruiting, developing people, and forecasting discipline, and deals are won by answering why-the-customer-acts, why-us, and why-now.
- Performance follows inspection rather than expectation, and bad news is filtered and diluted as it climbs an organization, so leaders must dig in immediately.
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
People fail for only two reasons, lack of skill or lack of will, so leaders can map team members on a skill-will matrix to diagnose gaps; a previously skilled person can lose will after being passed over or from personal issues.
“there's only two reasons why people fail... skill or... will”
Databases are extremely sticky because applications built on a database are very hard to migrate, which explains why many Oracle databases persist even though Oracle is not well liked.
“there's still tons of Oracle databases around because it's such a sticky application”
Any sales process must answer three questions: why does a customer want to do anything (the pain or problem), why MongoDB relative to all alternatives, and why now, since without a compelling reason the customer just kicks the can down the road.
“any sales process is essentially designed to address three core questions”
A champion is someone who sells for you when you have left the room and has change-agent power and influence, unlike a coach who wants you to win but lacks personal risk or credibility, and you must ensure your champion is more powerful than the competitor's champion.
“you will got to make sure your champion is more powerful than your enemy's champion”
People overestimate new technology's impact in the short term and underestimate it in the long term because adoption follows S-curves, so extrapolating from early data points on the rise or on the downslope leads to errors, and AI will follow an upward curve while blockchain, crypto, AR, and VR remain unproven.
“people overestimate the impact of new technology in the short term but underestimate in the long term”
In enterprise software, the best technology does not always win, because large incumbents like IBM and HP have such strong account control that buyers default to the safe choice of a large organization even when a better product exists.
“the best technology didn't always win”
A B2B company needs both a great product and a great go-to-market organization, and the differentiation comes from marrying the two; a great product is necessary but not sufficient.
“the magic happens when you can marry a great product with a great go-to-market organization”
CEOs cannot fully open up to their board or team, because they worry about reactions or org changes that could marginalize people, so a CEO needs a mentor with whom they can share deepest worries and ask any question without being belittled.
“being a CEO is a very lonely job”
Raising too much capital breeds bad behavior because it removes pressure, while founding BladeLogic with only $6 million (raised right before 9/11, with $7 million still in the bank at S-1 filing out of $29 million raised) forced efficient decisions and fast resolution of underperformers.
“sometimes raising too much capital breeds bad behavior because it takes the pressure off”
When software became cloud and SaaS, small companies could access the best CRM, HR, and supply chain software that only Fortune 500 firms could afford, which grew the overall market, so AI will likely lower the cost of legal, accounting, and financial services and be a net positive by letting small businesses use top-tier expertise.
“small businesses will be able to leverage the best legal minds”
Product-market fit is best shown when customers come back and buy more, measured by net dollar retention, and public software companies should target around 120% or higher, meaning the same cohort spends 20% more a year later net of churn.
“you want rates of 120% and higher”
Surrounding yourself with the best people produces disproportionate performance gains, as shown by athletes who improve faster when placed on better teams, and the same holds for startups where quality investors and people have outsized effects.
“putting yourself in a position where you're working with the best people around you delivers higher performance”
In every technology transition, value accrues first at the bottom layer of the stack (telephony, the internet, and now compute and chips), and the returns for the application layer come later as apps that deliver customer experience, automation, or new business models mature.
“value always accrues first at the bottom layer of the stack”
In a high-growth business, the revenue or bookings forecast is a proxy for the expense forecast, so poor forecasting can burn through cash very quickly.
“your revenue forecast or your bookings forecast is a proxy for your expense forecast”
Developer productivity can increase roughly tenfold with AI, which means a step-function increase in the number of applications built, and since there is a backlog at every development team, the demand for software and data will grow rather than shrink.
“developer productivity grows 10x”
MongoDB's document-based organization aligned with how developers think and code, reducing the cognitive load of decomposing a customer or product into rows across many relational tables, which drove developer adoption.
“they came up with this novel approach of organizing data in documents”
MongoDB's cloud business was 2% of revenue when the company went public in 2017 and grew to 68% of revenue, while total revenue grew from roughly $30 million when Dev joined in 2014 to a near $2 billion run rate.
“that business, when we went public in 2017, was 2% of revenue, now it's 68% of revenue”
People perform to the level you inspect, not the level you expect, because human nature means people need challenge and scrutiny; a high degree of inspection creates a healthy dynamic and a bias to action.
“human nature doesn't operate that way. And you need people who can challenge and push you”
Bad news travels very slowly up an organization and very quickly down it, and by the time it reaches the top the sharp edges have been dulled, so a leader who hears bad news should assume the situation is far worse and dig in immediately.
“bad news travels very slowly up the organization, but very quickly down the organization”
Holding people accountable works in three steps: first be very clear on expectations, second when a miss happens the manager takes responsibility for being unclear and re-explains why it matters, and third if the miss repeats the manager can hold the person accountable without ambiguity.
“we have this three-step process that we encourage people to consider”
Unlike HRIS or CRM platforms that get standardized company-wide, MongoDB must win business application by application or workload by workload, which makes its cost of sales expensive and makes virality through developer self-selection the key to lowering it.
“We have to win business application by application or workload by workload”
A rule of thumb is never to answer an RFP for a deal you did not shape earlier, because an RFP usually means another vendor has already written the decision criteria.
“you should never be answering RFP because invariably... some other vendor has written the RFP”
A leader's quality compounds through hiring: strong leaders hire strong people, and as a result A-players hire A-players, B-players hire C-players, and C-players hire F-players, so a bad leader drives out good people and lowers recruiting quality.
“I find as As hire As, Bs hire Cs and Cs hire Fs”
Investors observe an inverse correlation between work and outcomes: their best companies require little board involvement, while struggling companies demand heavy help, so a board that hears from investors often is a warning sign.
“there's an inverse correlation between work and outcomes”
Most people fold when hard decisions are needed because they are social animals who avoid painful conversations, which is why many companies are average or mediocre, so excellence requires being judgmental about people, product positioning, pricing, and go-to-market.
“what I find is most people fold with the pressure to make hard decisions”
The most common hiring mistake is focusing on background, school, and skills, when the more important question is what drives the person; since the only enduring edge is psychological, leaders should assess grit, ego, comfort with low status, delayed gratification, and long-term orientation.
“the only enduring edge in life is psychological as human nature doesn't change”
A leader with authority who sees a problem and does not act is accepting mediocrity and penalizing the good people who work hard, so the leader becomes the problem.
“if you have the authority to solve that problem, you're not doing it, then you're the problem”
Switching costs between large language models are currently low, since users can move from OpenAI to Anthropic or LLaMA easily, but if LLMs embed persistent memory of user history, switching costs rise, making underlying data persistence stores increasingly important.
“the switching costs from going from one model to another model, it's actually quite low”
In a consumption business, revenue is recognized by usage, giving a real-time view of demand, but declining consumption can reflect customers' business slowdowns or can signal problems like misconfigured schemas, unoptimized indexes, or rising costs relative to usage, which the vendor can diagnose and fix.
“consumption business gives you a real-time view of demand”
MongoDB's biggest AI-era risk is that next-generation AI applications do not choose it as their architecture, similar to how startups no longer build on Oracle, so the company must keep its product roadmap responsive to avoid complacency.
“I can't think of a startup today who's building on Oracle”
Ittycheria's investment criteria are a big market, since outcome correlates with market size, a defensible technology advantage with proof points of durability, and a CEO who is smart and coachable, because experience cannot be compressed.
“I don't believe that there's a compression algorithm for experience.”
The early iPhone App Store was full of trivial apps such as flashlight and iBeer apps, but within a few years the platform produced transformative apps like Uber and Airbnb and enterprise apps, and AI will follow the same path from simple chatbots to real-time decision-making tools.
“the same thing is going to happen in the AI world”
A database business lives or dies on developer adoption, since without developers choosing the database there is no business, and this is why many modern challengers to Oracle, such as XML, in-memory, and graph databases, failed.
“if you don't have developer adoption, you don't have a business”
When MongoDB launched its cloud business in 2016, it was the first infrastructure service launched across the hyperscalers, which many thought was crazy because partnering with and competing against Amazon, Azure, and GCP could get them eaten alive, but customers wanted to build apps rather than manage databases.
“We were the first infrastructure service that was launched across the hyperscalers”
The common struggle for salespeople is not losing a deal to a competitor but having the customer never make a decision, so creating a compelling event to force a timely decision is essential.
“It's not that you lose a deal, but the customer never makes a decision”
Deep inspection at the front line is exhausting because it requires asking uncomfortable follow-up questions past superficial reassurance, and most people avoid it, so projects fail three months later when no one pressed further.
“at the face level, it is exhausting”
Most managers fire people too late because they introduce hope into the process, so the right response is to start the conversation and make a decision quickly once a hire is not working out.
“do you ever fire someone too quickly or too late, 99.9% of the time they'll say they fire people too late”
You can fool your peers and boss, but you cannot fool the people below you, who see how effectively you resolve problems and how rigorous your decisions are, so their view is the real measure of leadership.
“You can never fool the people below you”
MongoDB's capital allocation orients around what drives returns, investing where returns are high, divesting when investments are not working, and never spending like drunken sailors, which Dev attributes to having lived through bad markets.
“we never really spent money like drunken sailors”
Across mainframe, client-server, Internet, cloud and mobile, the cost of building apps fell dramatically, producing far more applications and data; AI should similarly deliver a step-function increase in developer productivity and therefore many more applications, benefiting MongoDB.
“the cost of building apps went dramatically down”
Career transitions reveal how thoughtful a candidate is, so asking why someone moved from company A to company B and why they chose their major shows how they will later prosecute deals, make product decisions, and generate ideas.
“those are some of the most important professional decisions you can make in your life”
Passive-aggressive behavior, where people nod in meetings and then roll their eyes afterward, is a form of duplicity, so a culture must let people speak constructively and intellectually honestly because the group is smarter than any individual.
“I view passive aggressiveness as a form of duplicity”
Bad news must be investigated and cleaned up rather than ignored, because when someone is fired the new leader often discovers many more hidden problems, meaning a leader only sees the tip of the iceberg.
“you're only seeing a tip of the iceberg”
Candidates who describe overcoming hardship in their upbringing, such as coming from a broken home or with few resources, often reveal a deep intrinsic motivation to build security for their families, which is a strong signal of future performance.
“this person is intrinsically motivated”
Many database startups offer single-purpose functionality with a few nice logos, but the real test is whether they can repeatedly win more business from the same customer or show a pattern of winning other customers with the same use case, since many customers try things as a science project.
“any customer can buy something as a science project. And this is a trap I see a lot in the database space”
The inputs that drive an exceptional sales organization are excellent recruiting of hungry and smart people, development of them in deal execution and forecasting, and consistent quarterly delivery, not a yo-yo pattern of great and poor quarters.
“the best organizations are very good at recruiting. They're very good at recruiting hungry people”
The returns on AI will show up through applications that customers see, measured by better customer experiences, cost removal through automation, or new or disrupted business models, rather than through the infrastructure spend itself.
“the returns will come in through the apps that customers see”
A sign of a great sales force is broad-based performance across reps, not one star, and an 80-20 pattern in which 20% of reps carry the team creates a poor culture.
“you don't want the 80-20 rule where 20% of the reps are killing it”
Datadog showed strong product-market fit because it closed many customers with a junior sales force and its existing customers grew about 5% month over month after purchase, meaning the sale was easy and expansion was rapid.
“the month-over-month growth of existing customers was 5% month-over-month”
Security will remain a durable area of investment because each new solution invites workarounds, and AI-enabled deepfakes will expand the attack surface, requiring verification of who a caller or emailer really is.
“the security vectors will have to expand”
MongoDB's unfinished business includes product demands for higher performance and scale, features customers request, and efficiency in go-to-market, all to meet investor expectations for the rule of 40 (high growth with increasing profitability), while competing against hyperscalers, legacy vendors, and new entrants.
“Investors want to see the classic rule of 40, high growth with increasing profitability”
If a sale requires jumping through many hoops, the product may be a unicorn that few customers will pay the pain to buy, so a simple acquisition motion is a better sign of a viable business.
“if it's a very complex go-to-market process, and you've to jump through 17 hoops”
Character flaws cannot be fixed by an employer, because parents shape character during childhood and a company only has a limited number of years with a person, so the recruiting process must understand the person's upbringing and motivations.
“if someone has some character flaws, the parents can fix their character flaws growing up”
Seventy to eighty percent of a CEO's time is spent on people issues such as who moves up or sideways, misalignment between teams, compensation, and culture, because the CEO leads through others who drive the business day to day.
“70% to 80% of my time is all around people”
Sales reps need roughly a six-month ramp to become productive, so a high-growth company must hire and plan territories two or three quarters ahead, and avoid letting veterans hoard the best accounts.
“you have to assume roughly a six-month ramp for someone to become productive”
Cisco was the world's most valuable company in March 2000 at roughly a 300x P/E multiple, and when the bubble burst that valuation looked unjustified; NVIDIA is not trading at a comparable multiple, so the comparison is imperfect even though infrastructure leaders capture early value.
“Cisco was the world's most valuable company”
Steve Walske, former chairman and CEO of PTC, mentored Dev after he was referred by Bessemer; Walske shaped Dev's view on marrying product with go-to-market, board management, capital raising during the post-9/11 downturn, and recruiting CROs including John McMahon.
“he really informed my leadership philosophy around marrying product with go-to-market excellence”
Dev's own drive to prove he is good enough stems from his upbringing, including a divorce stigma in India, immigration, and being a Christian minority, and this motivation persists through setbacks because one does not lose one's upbringing.
“I don't think you lose your upbringing”
The decade up to 2022 was a very good market that made spending easy to justify, but Dev's experience in bad markets is why MongoDB keeps discipline, since hindsight shows strong markets masked poor spending habits.
“for the last 10 years until '22... that was a really good market”
Because a CEO cannot attend every meeting, when bad news surfaces Ittycheria samples by acting on it personally, such as giving one leader feedback on a team member, which triggered Slack messages from every leader and signaled that unaddressed problems will not be solved by anyone else.
“So I do sampling, whenever I see some bad news.”
MongoDB's strategy is to remove the database as a constraint so developers can focus on business logic, which means owning enough workflow while partnering elsewhere, with AI expected to make the power of applications far greater.
“our whole goal, our whole strategy is to really actually get the database out of the way of the developer”
Sequoia is disciplined and founder-friendly in a way that avoids reputational-driven interventions such as forcing a COO on a CEO, and Roelof Botha listens carefully as the largest shareholder, with a high signal-to-noise ratio, rather than using ownership to dominate the boardroom.
“the signal to noise ratio was very, very high”
Dev's goal for employees is that their MongoDB experience changes the slope of their career trajectory, so the skills, relationships, and experiences gained shape them long after they leave, rather than treating the company as a job.
“don't come to MongoDB for a job, come to MongoDB to build a career”
Companies that are low on inspection are the default inertia mode, and MongoDB is intentionally high on inspection, which is a deliberate difference from the average company.
“the average company is low on inspection”
BladeLogic's sales force was so strong that about 35 people from it became CROs at other software companies, a track record Dev believes no other company has matched.
“about 35 people out of that sales organization became CROs”
About 70% of a development team's time is spent working with data, specifically presenting the right information to the right user with the right context, which makes the database central to how applications are built.
“about 70% of their time is spent working with data”
Once a manager reaches a set number of reps, the team should be divided and another manager promoted or hired, which is a repeatable way to scale the sales organization.
“once you get to a certain number of reps for manager, you divide and promote or hire another manager”
When Dev joined MongoDB he assessed the inherited leaders by meeting them across the business and comparing the quality of their teams, attrition data, and the quality of people they were hiring.
“I looked at like attrition data as well as the people that we were hiring”
Dev raised BladeLogic's first round of capital five days before September 11, 2001, when the bubble had burst, and raised $29 million in total while using a direct sales force.
“I got my first round of capital five days before 911”
When MongoDB went public it was trailing $100 million in revenue at negative 38% operating margin, and last year it reached 16% operating margin while approaching a $2 billion run rate, a roughly 55-point improvement while growing quickly.
“When we went public, we were trailing $100 million in revenue and negative 38% operating margins”
Although databases are among the stickiest technologies, MongoDB does not take that for granted, and invests in a customer success function engaging customers almost daily or weekly because consumption businesses cannot sell and walk away.
“databases are very sticky... we don't take that for granted”
The jury is still out on blockchain, crypto, AR, and VR, which Ittycheria does not consider proven transformative technologies, in contrast to technologies that have followed the S-curve to take hold.
“I think blockchain and crypto, the jury is still out about those technologies. AR and VR, I think there's still some questions”
The thing that gives Ittycheria the most joy is employees telling him that the value they created at MongoDB has given them economic security for their kids, homes, or savings, and he tells people to come to MongoDB to build a career rather than just take a job.
“don't come to MongoDB for a job, come to MongoDB to build a career.”
MongoDB's next focus is navigating a multi-year AI transition, which Dev sees as exciting even though running the business is a daily routine, and his deeper satisfaction comes from employees who gain economic security and career development from their time at MongoDB.
“I think it's going to be a multi-year transition.”
The kindest thing anyone has done for Dev was believing in him when he did not believe in himself, from his mentor Steve Walske, investors, colleagues and his wife, which matters because of his persistent inner doubt about being good enough.
“I think believing in me.”
MongoDB is a multi-product, multi-channel business with direct enterprise, mid-market, self-serve, and partner channels, and Dev worries about over-complexifying the business, but a one-dimensional channel model would not work given the size of the market.
“are we over-complexifiing the business? I think there's value and simplicity”
NVIDIA is not trading at a 300x P/E multiple like Cisco did in 2000, and the surge in compute demand means that the more compute available, the bigger the AI breakthroughs, so the bottom-layer comparison is weaker than many assume.
“NVIDIA not trading at 300x P/E multiple. There's obviously tons of demand of compute.”