Asana (ASAN) 2027财年第二季度业绩电话会议：AI驱动25%的净新增ARR

核心要点

• 2027财年第二季度营收同比增长10%至2.164亿美元，高于Asana指引区间的上限。StackAI对已公布的增速贡献了约50个基点。
• Non-GAAP营业利润率达到10%，同比增长约300个基点。调整后自由现金流为4230万美元，利润率为20%。
• 整体以美元计算的净留存率升至97%。核心客户以及年支出至少10万美元客户的净留存率均为98%。
• AI Studio与AI Teammates创造了约25%的净新增ARR，高于上一季度的17%。年支出至少10万美元的客户中，已有超过25%购买了其中一款或两款产品。
• Asana将于9月中旬开始将新客户、自助服务客户以及销售引导续约的客户迁移至智能体工作管理（Agentic Work Management），并将AI Teammates和Asana Dash嵌入到付费套餐中。
• 管理层预计2027财年营收为8.585亿美元至8.635亿美元，Non-GAAP营业利润率约为10%。该展望包含因向基于消费的AI定价过渡而产生的120万美元收入确认时点逆风。

核心财务业绩

剔除2026财年第二季度签署的一项大型多年期合同，当前RPO增速从上一季度的8%加速至约11%，而总RPO增速从7%加速至约12%。

业务与运营业绩

AI采纳是本季度业务扩张的主要驱动力。AI Studio与AI Teammates贡献了约25%的净新增ARR，而第一财季为17%。若剔除Asana最大的AI交易，该贡献率接近22%。

Asana与一家《财富》500强媒体公司签署了一项为期三年、价值数百万美元的扩容协议。AI Studio与AI Teammates占合同价值的近一半，并超额弥补了席位数量缩减的影响。管理层将该协议视为基于消费的AI收入能够减少公司对员工人数和席位增长依赖的证据。

客户案例包括Indeed，该公司借助AI Studio每年节省了超过1,400小时的高管时间，创造了约30万美元的成本节省与额外产能。Washmen则报告称，在使用AI Teammates后，理赔解决时间从三天缩短至六小时，降幅达90%。

美国业务同比增长10%，为两年多来首次恢复双位数增长。管理层将这一加速归因于科技客户预订额和留存率的提升、AI采纳以及新客户的获取。科技垂直领域连续第二个季度实现同比增长，而非科技客户的增速继续快于公司整体水平。

智能体工作管理（Agentic Work Management）将融合AI Teammates、AI Studio与Asana Dash。付费套餐将包含一定的AI请求配额，且不会改变层级定价。Asana选择“请求次数”作为其消费单位，并由公司管理模型选择与路由，以平衡输出质量和成本。

Asana客户管理（Asana Client Management）和Asana服务管理（Asana Service Management）处于早期体验阶段。Command计划于9月晚些时候进入早期体验，并计划在今年晚些时候进行更深层次的OpenAI Codex集成。管理层预计这些应用在2027财年贡献的营收微乎其微，目标是在2028财年带来更显著的贡献。

管理层业绩指引

全年展望假设StackAI对营收增长贡献约50个基点，固定汇率提供约20个基点的顺风。管理层目前预计包括StackAI在内的AI产品将占全年净新增ARR的20%左右，高于此前约15%的目标。

Asana预计年末毛利率将处于80%的中段水平。由于在确认相关消费收入之前产生的成本，以及低毛利率AI产品占比的增加，公司预计在第三和第四财季将面临约150个基点的毛利率压力。

第三财季还包含约300万美元用于智能体工作管理及智能体应用发布的增量支出。管理层预计该支出将在发布后恢复正常，从而使营业利润率在第四财季恢复环比扩张。

风险与关注点

• 较低的产品驱动增长（PLG）预订额继续对营收基数产生叠加影响。管理层预计第三财季营收增速将面临约100个基点的逆风，第四财季为150个基点。
• 年支出低于5,000美元的客户群仍是公司整体NRR的主要制约因素。Asana正在将获客支出重新调配至匹配度更高、生命周期价值潜力更大的客户、行业和使用场景。
• 基于消费的AI会计处理带来了更大的波动性，因为收入是在使用请求时确认的。预计AWM套餐过渡将在下半年将120万美元的收入推迟至未来期间，但不会影响ARR、预订额、账单金额、递延收入、RPO或现金流。
• AI基础设施、开发成本以及毛利率较低的StackAI业务拖累了毛利率。管理层预计模型匹配、路由和其他交付优化随着时间的推移将改善AI的经济效益。
• 2027财年指引假设当前的PLG趋势不会复苏，且Client Management、Service Management和Command的贡献微乎其微。

分析师问答亮点

管理层解释称，智能体工作管理将通过自动推荐相关的AI Teammates来减少AI探索摩擦。Dash可以在用户交互期间推荐智能体，任务输入可以触发上下文建议，管理员可以使用工作图谱分析器在整个组织中识别有用的智能体。

关于PLG增速放缓，管理层表示，情况较3月份发现的约2个百分点的ARR逆风仅略有恶化。第三和第四财季的展望假设趋势不会较第二财季进一步恶化。公司正在优先保证理想客户画像的质量，而非漏斗顶部的数量。

关于Asana在服务管理、客户管理和软件开发协调领域的竞争能力，管理层强调了企业工作图谱、共享上下文、权限和审计追踪。公司表示，这些能力使AI智能体能够利用先前的工作流历史来解决或分发工作，而不是作为孤立的单点解决方案运行。

管理层还表示，模型匹配和路由已显著降低了AI交付成本，足以支持嵌入式AI能力的广泛推广。尽管推广会带来短期毛利率压力，但Asana预计优化和规模效应随着时间的推移将改善毛利率状况。

在定价方面，管理层描述了一种将可预测的订阅层级与基于请求的消费相结合的混合结构。公司承认在初期对消费的预测精准度较低，并表示希望在发布后观察使用情况、付费扩容和续约行为，从而获得更好的能见度。

业绩电话会议完整文字记录

---

完整财报电话会议逐字稿

管理层陈述

Operator

Thank you for standing by, and welcome to Asana's Second Quarter Fiscal Year 2027 Earnings Conference Call. [Operator Instructions] I would now like to hand the call over to Eva Leung, Investor Relations. Please go ahead.

Eva Leung

Good afternoon, and thank you for joining us on today's conference call to discuss the financial results for Asana's Second Quarter Fiscal Year 2027. With me on today's call are Dan Rogers, our Chief Executive Officer; and Aziz Megji, our Chief Financial Officer.

Today's call will include forward-looking statements, including statements regarding the expected release and benefits of our product offerings and our expectations for revenue to be generated by those offerings, our retention and expansion opportunities, our expectations for our financial outlook, including our fiscal year '27 full year guidance, strategic plans, our market position and growth opportunities, and our capital allocation strategy, including our stock repurchase program, among other items.

Forward-looking statements, including risks, uncertainties and assumptions may cause our actual results to be materially different from those expressed or implied by the forward-looking statements. Please refer to our filings with the SEC, including our Annual Report on Form 10-K and our most recent quarterly report on Form 10-Q for additional information on risks, uncertainties and assumptions that may cause actual results differ materially from those set forth in such statements.

In addition, during today's call, we will discuss non-GAAP financial measures. These non-GAAP financial measures are in addition to and not a substitute for or superior to measures of financial performance prepared in accordance with GAAP. Reconciliations between GAAP and non-GAAP financial measures and a discussion of the limitations of using non-GAAP measures versus the closest GAAP equivalents are available in our earnings release, which is posted on our Investor Relations website at investor.asana.com.

With that, I would like to turn the call over to Dan.

Daniel Rogers

We delivered a solid second quarter, exceeding our expectations on both revenue and profitability with continued improvement in the underlying health of the business. There's three things I want to point out this quarter.

First, the business continues to get healthier. Growth is accelerating, retention is improving again, and we saw a broad-based strength across industries and geographies. Second, while still early, our AI products are creating a new growth and expansion vector beyond our traditional seat-based model. Customers adopting AI Studio and AI Teammates are engaging more deeply, retaining better, and expanding faster than the broader customer base. And we believe this gives us an early validation of our opportunity to build meaningful consumption and outcome-oriented revenue stream. Third, we're acting on the learnings by bringing AI Teammates, AI Studio, and Dash together as a core part of the Asana experience through our Agentic Work Management product.

We want our customers to experience these capabilities early and naturally as part of how they work every day rather than a separate AI products that they have to discover and purchase. And we're going to be bringing that same orchestrated execution across humans, agents, and systems with Asana Client Management, Asana Service Management and our Command products.

So let's have a look at this quarter. Improving health of our core business validates our strategy. It gives us confidence in the investments we're making to drive future growth. Revenue was $216.4 million, up 10% year-over-year and above the high end of our guidance. Reported net retention improved in every cohort we report. Overall NRR improved to 97% from 96%. In quarter net retention improved for the fifth consecutive quarter. Core customers NRR improved to 98% and our largest customers, those spending over $100,000 or more improved to 98% from 96%. That improvement is being driven by broader multiproduct adoption within the largest customers creating additional pass-through expansion. The technology sector delivered a second consecutive quarter of year-over-year growth.

Now while growth remains modest, we're encouraged by the continued acceleration in this vertical. That growth included another expansion with a leading AI lab this quarter, adding seats, in addition to the expansion with AI Teammates that we mentioned last quarter as well as a global streaming service to both expanded seats and added AI Studio. Outside of tech, the story has been consistent for more than a year. Non-tech continues to grow faster than the company's overall growth.

In fact, we added new customers across a range of industries this quarter, including one of the largest telecommunications operators in the U.S., a large insurance operator in the U.S., a Big 4 professional services firm, one of the world's leading law firms, and an iconic American luxury jewelry brand. We also saw encouraging acceleration in the U.S., where revenue grew 10% year-over-year in Q2, returning to double-digit growth for the first time in over 2 years. This growth acceleration is attributed to improvement in both bookings and retention in our tech customers, which are concentrated in the U.S., strong adoption of our AI products, and acceleration in new logo acquisition.

Internationally, Darktrace and a leading U.K.-based financial service company were notable new logo wins for our EMEA team, and Delivery Hero expanded its relationship with Asana, including our AI products. Looking now at our AI product momentum. Momentum across our AI products continued to build this quarter. And while still early, we're seeing an encouraging validation of the opportunity to build meaningful consumption and outcome-oriented growth and expansion revenue streams alongside our traditional seat-based model. AI Studio and AI Teammates, in fact, drove about 25% of our net new ARR, up from 17% last quarter. This is above our 15% full year target, which we set in March. We find customers that are adopting our AI products, engage more deeply, retain better, and expand faster than the broader customer base.

This shows up most clearly in our largest accounts. More than 25% of our $100,000-plus customers have now purchased AI Studio or AI Teammates. This has been a key contributor to the NRR expansion we're seeing up market. Also seeing clear evidence that AI products can mitigate seat-based pressure while creating new expansion opportunities tied to usage and outcomes. This quarter, we signed our largest AI expansion deal in Asana's history, a 3-year multimillion-dollar agreement with a Fortune 500 media company, spanning AI Studio and AI Teammates with AI products representing almost half of the total contract value.

What's particularly important is the role our AI product played in the expansion. The customer is operating with a smaller workforce which historically would have resulted in a seat contraction. Instead, the investment in AI Studio and AI Teammates more than offset a smaller footprint, resulting in a modest overall expansion with also the additional upside potential of consumption growth over time. And they're already seeing measurable value.

In fact, in one creative marketing workflow, AI Teammates have already reduced the content operation cycle time by 30%. This is an important example of how our AI products are creating new growth vectors beyond seats, allowing us to expand with customers, based increasingly on the work and outcomes delivered through Asana rather than changes in headcount. We're seeing customers move beyond individual use cases to make Asana a core part of their broader agentic enterprise strategy, coordinating humans and AI across the workflows that run their businesses.

Asana is becoming the operating system for human agent teams for them. Let me share a couple of examples of what that looks like in practice. Indeed is a great example of how enterprises are using our AI products together to remove manual coordination at global scale. The world's #1 job site deployed AI Studio to automate project discovery and the technical scoping for its analytics teams. It also runs the dynamic intake and triage across the 70-person in-house creative agency, which operates in more than 60 countries and 28 languages. Annually, that work reclaims more than 1,400 hours of senior level time. It's cut lead time from raw request to active project by 60%. It's reduced manual ticket management by more than 40% for the creative team and delivers roughly $300,000 in savings and unlocked capacity. Indeed is also piloting AI Teammates as an autonomous brand auditor, matching localized content to global brand guidelines across dozens of languages.

Washmen, a UAE-based textile care business is another example of AI Teammates running an operation end-to-end using AI Teammates to agentify their customer support and returns process. So when a garment comes in, one teammate researches its retail value. The second reviews the care plan for risk. Third, checks it against every past claim, and the fourth handles compensation and drafts the customer message.

A person steps in only when a teammate escalates, the result is 90% faster claim resolution, taking it from 3 days down to 6 hours. These kind of results reinforce our belief that our AI products create the greatest value when they're being embedded in business-critical workflows with a shared context that enables people and agents to coordinate and execute together towards outcomes. This principle is at the heart of what we're bringing to market in mid-September with Agentic Work Management. So let's take a look at Agentic Work Management.

Let me explain what we mean here because this is a real meaningful evolution of our product. Not simply label on traditional work management. Individuals have experienced significant productivity gains from AI, but most organizations haven't yet translated that into the productivity gains at the enterprise level. AI often sits outside the workflows that run the business, requiring people to find the right agent, provide the right context, and bring the output back into the work.

With AWM, we closed that gap by putting people and agents and systems on the same plan. Historically, customers use Asana to coordinate work between people, to provide visibility into those tasks. With AWM, they can orchestrate execution across people and agents in the same context, same goals, and the same governance. AWM brings 3 things into every paid package tier. First, AI Teammates, including more than 30 prebuilt teammates for marketing, operations and IT. These are preapproved and ready to work and pretrained with no prompt engineering required.

Second, AI Studio, so that any team can build no-code workflow automations for intake, routing, approvals, and status. And third, Asana Dash, this is your AI chief of staff that knows a person's goals and priorities, pulls decisions out of meetings, e-mails and chat, and surfaces what needs their attention and keeps them that one step ahead.

So what does this mean for customers when AWM comes to market later this month? Well, beginning mid-September, all our new logos, self-service customers, and sales-led renewals will be moving to AWM. And they'll start with AI Teammates, AI Dash built directly into their package tier. This includes an allotment of Teammates and Dash requests.

Most importantly, rather than trying to find the right agent, the Teammates will surface themselves based on what a customer is trying to accomplish. This is deliberate. We want customers to experience the full value of Asana early. Similarly, the full allotment of request is designed to let customers put our AI products to work in their mission-critical workflows from day 1. By simplifying the purchase decision, we can get more customers to first value faster and create a natural path from demonstrated outcomes to deeper AI adoption to increase consumption and stronger seat retention and expansion over time.

We chose requests as the unit of consumption because we want our AI pricing to be customer-friendly, simple, and predictable. A request gives a customer a clear understanding of what they're buying with a consistent price per request, speed limits, usage visibility, and alerts. And behind the scenes, Asana is going to select and optimize the appropriate model. That complexity should be ours to manage, not the customer's.

So AWM is how we bring the operating system for human agent teams to customers today. People and agents running those cross-functional work that runs the business. Asana Client Management applies the same orchestrated execution to client delivery, service management to service delivery, and Command to product development, same platform, different kinds of work. We're not entering these markets with point solutions. Each is a purpose-built application built on top of the enterprise work graph that our customers are already running on. So each starts with that same shared context, memory and governance, the people, systems and agents need.

And the AI Teammates and automation a customer builds in one application carry into those others under the same permissions and audit trail. Each of these new products represents a large adjacent market and new buying center. So let's take a look at them.

Starting with Client Management, The promise here is simple. The complete client workflow coordinated across clients, account teams, delivery teams, AI, files, approvals, budgets, and projects. Nearly 1/3 of our customers today are already doing some form of client delivery or running a professional services team today. But they often run client delivery in Asana while managing the rest of the client relationship across disconnected systems communication and e-mail, statements of work and approvals elsewhere and resourcing and spreadsheet. That makes it really difficult for them to maintain a single view of client health, project profitability and team capacity.

ACM brings those pieces together. It has a branded client portal for requests, reviews and approvals, AI Teammates that draft statements of work, client-ready assets, and status updates, and time and budget tracking sits alongside actual work. Client Management is in early access right now.

Next, let's have a look at Asana Service Management. Traditional service management was built to route a ticket to a person and track it to resolution. Well, AI has changed that model, enterprises increasingly want service teams to resolve requests automatically, not simply route them faster.

Asana Service Management is one AI-native service platform for IT, HR, facilities and legal with 24/7 agents that can resolve routine requests through Slack, e-mail, or a portal before they even reach a human. Service Management builds on that with one front door for every department, a self-learning knowledge base that gets more accurate with every resolved case, and agentic resolution that moves Asana from a place where service work is tracked to a place where it's actually resolved. ASM is in early access now with strong feedback from IT design partners, particularly around the self-learning knowledge base.

Finally, looking at the Asana Command. As we know, AI has made code generation dramatically faster, but the coordination around that code hasn't kept pace. The spec, the handoffs, the release plans, the traceability. Increasingly, that's where the bottleneck now sits. Coding agents need more than the ability to generate code, they need context, a shared plan, and the decision history they can trust. Command provides that planning and orchestration layer built on the same enterprise work graph that already supports product and engineering planning teams today. That's the promise, ship faster with humans and agents in sync.

We designed Command as an open platform from day 1. So customers can orchestrate the agents and tools, they choose rather than being locked into any one proprietary agent ecosystem.

As SpaceXAI described it: "Command is a novel approach to a difficult problem. Coordinating work across many agents and tools modern engineering teams use. Its open platform design lets developers bring SpaceXAI into a broader orchestration there without being locked into a closed system."

Later this year, Command will also integrate deeply with OpenAI's Codex. This will bring parallelized cloud-hosted coding agents natively into how work gets planned, assigned and shipped. Command, reaches early access later this month.

Turning now to StackAI. StackAI is about turning your business processes into governed agentic workflows in minutes. Reading, writing, and executing across all the systems that the company already runs on. While Asana provides the plan, the shared context, and the people around that execution. Importantly, it gives us a more complete solution to enterprise AI transformation initiatives we're increasingly seeing from our IT and AI transformation buyers. And in that motion, we've already seen early wins including one of Australia's largest retailers. We believe these engagements are early validation of the opportunity to bring Asana and StackAI together for larger, more complex enterprise workflows.

In closing, taken together, we're expanding Asana in 2 dimensions. AWM gives us a path to drive deeper product adoption across our customer base and create meaningful long-term consumption growth alongside seats. While our new applications expand the workflow users and buying centers we can serve, all of it is running on the same architecture and advances our strategy to become the operating system for human agent teams.

With that, I'll turn it over to Aziz, to take you through the quarter and the outlook.

Aziz Megji

Thanks, Dan. Let me start with the quarter. Q2 revenue was $216.4 million, up 10% year-over-year, an acceleration from Q1 and above the high end of our guidance. StackAI contributed approximately 50 basis points to reported growth which was in line with the expectation we shared last quarter. Currency impact was immaterial this quarter. We have 26,778 Core customers which we define as customers spending $5,000 or more on an annualized basis.

Revenues from Core customers grew 11% year-over-year, and this cohort represented 77% of our revenues in Q2. We now have 890 customers spending $100,000 or more on an annualized basis. This represents a growth rate of 16 percentage points (sic) [ 16% ] year-over-year. As a reminder, these cohorts are measured using annualized GAAP revenue during the quarter and therefore, can be affected by the number of days in the quarter.

Our dollar-based net retention increased on every cohort we report. Our overall dollar-based net retention was 97%. Core customer NRR was 98%, and among customers spending $100,000 or more NRR was 98%. As a reminder, our NRR is a trailing 4-quarter average and therefore, a lagging indicator of more recent trends. This improvement is being driven by the continued strength in gross retention, healthier seat expansion within our largest enterprise customers, and broader multiproduct adoption, AI Studio and AI Teammates increasingly creating an expansion vector at renewal. As Dan discussed, that allows us to expand with customers in ways that are less dependant on seat growth alone.

Turning to self-serve. The PLG headwind we discussed last quarter builds throughout the year. The impact of lower PLG bookings compounds into the revenue base each quarter. So the drag on reported revenue growth increases even if the underlying self-serve trend does not deteriorate further. That pressure comes as several of our underlying growth acceleration levers are improving. NRR continues to strengthen. We're experiencing strong momentum with our AI products. Our U.S. business has accelerated and technology vertical has now returned to year-over-year growth for 2 consecutive quarters. It also explains the gap in our net retention.

Core and our $100,000-plus cohort are both at 98%, while company-wide NRR is 97%. That differential sits in the sub-$5,000 cohort which is concentrated in self-serve and skews towards customers outside our ideal customer profile. Getting company-wide NRR back to above 100% really comes down to 3 levers. First, gross retention improvement in the core and enterprise space; second, seat multiproduct and consumption expansion in those same cohorts; and lastly, improving ICP mix and driving stronger retention and expansion in the sub-$5,000 customer base. The first 2 are already starting to show benefits, and you see that reflected in our Q2 KPIs and financial results.

The third remains a key focus area and we expect the investments we are making there to contribute to improving NRR in FY '28, improving the growth in NRR within our sub-$5,000 customer base is centered on 2 areas. First, we're focusing our acquisition spend on the customer sizes, industries and use cases with the strongest fit and highest lifetime value potential. That includes becoming more targeted and verticalized with industry-specific team templates, AI Teammates and use cases designed to improve conversion and retention.

Second, we are increasing the surface area through which these customers can expand with us. AWM and ACM launched in self-serve in mid-September, bringing AI Teammates directly to our large PLG installed base while extending Asana into new workflows and use cases. We believe this creates a new vector to get deeper into critical workflows and expand these relationships beyond seats, which we feel will improve retention over time.

Now moving to profitability, where I'll be discussing non-GAAP results and year-over-year comparisons. We delivered a 10% non-GAAP operating margin in Q2, expanding approximately 300 basis points year-over-year while continuing to make significant investments in our AI products and agentic applications and the go-to-market capabilities to scale them. Our gross margin was 87%, which was down approximately 120 basis points from last quarter. This decline reflects 3 primary factors: first, higher AI infrastructure and compute costs attributed to onetime scaling and development costs for our new products, which accounted for approximately 80 basis points of the change. Second, the addition of StackAI, which has a lower gross margin profile, given its subscale accounted for approximately 30 basis points of the change. And third, the remainder of the gross margin impact reflects the mix shift from seats to our AI products.

R&D expenses were $50.7 million or 23% of revenue. Sales and marketing expenses were $88.2 million or 41% of revenue. G&A expenses were $28 million or 13% of revenue. Net income was $23.8 million or $0.10 per share on a diluted basis. We have kept our overall expense base relatively flat while adding capacity in lower-cost regions such as Poland and using AI products to increase productivity and expand capacity across our teams. We're seeing that most acutely in R&D, where AI is enabling our teams to deliver the most robust product road map in Asana's history without a commensurate increase in R&D spend. The combination of a more efficient talent footprint and AI-driven productivity gives us the capacity to continue investing behind our highest growth opportunities while driving operating leverage over time.

Moving on to the balance sheet and cash flow. At the end of Q2, cash, cash equivalents, and marketable securities were approximately $340 million. Our remaining performance obligations, or RPO, was $522 million, and current RPO grew 10% year-over-year. This represents 81% of total RPO and will be recognized over the next 12 months. The underlying RPO trends was stronger than the reported growth rates suggest. This is due to the comparison against the large multiyear contract we signed in Q2 of last year. Excluding that contract, current RPO growth accelerated to approximately 11% from 8% last quarter while total RPO growth accelerated to approximately 12% year-over-year growth versus 7% year-over-year growth last quarter. Our total ending Q2 deferred revenue was $350.7 million, up 12% year-over-year.

Adjusted free cash flow was $42.3 million or 20% on a margin basis. Note free cash flow benefited this quarter by approximately $5 million from stronger collections than expected. Before I turn to guidance, I want to connect the product strategy Dan described to the evolution of our financial model. In mid-September, we are including a base level of AI Teammates and Dash requests in the AWM tiers without changing tier pricing. This changes both for new and existing customers. We're seeding that usage deliberately, investing to drive adoption first, with the expectation that stronger retention, seat expansion, increasing consumption follow over time. Underpinning this shift, we have made significant investment in our monetization infrastructure and in-product experience, enabling AI native capabilities such as usage metering, overages, and consumption-based billing at scale.

Let me walk through how we reflected that transition in our guidance. There are 2 dynamics affecting revenue recognition as we transition towards consumption. First, going forward, all new AI Teammates sales will be consumption-based with revenue recognized as customer requests are consumed rather than ratably over the contract term. Because customers have flexibility in the timing of their consumption, this also introduces greater variability in the timing of revenue recognition.

Second, as we transition our core packaging from CWM to AWM and embed our AI products into the core subscription, a portion of subscription value that historically would have been recognized ratably is now allocated to AI consumption and recognized as that capacity is consumed. As customers ramp consumption over time, this shifts a portion of revenue recognition into future periods. The shift of new AI Teammates sales from ratable to consumption-based recognition, along with the AWM packaging changes creates a $1.2 million revenue timing impact in the second half. This is roughly split between Q3 and Q4.

Note this is just a timing impact. It does not change anything in customer economics, has no impact on ARR, bookings, billings, deferred revenue, RPO, or cash flow. In addition, this transition also creates approximately 150 basis points of gross margin pressure across Q3 and Q4, reflecting both costs incurred ahead of associated consumption-based revenue recognition and the growing mix of AI products, which currently carried lower contribution margins than our seat-based business.

Importantly, we're making these investments deliberately to seed AI usage and drive deeper utilization of the platform with expected benefits to retention and expansion occurring over subsequent renewal periods. As a result, we expect gross margin to be in the mid-80s exiting the year. We've already seen meaningful reductions in the cost of delivering our AI products through optimization and routing. And we expect those efficiencies to continue as we scale. Importantly, as you'll see in our operating margin guidance, we've been able to absorb the remaining increased costs through efficiencies and productivity gains elsewhere in the cost base while continuing to deliver margin expansion ahead of our expectations.

Note, this is all while absorbing approximately 1 percentage point of incremental operating expense as a percentage of revenue from the StackAI acquisition as we discussed last quarter. Second, the PLG headwind discussed earlier continues to weigh on the second half revenue growth profile. We estimate approximately 100 basis points of pressure to revenue growth in Q3, which increases to 150 basis points of pressure in Q4. Our outlook assumes that current PLG trends persist through the balance of the year and incorporates no recovery in FY '27 from the initiatives I discussed earlier.

Third, AI Studio and AI Teammates represented about 25% of net new ARR in the quarter or closer to 22%, excluding the large deal. Including StackAI, we now expect AI products to represent approximately 20% of that new ARR for the full year, which is up from approximately 15% of net new ARR, which we discussed in March. We're deliberately prudent with this target because seeding every customer with AI Teammates and Dash starting in mid-December may delay some consumption package purchases by a matter of months. This metric captures only new consumption and capacity package purchases, not the requests and credits included within the AWM tiers. No attribution is being made from the AWM packaging change. Fourth, we continue to assume minimal FY '27 revenue contribution from Client Management, Service Management and Command. Given enterprise sales cycles and deployment time lines, we expect the financial contribution to become more meaningful as a key growth driver in FY '28.

Finally, Q3 includes approximately $3 million of incremental AWM and agentic application launch investment consistent with what we discussed last quarter. That investment is concentrated in global brand and marketing and AI go-to-market activities around our September launches. We expect that spend to normalize following the launch with sequential operating margin expansion returning in Q4.

Now moving to guidance. The guidance I'm giving includes all the assumptions I mentioned above. For Q3 fiscal 2027, we expect revenue of $217 million to $219 million, representing 8% to 9% growth year-over-year. This includes a $700,000 headwind to revenue from our AWM packaging transition. We expect non-GAAP operating income of $18 million to $19 million, representing an operating margin of 8% to 9%.

In addition, we expect non-GAAP net income per share of $8 -- $0.08, assuming diluted weighted average shares outstanding of approximately 236 million shares. For the full fiscal year 2027, we expect revenue to be in the range of $858.5 million to $863.5 million, representing growth of 9% year-over-year at the midpoint of the guidance. The full year revenue reflects the outperformance from our Q2 results and the expected contribution from StackAI of approximately 50 basis points to growth, same as last quarter.

In addition, as mentioned above, it included a $1.2 million headwind to revenue from our transition to AWM and consumption. We expect an approximately 20 basis point tailwind to our full year revenue in constant currency which is consistent with what we shared last quarter. We expect non-GAAP operating income of $84.5 million to $86.5 million, representing an operating margin of approximately 10%. And we expect non-GAAP net income per share of $0.37, assuming diluted weighted average shares outstanding of approximately 239 million shares.

As we look ahead, AWM brings our AI products to our broader customer base, creating new expansion opportunities as adoption and consumption grow. We're investing ahead of those benefits while maintaining our margin commitments, creating the foundation for stronger growth and operating leverage over time.

With that, operator, we are now ready for questions.

Operator

[Operator Instructions] Our first question comes from the line of Patrick Walravens of Citizens.

分析师问答

Patrick Walravens

Great. And Dan, congratulations on all the progress on the product side around Agentic Work Management. There was one thing in your prepared remarks that stuck out to me, and I would love to hear more about it. You said, rather than having to find the right agent, the right teammate can surface based on what the customer is trying to accomplish. That sounds like a very good idea to me. How is that going to work and maybe you could share a simple example of a teammate surfacing to help the user?

Daniel Rogers

Yes. Thanks, Pat. And you're right, that is a good idea. And we think it's a bit of a game changer. Just to kind of level set on AWM, so Agentic Work Management. So this is the evolution of collaborative work management. The big idea here is that we think humans and agents are going to be working together and coordinating together to drive orchestrated execution. And we spent really the last, I'd say, 6 months figuring out how we want AI Studio, AI Teammates and our new AI chief of staff that we call Dash to appear to our customers. And what we found is, the more we can bring that directly into their experience, the better. So you'll see in September some, I'd say, innovative ideas on how we create this amazing experience.

So innovation number one is our AI chief of staff, Dash, as you ask it questions and interact with it, it will suggest the right teammates to help you complete your execution of that task. Number two is through input nudges. We'll actually recognize the type of task that you are trying to complete and suggest one of the prebuilt pre-skilled teammates that can help you. And that's because, of course, we've got all of this great work graph history, so we know exactly what kind of work you're trying to do. And it's no coincidence that we've built these 30 prebuilt teammates, those are exactly the kinds of work that our customers are doing today. So that matching will happen.

And then finally, if you want at the administrative level to do what we call a work graph analyzer, you can actually do that across all of your work and the admin can easily see which teammates could be most useful to help. And this is all in response to the idea that -- today, one of the biggest hurdles of agentic enterprise is actually the discovery of the agents being able to find the right ones that will work for you.

Operator

Our next question comes from the line of Steve Enders of Citi.

Steven Enders

Okay. Great. I guess I want to dig in a little bit more just in terms of the factors being included in the guidance outlook on the revenue side, in particular. And I guess, one, better understand the PLG headwind dynamics and I guess, what exactly maybe change there in the guidance here versus last quarter? And then I guess with the headwinds that we're talking about on the packaging side as well, just how should we think about that continuing, I guess, beyond Q4 and going into next year for the potential impact that these factors could have here?

Daniel Rogers

Yes. Thanks. So just to frame that up, I'd say, looking forward, 2 things we're really excited about, one is work-in-progress. So what are we excited about? The first is, as you see, we are now manifesting our vision as the human-agent operating system. We've got so much good stuff ahead of Agent Work Management. And then you see all of these other buyer-specific products of Asana Client Management, Asana Service Management, Command, and Stack. So really 5 new products that we're excited about.

Number two is you saw the upmarket strength, and you saw that this quarter, manifest as an increase in NRR across our $5,000 cohort across our $100,000-plus cohort getting up to 98% now. and AI adoption across the board, whether that's to every customer, which was -- you saw us say, 25% of our net new ARR is now coming from AI products. And then also for those large customers, in fact, over 25% of our greater than $100,000 customers have AI attached. So real excitement there, and then the work-in-progress is PLG. The dynamics changed a little bit.

The thing that we're now, I'd say, encouraged by is customers do still want to engage digitally, a digital discovery, digital playing, and many customers want to fully use and consume in a digital engagement. That remains true. What is new is the top of the funnel can get very clogged up with, I'd say, heavy tire kickers. And the best thing that we can do is focus all our efforts in making sure that the customers that are coming in and actually paying are the right customers for us that they're our ICP. So you'll see us focus a lot more of our efforts, a lot more of our marketing dollars on our ICP. And as they do so, as we get the right ICP into our funnel, because of that product strength, we now have so much more to delight those customers with. AWM will be in our PLG funnel. ACM will be in our PLG funnel. And so a lot more customers will have a lot richer and deeper experience early as they get used to Asana.

Aziz Megji

And Steve, just to add on to Dan's point, we're really encouraged what we're seeing about upmarket, just another KPI I'll call out is just the growth in RPO and cRPO. So if you actually back out the large customer renewal, multiyear renewal we had in Q2 '26 of RPO and cRPO. RPO accelerated from 7% year-over-year growth last quarter to 12% this quarter and cRPO from 8% to 11% this quarter. And that's really the best proxy for upmarket and enterprise growth. So we're seeing really strong traction there.

Also with our $100,000-plus customer cohort, that accelerated to 16% year-over-year on a customer count basis from 12% last year. And importantly, we're driving this upmarket strength with efficiency. Our sales and marketing spend has been really flat over 2 quarters. So we're seeing stronger sales efficiency there. So as you think about how that upmarket strength is manifesting in our consolidated growth and our guidance.

As Dan called out, the PLG piece is really masking that. So we called out a 2-point headwind to ARR back in March. That actually gap has widened a bit. And the impact of the Q4 headwind, the Q1 headwind and now again in Q2 on revenue growth compounds each quarter, so that ARR impact gets greater each quarter, where in Q3, it's about 1%. And in Q4, it grows to about 1.5 percentage. So that's underlying our guidance. And then you add the packaging transition to AWM, having about a $1.2 million impact in the second half or 30 basis points. That's just timing, and a lot of that is just created because it's the first quarter we're moving to that. It will normalize and should normalize in Q4 and subsequent in 2028, and we'll get that timing impact back in subsequent quarters.

So I think you asked whether that will grow or have a bigger headwind going forward. It won't. It actually have the biggest headwind in Q3, Q4 and then normalize thereafter. So if you take those 2 things in account and you think about our guide, especially with the $1.2 million, we beat Q2 by about $2.4 million. We raised $1.5 million. We had this $1.2 million impact we didn't foresee in the last couple of quarters. So in absence of the $1.2 million impact from the transition from CWM to AWM, we would have rolled the full beat and then some. So just putting into context how we're thinking about the guide. And just to reinforce these new products that were coming out of super excited, but we have not factored any contribution from them in our FY '27 guidance.

Operator

Our next question comes from the line of Billy Fitzsimmons of Piper Sandler.

William Fitzsimmons

I think great segue here. In terms of, Dan, a lot of new products rolling out in the second half, Command by Asana, Service Management, Asana Client Management. These products obviously expands your TAM. But in some cases, you're competing against new vendors. So Dan, I love that you could kind of talk about what is Asana's right to win in these spaces. And you touched on this a little bit, but what has to be done from a go-to-market standpoint as these products go GA to kind of get them out to customers.

And then as these -- I appreciate that last point there. So to be crystal clear, it sounds like potentially of adoption for these new products is better than expected. It could be a source of upside in the back half. Is that fair to say?

Aziz Megji

Yes, I'll start off. Dan goes, that's fair to say.

Daniel Rogers

Yes. So -- thanks for the question. Returning to your first piece about new products, new TAMs and what's our right to win in those areas. I guess the first piece to think about is, I wouldn't think of them as just single products. This is a platform. The platform is an orchestrated execution across every team. And the platform itself has many of these differentiators built-in, really orientating around the work graph. So the platform itself promises instant productivity for any of the agents that run on it.

Why? Because we can quickly recognize all the relevant work, we can recognize who work needs to get routed to. It also promises increased velocity because there are a lot less handoffs if you know exactly who's supposed to get it next. There's no back and forth of e-mail and Slack. And then it promises the ability to control and manage the enterprise risk of those agents because every agent is auditable. So if you think about that, now apply that to those new products, so we already know a lot about these workflows. It turns out we've served IT teams. We've served R&D teams. We've served HR teams. We've served client delivery teams. We know exactly what tasks and work is and what the workflow looks like. So we get to bring an agentified solution to those workflows now based on all of the deep, rich data we have on how those workflows actually travel. So I'll give you kind of one example.

Let me do this for Asana Service Management. So Asana Service Management on Asana looks like a request might come in through a single portal or it might come in through Slack. Well, we'll understand the context of that request because we have this rich data. So instantly, we're now able to do one of two things, either a, resolve it instantly using AI, or b, route it to exactly the right person that we know is capable of dealing with that, with all of the full project context and history. Then when we actually make a resolution the resolution isn't just trapped in e-mail as an example, but part of the work graph itself. And now when the next request comes in, we know exactly how that in turn was solved in the last time.

So this is a kind of dynamic learning system that's all baked off this orchestrated execution platform. So that's our right to win. And so what does that lead us to? Yes, sometimes we'll be working alongside some of those point solutions. And sometimes, our customers may want to consolidate their spend on Asana.

Operator

Our next question comes from the line of Elizabeth Porter of Morgan Stanley.

Elizabeth Elliott

I wanted to follow up on your comment about Asana being able to select and optimize the appropriate AI models for customers and you guys taking on that complexity as opposed to pushing it down. So what is the impact to your efficiency to be able to deliver AI and more cost effectively. Is this something where you could start to see greater savings that benefit the margin or more likely pass through in order to drive more share and usage within AI?

Daniel Rogers

Yes. Thanks. I'd say, look, this is a growing competency and we're getting rather good at it. And I would say the piece that we've gotten rather good at over the last, say, 6 months to a year is figuring out which types of tasks should go to which types of model. And so something that may come in as a, I'd say, a generic request or a net new task type, we're doing pretty good categorization now of passing that out into the right model, to both solve for quality and cost optimization.

And so this will in turn lead to a much better gross margin profile as we're able to deal with that request, and we'll talk about request in a second is the unit that we're charging customers on so that we can deal with that request most efficiently, both in terms of efficacy of the outcome for them but also the cost delivered. And so yes, in the beginning, I'd say the gross margin burden, we've taken that a lot on our shoulders. But over time, you'll be able to see, I'd say, getting a much better gross margin profile from that.

Aziz Megji

Yes. And just to add, as we were determining the scope of the AWM launch, whether this would be new customers only or taking it to specific segments or bringing it to the full entire base like we are. The progress we have made reducing the cost of delivering our AI products particularly through the model matching and routing that Dan just mentioned, gave us confidence that we could go to the broader base while keeping the cost of that rollout manageable and mitigatable.

And you've seen that while it's having a 150 basis point impact into COGS in the second half because we're investing ahead of the benefits, we've been able to rationalize other places in the cost base to still deliver the margin expansion above our expectations. And so that was an important determinant of how broad we were going to go, and how broad we're going to go allows us to spark that adoption and that flywheel of adoption leading to better seat dynamics leading to consumption much sooner and much broader.

Operator

Our next question comes from the line of Jackson Ader of KeyBanc.

Jackson Ader

Great. The question I had was about the seeding the market in AWM and kind of trying to reduce the friction for AI adoption across your 3 AI products. I'm just curious, like what friction are you hoping to alleviate by going to this kind of embedded packaging, was like price a hurdle? Is there so much noise from every software vendor or AI vendor that like people didn't necessarily know what they could access via Asana? Like what is it that you're hoping to alleviate by embedding this in everybody's package?

Daniel Rogers

My short answer would be yes, and then I'll expand on that a little bit. So there's a great productivity gap in AI, which is, individuals have seen massive improvements in the productivity by interacting with chat agents. They become much more productive in code generation, much more productive in document generation.

But oftentimes, enterprises haven't been able to translate that into real productivity. Why? It's because the AI is not actually part of their core workflow. It's not part of what teams do every day as teams. It's not part of the handoff process between teams. It's not part of the, let's say, coordination that's required to actually get work done in an enterprise. So what are we trying to solve? It's really that. It's how do we embed AI more deeply into the workflows that actually matter to our customers. So yes, there's a discovery part to that. We want to make sure that the agents are imminently discoverable.

But also, anything that the agents do actually operates within the context of a team that they are actors within the same work pattern as your humans. And so that's literally why we call them teammates. They are things that multiple people can interact with and improve upon and to interact with humans in the loop every time. So these are going to be much more deeply embedded in your day-to-day work. And because they're so discoverable, we think the cost of discovery has gone down, but also your ability to try these things out has also gone down.

And that ability to keep the multiplayer basically means everyone gets to take part, everyone gets to make them better over time. And when you add the work graph to it, you get this nice additional benefit, which is all of the work that you do to make your agents better, all the work you do to make your workflows better, make the very next run once again better in turn. And so benefits kind of compound and that's often what's missing in some of the single-player chat interaction today.

Operator

Our next question comes from the line of Rob Oliver of Baird.

Robert Oliver

Great. With 25% of net new ARR now coming from Teammates and Studio and would -- really, I think, underscores the case you guys have laid out for now, now being the right time to kind of transition here to AWM. I'm curious, you talked about and Aziz, you mentioned in detail, I appreciate all the detail, some of the impact on rev rec as the move consumption happens. You guys also mentioned in the prepared remarks, outcome-based pricing. And I would love to get some more color on how outcome plays into your thoughts and expectations about AWM as it ramps and how that potentially influences your ability to forecast the business?

Daniel Rogers

Yes, I'll try and describe some of the philosophy here. So our customers want predictability in pricing, but they also want things to tie as closely as possible to the value that they're achieving. Predictability, definitely comes to a, let's call it, like a subscription-type model. But in order to tie to value, yes, we need to more and more tied to the outcomes that they were delivering together. And so that's really where the hybrid model come in. So how should we tie our pricing to value.

Well, we've decided that the unit that we're going to anchor on is requests. We've seen, of course, other companies with their endeavors around tokens or around credits or putting the burden on the customer themselves to choose the model and do model optimization. We, kind of, say we want to [indiscernible] all of that. We think request is the most customer-friendly possible unit.

Why? Because it's literally how you interact with Asana. You will ask it or your teammate to do something or help with something and then fulfill that request. And so we think it's a very natural idea that is, honestly, as customer-friendly as we could imagine. So the hybrid model is essentially a predictable piece that really does scale up and down with the size of the organization. And then also a knowable piece, which is how many requests do you want this system to deliver to you the outcomes of. So yes, we think that's the right customer-friendly mix.

Aziz Megji

Yes. And then on the forecasting, I'll be honest, our forecasting position on this in a year from now will be better than it is today. So that we've taken some prudence in how we've built this AI product target, raising it from 15% to 20%. The seeding should accelerate adoption in users, but it can push out the timing of incremental paid consumption. So we factored that in and how we have designed the 20%. And we'll learn a lot more post launching in a couple of weeks about how customers are adopting how fast the seeded credits are leading to expansion.

And then upon renewal, how they're impacting and influencing the seat renewal and seat expansion, which is a -- it's not part of that AI metric, but it's an influence and an attribute of seeding that we look to drive over time.

Operator

Our next question comes from the line of Taylor McGinnis of UBS.

Taylor McGinnis

So, given that it sounds like up market has been pretty strong and the weakness is in the PLG motion. I'd love to ask you a question on that and what you're seeing in terms of top-of-funnel activity there. Were those demand trends stable? Or have they become more challenging in 2Q and 3Q? And just as we think about the 150 basis points of impact of 4Q revenue, does that mean that in FY '28, you'll see a similar headwind of 150 basis points? Or how should we think about that as we look beyond this year?

Aziz Megji

Yes. So the impact -- so to answer, kind of, is it getting worse in Q2, Q3? It is, but not materially. So I think we called out the 2 points of ARR headwind back in March when we reported Q4. That's gotten a little bit worse, but more to the tune of about 50 basis points. The impact we called out on revenue is really from Q4, Q1, Q2. We don't expect and have not factored in Q3 and Q4 to further deteriorate from what we saw in Q2.

And all the efforts that Dan outlined in terms of driving the right top of funnel, not just the volume but the ICP mix, whether it be the size the industry of the customer. We see that the right ICP drives the right LTV. And then with new products and additional surface areas to procure ACM, AWM, the expansion opportunities with Teammates and Studio. It just amplifies that. So now you have a higher LTV customer with more to buy. It just creates better ACV and expansion outcomes and retention.

So and as we called out, the real inhibitor right now to getting to 100% plus NRR we're seeing is in that self-serve cohort, which is concentrated in less than $5,000. And if you look at, our Core is at 98%, if you kind of back in what that means on in-quarter based on the improvement, our in-quarter is trending towards 100%. And really what's driving down the consolidated is that sub-$5,000. So we don't expect this headwind to persist in the same level in FY '28.

Operator

Our next question comes from the line of Rishi Jaluria of RBC.

Joshua Trautman

This is Josh standing in for Rishi. You guys mentioned looking to improve the sub-$5,000 customer base. And I just wanted to sort of dig into that a little bit. I was curious around how you're balancing developing the product to be -- to appeal to a broader audience and sort of being out of the box for giving a customer size while also balancing the specialization that comes with verticalization. And just a little bit more context around that would be great.

Daniel Rogers

Yes. Well, I'll say all of our 5 new products serve really every segment rather well. And it's about how deeply you adopt it and which kinds of workflows you will use against them. So if you take Agentic Work Management as an idea. Well, it turns out, if you're a small business, you're a large business, you will want to have pre-built agents that are working alongside you.

Which ones you pick from that menu of 30 will depend, of course, how thorough you've built out those departments because these are essentially like packaged up agents that are prebuilt for your department. And so if you have a well-tuned, let's say, campaign department, then you're going to absolutely love the campaign orchestration agent. If you have a well-tuned launch process, you're going to love the launch agent. But similarly, if you're a small business and potentially you want to improve your reporting, then maybe you're going to use the reporting agent.

So I don't think the size of the company or really how they engage with us is going to gate how much they love these products. And then I'd say things like a Asana Service Management, if you have a, let's say, a large service department or you have a lot of manual service requests. Clearly, you're going to get a lot more value from that than if those departments are maybe immature or haven't started yet. So I'd say all of our products really serve all of those segments, and that's really part of the strength of Asana is, we have a great digital discovery, digital trial, digital experience and both small businesses and large businesses come to know us through that digital engagement.

Operator

Thank you. I would now like to turn the conference back to management for closing remarks.

Eva Leung

Hi. Thank you, everyone, for joining the call today. We are on the road attending the Citi and Piper Sandler conference in the coming weeks, and we'll also have a marquee Work Innovation Summit in New York on October 14. Hope to see you all there. As always, if you have any questions, please reach out to me at ir@asana.com. Thank you very much.

Operator

This concludes today's conference call. Thank you for participating. You may now disconnect.