Atlassian’s Land-and-Expand Model: What B2B Software Founders Can Learn

Many software founders describe their company as product-led. Atlassian’s latest annual filing shows what that phrase looks like after the slogan is stripped away: a customer starts small, the product carries much of the early selling work, and revenue grows when the software spreads to more users, teams and workflows.

The Australian-founded collaboration-software company reported more than 350,000 customers at 30 June 2026. It generated US$6.572 billion in revenue during fiscal 2026, up 26% from the previous year. Subscription revenue accounted for 95% of the total, and more than 90% of revenue came from customer accounts that already existed at the beginning of the fiscal year.

That last figure is the centre of the case study. Atlassian did not depend only on finding a new buyer for every additional dollar. Its filing says the year’s revenue growth was driven mainly by greater demand from existing customers, while subscription growth came primarily from more paid seats and price increases. This is the economic promise of land and expand—but it also reveals the product, service and trust obligations hidden inside the model.

The first sale is deliberately small

Atlassian says a typical customer begins with a free edition or one product used by a limited number of people. Its website is the primary point of engagement for new customers and processes thousands of commercial transactions each day. Automated purchasing and low-touch support reduce friction before a traditional sales team becomes economical.

For a B2B startup, the principle is more useful than the label. The first version should solve a recognisable job without requiring weeks of procurement, configuration and training. A bookkeeping application might begin with invoice capture. A logistics tool might begin with proof of delivery. A compliance product might begin with one evidence register. A narrow entry point gives a customer a reason to try the product and gives the supplier real usage evidence.

Self-service is not the same as no service. Someone must still design onboarding, explain pricing, protect customer data, answer billing questions and help users recover when something fails. The work moves into the product and operating system rather than disappearing. If activation depends on a founder manually repairing every account, the company has a sales-assisted service disguised as self-service software.

Expansion is the primary growth engine

Atlassian measures whether small accounts become economically larger. The number of customers producing more than US$10,000 in cloud annualised recurring revenue rose from 51,978 at 30 June 2025 to 57,334 a year later. The company says these customers represent the majority of cloud revenue.

Expansion can come from several directions: additional employees using the same product, another department adopting it, an upgrade to a higher-value edition, or the purchase of another connected application. Atlassian’s direct sales team concentrates on those larger relationships. That makes its model hybrid rather than purely product-led: software and automated systems handle much of the landing and early growth, while people help navigate complex enterprise expansion.

A smaller SaaS company needs similarly honest metrics. Registrations and website traffic are weak substitutes for activated organisations, retained paying accounts and expansion. Useful measures include time to first value, the proportion of invited users who become active, retention by customer cohort, support cost per account, revenue expansion from existing customers and the percentage of customers using more than one paid capability.

One product becomes a connected system

Atlassian’s portfolio includes Jira, Confluence and Jira Service Management alongside collections and other applications. The products are designed to pass work and context between teams. That connection supports expansion because a customer can add another workflow without choosing an entirely separate operating environment.

This is not a licence for a startup to build a suite too early. Every adjacent module increases development, security, documentation and support obligations. The practical sequence is to earn trust around one repeated workflow, identify the neighbouring task where customers already experience friction, and add a connection only when it improves retention or willingness to pay. Breadth is valuable after the core product is credible; before that, it can dilute the reason to buy.

The marketplace extends the product—and the risk boundary

Atlassian also operates a marketplace for third-party applications. Its corporate marketplace page advertises more than 4,000 apps and integrations, while the annual report describes the marketplace as a way for customers to discover add-ons and for developers to reach Atlassian’s customer base. Forge, Atlassian’s cloud development platform, gives developers hosted infrastructure and tools for building those extensions.

An ecosystem can deepen a software platform without forcing the core company to build every specialist feature. It can also create new distribution for independent developers. But the platform remains exposed when an external app is insecure, unreliable or abandoned. Atlassian’s risk disclosures explicitly warn that third-party defects or vulnerabilities can disrupt customers, expose data and damage its reputation.

A young platform considering an app marketplace therefore needs more than an upload form and a revenue share. It needs review standards, permissions that limit unnecessary data access, security requirements, incident contacts, version support, customer disclosure and a credible removal process. The commercial benefit of more integrations must be weighed against the expanded trust surface.

Cloud migration changes both the offer and the economics

Cloud products generated US$4.411 billion of Atlassian’s fiscal-2026 revenue, compared with US$1.831 billion from Data Center and US$331 million from Marketplace and other revenue. The company stopped selling new Data Center licences to new customers in March 2026 and plans, subject to limited exceptions, to end maintenance and support for those on-premises versions in March 2029.

Moving customers to cloud gives a software provider greater control over releases, infrastructure and integrated services. It can also make new AI and automation features easier to distribute. The transition is not costless. Atlassian warns that cloud hosting fees can pressure margins, that some customers may lack the security or control they need to migrate, and that cloud revenue may be lower in the initial year than revenue recognised from a Data Center contract.

The lesson for founders is that recurring revenue quality cannot be judged from an annualised number alone. A cloud provider must understand gross margin after hosting and support, data-location and compliance obligations, renewal behaviour, the cost of migration and the amount of contract revenue deferred into future periods.

Product investment is expensive, not magical

Atlassian spent US$3.269 billion on research and development in fiscal 2026, equal to 50% of revenue, and US$1.541 billion on marketing and sales, equal to 24%. More than half of employees were involved in research and development at year-end. Those figures support the company’s emphasis on product and platform investment, but they are not a benchmark an early-stage company can safely copy.

The same accounts show the tension. Atlassian delivered an 85% gross margin and US$1.319 billion of free cash flow under its stated non-GAAP definition, yet reported only US$10.4 million in GAAP operating income and a US$53.8 million GAAP net loss for the full year. Stock-based compensation and restructuring charges materially affect the distance between different profit measures, so readers should compare definitions rather than presenting one margin as the whole result.

What founders should take—and what they should leave

The transferable model has five parts: make the first useful outcome easy to reach; let price and packaging support a small initial commitment; measure whether customers expand because the product becomes more valuable; introduce sales assistance when organisational complexity justifies it; and treat integrations, security and support as product responsibilities.

What cannot be copied is Atlassian’s scale. A company with hundreds of thousands of customers, billions in recurring revenue and a large research organisation can fund a product portfolio, developer platform and enterprise sales motion that would overwhelm a young firm. Product-led distribution is not a shortcut around customer acquisition, and land and expand is not proof of expansion until retained cohorts actually spend more.

Atlassian’s fiscal-2026 evidence makes the business logic visible. The website helps the company land customers, connected products create routes to expand them, subscriptions spread revenue across time, direct sales supports complex accounts, and the marketplace increases usefulness beyond the core product. The difficult work is keeping every layer reliable enough that a small initial adoption becomes a larger relationship instead of a quiet cancellation.