Atlassian's go-to-market spend grew 35.8% against revenue growth of 26.0%, a sign that even bottom-up companies eventually need sales teams. The threshold for hiring has tripled to $100M ARR, but the sales build is still inevitable. Alpha Scores: CRM 63, TEAM 43.
Public filings from four of B2B software's biggest names confirm a pattern that the private market has been living with for years: sales and marketing costs are not going away, and for some companies they are rising.
Atlassian spent 35.8% more on go-to-market in its most recent reported quarter than the year before, while subscription revenue grew 26.0%. That nine-point spread pushed GAAP sales and marketing to 39% of revenue. Salesforce and Datadog each saw their S&M ratio decline year over year, but both still sit above 40%. HubSpot, the other company in the comparison, dropped to 46% from 51%.
The split is not about sales skill. It is about business model. Companies that sell per-seat with a manager-level buyer spend a third or more of revenue on go-to-market. Companies where usage expands on its own can spend less on sales and more on product. But even the lowest-spending of the four, Atlassian, is moving toward the middle. Twenty-five years in, with $6.6 billion in subscription ARR and roughly 85% of the Fortune 500 as customers, the company is still adding go-to-market headcount faster than revenue.
Why the threshold shifted
Replit CEO Amjad Masad posted recently that his company is now more than half salespeople, a number that sounds extreme until you look at the base. For years Replit had no dedicated sales team, just one engineer handling inbound purchase requests alongside three other jobs. The demand came from companies whose employees were already using the product. Masad said the thought of handing a deal to a competitor was the worst feeling he knew.
His experience echoes what Slack faced in 2015. At roughly $30 million ARR, Slack had no sales reps. Stewart Butterfield, then CEO, described the job of handling enterprise procurement as "midwifing the sale". Slack called those people account managers, not sellers. By the time Slack hit $1 billion ARR, the majority of revenue came from enterprise deals and the company was both sales-led and product-led.
What changed between Slack's timeline and Replit's is the revenue threshold at which a sales team becomes unavoidable. A decade ago, companies hit the wall around $30 million to $50 million ARR. Now the wall is north of $100 million, according to SaaStr founder Jason Lemkin. The demand piles up while you wait.
The private market data
SaaS Capital's 15th annual survey, completed in March 2026 across more than 1,000 private B2B companies, found median selling costs at 15% of ARR, up from 13% the prior year. Support and customer success spending rose to 9% from 8%. R&D held at 22%. Selling was the line that moved.
Equity-backed companies spend 70% more on sales and 100% more on marketing than bootstrapped peers at the same revenue. Taking venture money buys the obligation to build the team.
What AI does and does not change
Emergence Capital surveyed 560-plus venture-backed B2B software companies. SDR and BDR teams took the biggest hit: 36% of companies cut that headcount over the prior 12 months, and only 19% added. But only 14% cut sales engineers. Account executive headcount grew at 28% of companies. Professional services expanded at 34%.
The cheapest, most automatable layer is compressing. The expensive, technical, relationship-carrying layers are expanding. Vercel COO Jeanne DeWitt Grosser told SaaStr AI 2026 that a lead qualification agent took a 10-person function down to about one and a quarter people, running at roughly $5,000 a year in infrastructure. The displaced people moved into higher-value roles. SDR quotas went up 30% that quarter.
The cost of waiting
Companies that defer the sales build defer the sales infrastructure with it. Account data, usage signals wired into the CRM, a real definition of an account versus a workspace – all are cheaper to build at 100 customers than at 100,000. The companies that hire a sales team late often do so under time pressure, with a founder who has never run a compensation plan or a forecast meeting.
Masad's conversion came from a specific experience: watching a deal get won that he knew would otherwise have been lost. "What changed my mind wasn't a book or a mentor. It was watching the dinners work," he wrote. Founder-led selling does not scale, but it is the only reliable way to learn what a real objection sounds like.
AlphaScala data
Salesforce (CRM) carries an Alpha Score of 63, labeled Moderate, in the Technology sector. Atlassian (TEAM) scores 43, labeled Mixed. Both trade on the back of subscription models that still require significant go-to-market investment to convert enterprise buying intent.
Atlassian's own trajectory illustrates the point. The company has long argued it does not have a direct sales force, leaning on a channel of resellers and channel partners. That third-party selling model kept reported S&M ratios lower than peers. But the growth in go-to-market spending suggests the channel itself requires more internal support, and the unconverted demand inside accounts Atlassian has already won is large enough to justify the build.
"I'll call whoever needs calling. I'll get on a plane and because of that, we rarely lose," Masad wrote. That sentence captures the tension. Founders who sell personally learn the objections. Founders who hire a VP to run a motion they have never run cannot evaluate the results. The first sales hire is harder to make at a $200 million company with no sales culture than at a $5 million startup where the job is obviously to build from zero.
Prepared with AlphaScala editorial tooling from the source reporting linked above. Indexable analysis may include a cited Alpha Score value. Publishing checks screen each story before release. Educational coverage, not personalized advice.