Ramp 增长引擎内幕：按天计时、看回收期、把实验失败做彻底

Batchu confirms Ramp's reputation: an analysis by Packy McCormick showed it was the fastest company to reach 100 million dollars of annualized revenue at the time, in about two years, and last year it grew 4x. Okta data on SMB and mid-market software also named Ramp the fastest-growing, he says, while the company stays under roughly 500 people, which implies unusually high revenue per employee.

Beyond product quality, he highlights something he calls cap table as a growth strategy. Co-founders Eric and Karim had a prior successful exit and a strong reputation, and they brought many early-stage founders and operators onto the cap table. Many early customers were those founders' companies, such as the founders of Eight Sleep and Pod, and some investor firms became customers too.

Investor introductions mattered less than one might expect, Batchu says, because companies run their own decision processes and weigh the view of another customer who actually uses the product more heavily. Today the vast majority of Ramp's customers are mid-market and enterprise, not startups. And while existing customers expand and the product suite grows, he says the vast majority of growth then and now comes from new customer acquisition.

Batchu declines to share exact percentages but says Ramp's distribution across outbound sales, paid marketing, field and other channels would not look far off from similar companies. The difference is that each investment is driven by technology and data. Sales is highly efficient because a dedicated growth engineering team supports it with third-party data and AI automation, work that began almost two years earlier, before AI became a buzzword.

That team shares a goal with sales: pipeline driven and the payback period of the channel. Engineers feel ownership of the quota and not just product metrics, which lets them bring bottom-up projects that improve efficiency and top line. Their work spans finding the right prospects, sending the right messaging, prioritizing responses and drafting replies.

The wider growth organization has channel teams for paid marketing, lifecycle CRM and field marketing, plus a product engineering team for growth and sales, self-serve activation, and SEO and website teams. A small skunk-works team handles cross-channel experiments, such as TikTok, Reddit, referrals and first-party events, with small bets that can be scaled if they work.

Batchu says investor Keith Rabois is right that Ramp's velocity is unprecedented. The core, he says, is a focus on cycle time and getting people to think in smaller units of time. CEO Eric shows the number of days since founding, 1,529 at the time, at every board meeting and all-hands, to remind people they work in days and not quarters. The growth team, though cross-functional, works like a product team on two-week sprints and cross-prioritizes together.

Coming from Instacart and Opendoor, he noticed how fast people answer on Slack, and that even unfinished threads end with a clear owner and deadline. The company builds in public, so work is visible across teams. To avoid constant interruption, people use calendar blocking and calendar audits, with a template from the people team, on the idea that you ship your calendar.

On burnout, Batchu agrees with Lenny that hard work matters early in a career. In his experience, highly successful people are driven by autonomy, flexibility, mission alignment and happiness in the work, not hours. He emphasizes appreciating the team and celebrating wins, and mentions that nearly everyone participates in the Zoom chat during all-hands.

Batchu observes that growth leaders used to come from marketing or product, but more now come from investing and analytics backgrounds because growth has become a diversified portfolio of bets at reasonable ROI. His aim is to make all channels more efficient while allocating more to efficient channels that can scale, including pushing awareness through owned and earned media. Lenny adds that most companies start with one growth engine and eventually end up running all of them.

One surprise is PR as a growth machine. Fundraising announcements, when used to create a market moment, have driven a non-trivial amount of top-of-funnel. Both men stress that PR works only when there is something genuinely interesting, so Ramp adds extra color about the business to funding news.

On newsletters versus traditional press, Batchu says newsletters and podcasts reach tech founders and help recruiting, but most Ramp customers are no longer tech or startups. Traditional PR therefore remains important because it targets a different audience.

Batchu calls the debate over team structure a red herring and argues that culture, rituals and cadences matter more. A great growth team has one, at most two, simple north star metrics, defines data-driven hypotheses, and runs quick MVP-style projects where people can fail, learn and iterate fast. The goal, especially early on, is a group that generates and evaluates ideas effectively.

He likes two metrics, one for volume and growth that is motivating and something the team can directly affect. At Instacart the north star was monthly active orders, because revenue sits too far downstream. With over 300 engineers in the consumer growth organization, each local team had its own influenceable metric, such as checkout speed, and finance and data built a translation layer converting each into MAO impact.

That common currency made cross-prioritization easier, answering questions like where extra engineers or budget would produce more MAO per dollar. Batchu admits the factors are imperfect, so the team treated them as a guide, used judgment for marginal calls within roughly 5 to 10 percent, and refreshed them every six months. Teams also used long-term holdouts for each surface area to measure cumulative impact, including a permanent holdout of Instacart users who never saw ads.

At Ramp the approach is similar, and Batchu says it grows more important as companies get bigger. Recently the growth north star was dollars of SQL pipeline, so a landing-page copy change would be translated from conversion rate into pipeline dollars to decide whether it was worth the effort.

Batchu says there should be two success metrics, one on volume and one on efficiency. A good volume metric links clearly to value creation, like Facebook's MAU or Instacart's monthly active orders, yet remains intuitive and traceable from small projects. For engagement he recommends a user's escape velocity: 10 friends in seven days at Facebook, three orders in the first month at Instacart, and four events in 30 days for Ramp activation.

He relates this to Gibson Biddle's growth, engagement or monetization framework and to the speed, quality and cost trio used at Opendoor, noting that you cannot optimize all at once. His view of a growth team's journey is to build a system that moves fast, then improve quality, then optimize cost.

On efficiency, he argues CAC focuses on cost and not value, so cutting it can mean bringing in less valuable customers. LTV-to-CAC is better but hard for a four-year-old company to predict, and mistaken churn assumptions can destroy value. He prefers payback period using contribution margin, the profit after all variable costs, such as a 5,000 dollar acquisition cost against 500 dollars of monthly profit for a 10-month payback.

Batchu's general path for B2B is founder-led sales, then the first salespeople, then low-cost targeted marketing such as content, community and small events, then PR, and only afterwards paid and brand efforts. SEO starts around the time of paid. Channels get more expensive, more effective as you learn the customer, and more scalable as you go down the list, and SEO takes time to build and can flail without domain authority.

On experimentation, Batchu says growth experiments succeed about 30 percent of the time, so teams should celebrate failure as long as they learn, which requires failing conclusively. Otherwise the same idea returns with each new executive. Because B2B lacks the big samples of consumer products, he recommends maximizing the treatment effect by applying every plausible tactic, and rationalizing cost later if it works.

His example is account-based marketing, which companies often test halfway and then abandon, only to repeat it later. At Ramp the team maximized both the number of target accounts and the variety of touches. He says the approach matters most for large cross-functional tests, while quick tests like website conversion or email tweaks can be redone cheaply.

For tools, Ramp uses Airtable for sprint planning and scoring with its translation layer, and Mutiny for website personalization using third-party data. Batchu says Ramp is thoughtful about build versus buy and buys external tools for non-strategic needs, while building prospecting, lead scoring and customer communication automation in-house. At Instacart and Opendoor, by contrast, the teams built their own experimentation systems.

On hiring, he describes two approaches. One is a network search for the best person in a field, asking each contact for the best person they know and focusing on companies one or two stages beyond yours. The other is data-driven: for example, using Similarweb to see which companies drive the most traffic through email and recruiting from those teams.

Batchu holds a spicy view on compensation: small teams of strong people beat large teams of mediocre ones, so companies should be able to reward 10x operators with 10x pay. Talent density depends on retention and performance management as much as hiring, since interviews produce many false positives. He adds that parting ways with people who are not a fit is almost never about bad actors, but delaying it lowers the talent bar.

Batchu says his love languages are spreadsheets and frameworks, and he recommends MECE, mutually exclusive and collectively exhaustive, when brainstorming solutions. His example breaks a revenue slowdown into revenue per customer and number of customers, then into products, new acquisition, activation and retention. Structuring a problem this way ensures nothing important is missed.

In the lightning round he recommends Never Split the Difference by Chris Voss, whose core lesson he describes as listening to what the other person really wants. His favorite interview question is what something you are bad at but still do, since candidates who cannot answer may run at the first sign of trouble. A process change that mattered was merging separate sprint planning for product and marketing into one cycle, so lifecycle marketers joined the activation team.

For saving money, he notes that software contracts are negotiable, especially near quarter end. His second tip is to hire slower, choosing for slope over intercept and hiring only when teams are stretched. He says small teams accomplish more, as Ramp does against competitors of similar revenue, and notes Ramp is hiring.