How's everybody doing? Welcome to HubSpot's Analyst Day here at INBOUND in San Francisco for the first time, which we love. We can totally get used to this. It's great to see a ton of familiar faces in the room here. Appreciate everybody making the trip, some places from near and other cases from afar. I also want to extend a warm welcome to those of you joining the webcast. I'm Chuck MacGlashing, and I have the pleasure of leading the Investor Relations team here at HubSpot. I hope you all had a chance to catch Yamini and Karen's keynotes from a bit earlier. I think they did a phenomenal job of highlighting all of the innovation that we've delivered over the last year for our customers, right?
A ton of new product announcements today that we're incredibly excited about, and of course, just why we're so excited about the year ahead. Are we getting feedback here? OK, I'll keep going. Our Safe Harbor Statement in all of its glory. No analyst day would be complete without one of these and a bunch of, yeah, just a lot of words. But you can find this up on our website at ir.hubspot.com. We'll be loading up all the presentations from today at the end of the day, so just ask that you be patient as we get that done post the sessions. Onto the agenda and the fun stuff here. So in just a moment, our Chief Executive Officer, Yamini Rangan, will take the stage to share our strategy and vision for driving durable and efficient growth in an AI era.
After Yamini, our CEO. Our CFO, Kate Bueker, will take the stage to walk you through how we're executing and kind of delivering against that vision. And then Yamini, Kate, Dharmesh, and I will take the stage for about 45 minutes' worth of Q&A up here, so get those questions ready. I would say Dharmesh is being a real trooper, so he's going to come up here and do Q&A with us. He has his main stage keynote right after, and so if you see him kind of rush out of here pretty quickly without exchanging a ton of pleasantries, you'll sort of understand why. And then around 1 o'clock, maybe a few minutes after, there'll be an announcement. We will pack up our stuff and head down to the main stage to watch Dharmesh's keynotes on learning, building, and growing at a period of unprecedented change.
One other thing I'd call out too is just after we finish up, there will be about a 30-minute period of time where we'll have an ability to connect with the team and mingle a little bit just before we head down to the main stage there. So with that, it's great to see everybody excited about Act Two here of Analyst Day, and let's go ahead and get started. Thank you.
I can almost with certainty tell you that every single representative, whether they'd be from marketing, sales, or service, have a much better experience in utilizing HubSpot. We have a CRM that's actually usable for all of our organization.
With HubSpot, we're able to take our creativity and our insights as marketers, pair it with some really easy-to-use tools that make it easy to deploy customized experiences for these audiences.
The cool thing about Breeze Studio is that the tools that are integrated really help our team save time. It really made our content creation process so much more efficient. It is a real time saver.
Customer Agent is phenomenal. The ability for it to just browse my knowledge base and be able to provide a would-be customer in a chat, "Hey, I need an answer to this question." It searches the knowledge base, answers the question. I love that.
HubSpot's value as a connected platform, having all of those hubs, all of those pieces of functionality in one place. We're all working off of the same source of truth, and it's completely transformed how we collaborate and measure impact.
Please welcome to the stage Yamini Rangan.
Hi, everybody. Welcome to Analyst Day. How's everyone doing? Great. Good to see so many of you here in San Francisco, my hometown. Nice to be here. And welcome to everyone who's joining us online as well. I'm really excited to share how HubSpot is leading in a world that is getting transformed with AI. Now, our industry is just evolving at hyperspeed. How we build product, how we deliver value, and what customers expect from software is getting redefined in real time. And in moments like this, clarity matters. So today, I want to provide a clear view of how HubSpot is leading with AI and why we're so excited about the opportunity ahead. Now, here's what we hope you will take away. We are transforming our platform to be an AI-powered customer platform. We have rich customer context, which is our platform advantage.
We are reimagining marketing beyond search with a new playbook, products that support it, and an ecosystem behind it. And we are scaling up market and down market to drive durable growth. And we are transforming as a company to be AI-first. Now, all of this positions us to lead and win in the era of AI. But there's a lot here, so let's dive in. We're going through a big technology shift, but more importantly, that is raising the bar for customers' expectations of us. And it is providing new ways in which companies can grow. Customers today expect us to resolve tickets, write blogs, schedule meetings, just like they would a coworker. So customers are expecting not just software that does the work for them, but actually does help them get more accomplished to grow. That's a big shift. And that unlocks a huge opportunity for HubSpot.
When we look at this opportunity, we are moving from delivering software to delivering work. We're no longer limited by the software budgets. We are now tapping into the work budgets. And the way we are doing it is by augmenting all of those manual tasks within go-to-market teams with AI. And that means for HubSpot, it's not just defending a fixed market. It is leading and winning in a much larger, fast-expanding market. We see a generational opportunity here to deliver products that do work, not just support the work, to reimagine marketing beyond search with a new playbook, products, and the whole ecosystem, and to power millions of organizations that are now scaling with AI. We saw the shift early. We moved fast, and we are transforming our platform, our pricing, and our entire company to lead and win in this new era of work.
Let's get started with how we are transforming our platform. Our product strategy has been very clear and consistent. We want to be the number one AI-powered customer platform for scaling companies. For more than a couple of decades, we've built for SMBs. We know them. We know their goals, their challenges, and how they grow. Now, we want to be the best at applying AI to help them grow. That means we are transforming ourselves into an agentic customer platform. What does that mean? Before AI, we helped our customers take actions in hubs. Now, we are helping our customers take action in both hubs and agents. Before AI, the context wasn't structured data: company, contacts, tickets, deals. Now, it is in structured data, unstructured conversations, and all of the external intent signals. Before AI, we enabled our customers to collaborate across their go-to-market teams.
Now, we are enabling hybrid teams of humans and agents to get work done. This is a complete reimagination of how companies can scale, and it actually allows HubSpot to add more value to our customers. And the way we are doing it is by bringing together three interconnected layers: context layer to know your customer, action layer where work happens, and an orchestration layer to connect everything together. Each layer is powerful on its own and stronger together, so let me spin through each of these layers, starting with the context layer. This is the brain of our platform. This is where all of that rich data comes together to know your customer, and at the center of it is the Smart CRM, a single source of truth for marketing, sales, support, operations teams to get accurate information about their customers and know them.
Now, this is so powerful that I really want to emphasize it because point solutions and point apps, they do not have all of this data across the entire customer journey. And acquired, cobbled-together solutions, they just cannot bring this data together. We bring together structured data, company, contacts, deals, tickets, unstructured data from conversations that are typically in free-flowing emails, chats, Zoom calls, and external intent signals, real-time intent data across 200 million company profiles that fill in all of the gaps about customers. Without this unified context, AI cannot deliver work, and humans cannot drive growth. That's why it's important. The next is our action layer where work happens. All that rich context comes together for humans and agents to get work done. Now, hubs like our Marketing Hub, Sales Hub, Service Hub were built for go-to-market teams to get work done. They launch campaigns there. They send emails.
They close deals. They drive growth. And that has been consistent with our strategy. Now, in addition to that, we now have Breeze Assistant, which is the AI chief of staff for every go-to-market employee. They're aware of the role, the priorities, and the context of the go-to-market employee and can provide insights and answers to help them do work. And then agents do work. They are the AI coworker that can now take on more jobs like support, prospecting, account research, and then allows the rest of the teams to do high-value work. Now, there are a whole universe of agents out there. So I want to try and simplify HubSpot's approach to agents. We build featured agents for go-to-market teams to get work done. Now, our Prospecting Agent will do account research and outreach. Customer Agent will help support tickets, but also answer questions across sales and marketing.
And the newly launched Data Agent today, it enriches customers' data with AI. Each of these agents will get work done, but they can also call on each other to do more for our customers. Then, we are providing the ability for our customers and partners to build custom agents. Today, we launched Breeze Studio to build those custom agents and then list them within the marketplace, Breeze Marketplace. Breeze Studio, Breeze Marketplace, and the whole first wave of these new agents went live today. And then, we are empowering an entire community of citizen builders on Agent.ai to create a network of agents for every go-to-market task. We have more than two million users on agent.ai, and they've built over 46,000 agents. By being the place where work happens, we're being essential. HubSpot is now essential in the AI era. Okay, let's talk about the orchestration layer.
This is to connect everything together. Now, this is where work flows across agents, humans, and other applications. Before AI, we had a whole set of APIs to connect, and now, we have APIs, MCP servers, and also connectors to bring HubSpot everywhere. Now, more importantly, the orchestration layer is the place for intelligence. Now, think about this. In a hybrid team of humans and agents to work together, you need a few good handoffs, and for those handoffs to work, you need memory about roles, about relationships, and the past work that has been done. You need evaluation of whether the output was good or not good, and you also need reasoning to know how the decisions were made. All of that intelligence is within the orchestration layer.
The orchestration layer is also home to Studio, which is the place where you can build custom agents, both customers and partners. Now, you put all of this together. This is our AI-powered customer platform, delivering value for customers. We're setting the pace of innovation, and we are staying ahead in building this agentic customer platform. Today, at INBOUND, we launched more than 200 new updates and features across the entire platform. Lots for you to go and see within the demo stations. But let me highlight the top few. We have transformed Operations Hub into Data Hub. This is where all of the data is going to come together. This is where we will enrich it with AI and activate it within workflows and automation for our customers. Marketing Hub is getting reimagined to support the Loop.
We launched the Breeze Studio Marketplace and a whole wave of new agents within Breeze. We unbundled the Smart CRM. We're going to talk about this in a few minutes. But Smart CRM now has AI data and platform value in it, including projects. And we launched our AI-first CPQ within Commerce Hub. We've always had this vision of making it not just easy for sellers to sell, but also for buyers to buy. And that is exactly what we did with the AI-first CPQ. We love launches. We like it. But what we like even more is adding customer value and driving customer adoption. That's what we care about. And we're beginning to see clear customer adoption with our AI strategy. Now, value for customers comes from embedded AI features across all of our hubs. It comes from agents that are now doing work for our customers.
It comes from Breeze Assistant and connectors that are just becoming part of everyday workflow. There are a ton of momentum stats here. I'm not going to go through all of them. But let me summarize. Embedded AI is driving growth for customers and therefore driving growth for HubSpot. You can see that in the Content Hub, driving value for customers. And therefore, the attach rate to Marketing Hub is now 50%. You can see that in Sales Hub, Service Hub, Marketing Hub, where it's adding value to every seat. And therefore, we are seeing strong seat upgrade motions. In terms of agent, we're beginning to see clear traction. We have over 5,000 customers on our Customer Agent, which answers questions across the customer journey. And the resolution rate now is 65%. Many of you will remember at earnings last month, we announced a 55% resolution rate.
Now, it's 65%. Best in class, and the same thing is happening with Prospecting Agent and Content Agent, gaining clear traction, and then Breeze Assistant and connectors, they provide insights from HubSpot to everybody everywhere. We're the first CRM to announce connectors with OpenAI's ChatGPT, Claude, and today, Gemini that we just announced this morning, and here's our focus. We want to bring customer context into the LLMs, and we want to bring AI referrals, leads, as well as actions back into HubSpot from those LLMs. That is how we're thinking about this, so we like this adoption and the momentum we are seeing with customers, but the question for all of you is, is this durable? What makes HubSpot unique and defensible in the AI era? The answer is context, and what is context? It is what AI needs to do great work. Now, think about this.
LLMs can, of course, write a sales email, but without knowing who the buyer is, what the last 10 conversations were, who the competitor is, and what's the best way to handle objections for that competitor, that email will be generic. LLMs know everything that's in the public web, not what's in the sales pipeline, which, by the way, changes every minute. Agentic customer platform, we have that context. We know the relationship of the users, their roles, the tasks, and the logic that is needed to take action on their behalf. And the richer that context, the better the outcomes that AI can drive. And nobody has richer customer context than HubSpot. And that is because we bring together data, user, and domain context. So let's talk about this. Now, data is what AI needs to do work, not just guess about the work to be done.
HubSpot has 19 years, 270,000 customers' worth of those touch points. Every campaign launched, every email sent, every deal closed, every CPQ transaction across the entire customer journey. That is the data that AI needs in order to do great work. We also need the user context so AI knows who is asking and what permissions they can take based on the role. HubSpot maintains that state for the user, their roles, their preferences, their teams, the actions they can take. All of that is within HubSpot. We are built to maintain state across thousands of users in a business context, something LLMs are not designed to do. Then, we have the domain context. We have the deep knowledge of SMBs and what drives growth within SMBs, what prospecting, good prospecting looks like, which campaigns convert, and what the good handoffs look like.
And we are building all of that into HubSpot. So you put these three pieces together: data, user, and domain context. That is HubSpot's unique advantage: rich customer context that powers every action customers take in order to drive growth. That's pretty exciting. But just don't take our word for it. Let's actually listen. And before we do that, let's actually talk about how this shows up within the market. Now, the question becomes, how does this context show as we compete? And how do we win when we look at up-market incumbents? Within up-market, we win because we are easy to use, fast time to value, and we deliver that unified context. And we lower TCO. Now, here's the thing. An example is Lendlease, who is a global real estate and construction company.
Now, here, we consolidated 20 different systems, including a legacy CRM, and we were able to reduce the total cost of ownership by 80%, and we completed this whole transformative project in under seven months, and take Liquidity Services. They are a global reverse supply chain company. They consolidated eight different systems, including a legacy CRM, and reduced their costs by 50%, and they were able to drive the pace of customer conversion up. When companies look to consolidate systems, lower costs, and increase growth, they choose HubSpot. Let's also talk about down-market incumbents. Now, smaller customers, they do tend to get started with simpler systems to manage their tasks or projects or to move away from spreadsheets, but as they grow, their needs grow as well.
They choose HubSpot because of the depth of functionality, the breadth across marketing, sales, service, and ops, and the unified context that we provide. For example, Viitag, they started with a simple CRM, and they moved into HubSpot in order to get that better visibility and to get more sophisticated sales reporting, sales forecasting functionality. And Catalyit, they consolidated across 100-plus automations to get a clear single view of the customer. So the breadth and depth and the unified context matters. And then, we also have AI startups. There are lots of them here. Now, point solutions and point agents, they are good at something very specific, like support tickets. But the minute you ask a question about sales or marketing, they don't have the context for it. HubSpot provides that full context. Our agents, whether they are support, prospecting, sales, Content Agent, doesn't matter.
They provide and pull from that same unified context. And here's the thing. Over the past few quarters, we've seen a whole consolidation trend. And when we talk to our customers, they don't want Point Agents or Point Applications. They want one connected, consolidated platform. And that is exactly what we are building. And the other thing is we look at the AI startups as our innovation lab. If there is a use case that works and is gaining traction, then we build it immediately into our platform. That speed, combined with the unified customer context, that is our advantage. So let's actually hear from a couple of customers on why they choose HubSpot.
If you're looking for a fast, easy-to-deploy CRM that everybody's going to be able to get on with and instantly gain value in a very short period of time, HubSpot is absolutely the solution for running your business. HubSpot has really become our customer operating system. HubSpot's native AI tools and its openness to integrate with others really provides the AI or customer context that no other platform provides right now. There's a couple of things that make Tessa untouchable. I don't have time to go to eight different systems to get five pieces of information. Part of efficiency is having everything where it needs to be. I mean, the feedback I'm getting is that their job is 100 times better because of Tessa.
Having Breeze Intelligence have access to all of our customer context, all of the interactions, it gives us the foundation we need to grow our AI capabilities over time. What I'm really excited about with Breeze Intelligence and Breeze AI is being able to leverage the AI that's pre-built into HubSpot to surface some of that structured data so it makes for meaningful content and context when we engage with customers.
Yeah, I hope that makes it super clear why customers are buying HubSpot. The next question is, what are they buying? And how are we evolving our pricing and monetization strategy to drive durable growth? Okay. Now, our pricing philosophy has stayed consistent. It's always value first before we monetize. We really focus on use cases that drive repeatable value for customers because when they grow, we grow.
And we are strategically patient because this approach of value before monetization has worked for us. But how we drive pricing is changing. Our pricing model is hybrid. It has hubs, seats, and credits. And we want to make it super easy for our customers to try, buy, and grow with HubSpot. Now, all of you know how subscription works. We've had that for many years. So let's actually double down and look at how we are evolving seats as well as credits to drive durable growth. And let's start with persona seats. Persona seats are simple. They provide access to very specific hubs like Sales Hub and Service Hub. So if you have a specialized role like a sales rep role, that's the workspace that you will go to and take actions to drive growth.
Now, when companies grow, they also buy other hubs, and they continue to add more seats, and when they grow, we grow. This has been a consistent driver for HubSpot's growth, and in the last couple of quarters, we have seen very strong seat upgrade motion for both Sales Hub and Service Hub. Next, let's actually talk about Core Seat. This is the next big driver of growth for HubSpot. When we launched Core Seats last year, the goal was simple. It was to unlock platform value. Now, customers buy HubSpot's Core Seat to get admin capabilities, to create new customer objects, powerful workflows, and it worked. Since the launch last March 2024, we've grown this to over 100 million in ARR with a strong seat upgrade motion, but now, we are making this Core Seat game-changing.
We are now, at this INBOUND, unlocking the power of the Core Seat with AI, data, and platform value. Breeze Assistant, projects, smart starts, the company data, and contact enrichment data for 200 million companies all now in this Core Seat. It is massive value for our customers. And here's the thing. That's why we've unbundled the Smart CRM, which means our customers can just buy the Core Seat and then add use cases as they go along. Easy to buy and easy to use. Here's the big shift. By delivering the AI capabilities and the data foundation that is needed to get work done, we're now making the Core Seat essential for every go-to-market employee. Super exciting. Okay. With that, let us talk about credits, which is our newest lever for growth.
Now, as HubSpot starts doing more work, we want to provide a flexible way in which our customers can pay that is not tied to the seats. And the usage-based pricing is the universal mechanism to do that. We launched it at INBOUND last year, and now it's expanding across the platform. And it includes things like AI agent actions, Data Hub syncs. And tomorrow, it's going to include anything that is usage-based within the entire platform. It's pretty simple. We have credits that's included within the tiers for every customer, and they can top it off with $10 or $100 increments. The key thing is this. It scales with AI and data value. Simple, flexible, and powerful. Now, you put this all together, you have persona seats, core seats, and credits.
They all grow with AI, data, and platform value for customers, and therefore, they are a durable engine for our growth. Pretty exciting. Now, let's actually shift gears and talk about something that is top of mind for everybody, which is the big shifts that are happening within marketing and the opportunity that's ahead. I hope many of you caught the spotlight, and you can see why we are so excited about the opportunity that's ahead. Now, here's a super quick recap from the spotlight. Search is getting disrupted. AI overviews are providing answers, and that means less traffic to the website. But at the same time, AI is also doing something completely different. It allows you to capture intent signals. It allows you to match it to the customer and therefore drive conversion. So marketing needs a new playbook. And that is exactly what we launched today.
It's a new playbook for growth within the AI era. We call it the loop, and customers can execute the loop with our products. Our vision for marketing is very clear. It is humans and AI in the loop. We express our authenticity and taste, and AI tailors it to our customers. We amplify and reach our customers, and AI evolves it and optimizes it. That's the future of marketing. It's also data-driven. All of that segmentation and personalization that you saw there, that is driven by data. Data is the foundation of everything that you need to do in the modern playbook, and it's AI-powered. Agents features that are built throughout the marketing workflow, and this is exactly what we are building. Today, we launched more than a dozen new features and updates to bring this loop to life.
If you look at this for Express, that's the newly transformed Data Hub. That's where you build your ideal customer profile, and within the marketing studio, you create the content to be able to get to your customers within Tailor, the Data Agent, a personalization agent, and AI-powered segmentation. All of that enables you to match the intent of your customer to the information that they need to see. Within Amplify, we launched new AEO tools to get LLM visibility for customers, and we have our Content Hub that helps diversify the content across multiple channels, and within Evolve, we have a number of optimization tools that are powerful and provide the abilities for customers to grow. All of this powers our new growth playbook. This is how we are reimagining Data Hub, Marketing Hub, and Content Hub to power this new growth playbook.
We are just getting started with this innovation. It's super exciting. Look, when Brian and Dharmesh started this whole movement with inbound, it was the playbook. It was a set of products that supported the playbook and a whole ecosystem behind it. And we are using a very similar approach with Loop. We have a clear playbook. We have a powerful platform, and we have a thriving ecosystem that is scaling it. This is such a massive opportunity for HubSpot, one that we are positioned to lead and win. Really exciting. Okay. Let's actually talk about upmarket and downmarket and how we are scaling both. Within upmarket, our strategy is clear. We want to deliver value to customers and therefore grow the value of customers.
And the way we are doing this is by focusing on three things: build the right product, grow the right ecosystem, and execute the right sales motion. Let's talk about our product. Going upmarket has been a multi-year focus for us. We want to build powerful tools that are super easy to use. And we have had a consistent set of innovation: opening up the markets for sensitive data, journey orchestration, multi-account management, new global data centers. All of this innovation proves that we scale with businesses. And then let's talk about our partner ecosystem. We have a very strong partner ecosystem. Just this year alone, we added 750 new solution partners. And we are focused on working with growth partners to serve upmarket customers and build with AI. Our partner activation, sourcing, co-selling, they're all up this year. And we have a solid app ecosystem, more than 1,900 integrations.
The average number of active integrations that customers have grown from 10 to 14 over the last five years. That shows that upmarket customers are using us as a platform. Together with this ecosystem, we are selling to more upmarket customers, delivering more value, and scaling faster there. Then we have the right sales motion. We have upskilled our sales team to sell complex deals with confidence upmarket. Large deals are up 36% year over year. 90% of the new deals for upmarket customers are multi-hub deals. That shows that the momentum is in deal size, but also product depth. Okay. We have a proven playbook upmarket. What about downmarket? Our strategy here is to drive volume, deliver value, and grow as customers grow. We do that by having a freemium product that adds a ton of value.
When businesses get started, they start on our freemium product, and today, we've added more AEO tools and AI tools into that free product, like our AEO Grader. That adds value, and from there, they convert into paid customers. We reduced the price of seats, we removed the seat minimums, and we have made onboarding AI-first, so free-to-starter conversion is now at an all-time high, 50% growth year over year in free-to-starter conversion, and then when they get into a starter or pro, we deliver compelling value. We become that customer operating system that customers depend on, and when they grow, we grow, so it's exciting. We have a solid distribution engine that is scaling upmarket as well as downmarket and providing durable long-term levers for us to grow. Now, as we grow, we also think about how HubSpot is scaling, especially with AI.
And this is where we recognized very early on that AI is not just transformative for our products, but also for the way in which we operate as a company. In order for us to lead, we got to move at AI speed. And we have to get the confidence into our customer and partner ecosystem. So we're transforming how we work internally. And this is how the transformation looks like in action. We're building faster. Our engineers are using Cursor, Claude Code, MCP servers, and their productivity is up. 92% of code changes that we did this year is with AI assistance. And productivity is up 46% more code ship per engineer this year compared to last year. And we are serving customers better. 75% of our web chat and more than 50% of our support tickets are now resolved by AI.
And AI is also scheduling meetings for our sales teams and increasing the win rate. And overall, at HubSpot, we're now becoming AI-first. Every HubSpotter is using AI in their everyday work. And we've dedicated more than 20 days to drive AI learning and innovation into the company. This is how we lead and win with AI. So let me close out with why I am excited about HubSpot's opportunity ahead. We're transforming our platform to be an AI-powered customer platform. We have rich customer context. That is our platform advantage. We have durable growth levers. We're scaling upmarket, downmarket, and we have hybrid monetization that scales with customer value. We are transforming ourselves to operate at AI speed. And we have the people, the passion, and the product in order to lead and win in this new era of work.
With that, it is my absolute pleasure to welcome Kate Bueker, our CFO, to come and share how all of this translates into durable, efficient growth. Thank you so much for your support, and I look forward to answering questions later.
Thank you. All right. Hello, everyone. It is nice to see you in person again here at INBOUND. I'm excited to share how our strategy translates into durable and efficient growth for HubSpot. In terms of our agenda, I'm going to start with a reminder of our key financial priorities and talk about how we have delivered against them over the last year. Then I'll show how our AI-powered customer platform enables growth for HubSpot today and also into the future. And finally, I'll talk about how we're balancing growth and profitability. I want to start by grounding us on our core financial principles.
Here at HubSpot, we are determined to build a big, profitable, and enduring business. As a result, our top priority is to deliver the fastest and most durable growth we can over the medium and long term. At the same time, we are building a healthy business that can scale over time and continue to deliver leverage. We have a consistent track record of delivering against these financial priorities. Since 2021, our revenue has grown at a compound annual growth rate of 24%. Our full year 2025 revenue guidance is for $3.1 billion, which represents a growth of 17% in constant currency. This revenue growth has been enabled over the last four years. We have grown our customer base at a compound annual growth rate of 22%. We had nearly 270,000 paying customers at the end of Q2, representing growth of 18% year over year.
Over this period, we have also delivered consistent profitability growth. We have grown our non-GAAP operating profit at a compound annual growth rate of 48% and expect to deliver $570 million of operating profit in 2025. This represents almost 10 points of operating margin expansion since 2021. Our growth has also translated into strong free cash flow. Since 2021, we have delivered $1.4 billion of free cash flow, and we have $1.9 billion of cash on the balance sheet at the end of Q2. We've used our free cash flow to retire our convertible notes, to acquire technology and teams to accelerate our product roadmap, and to opportunistically return capital to shareholders. Since our earnings in early August, we have bought back an additional $175 million of shares, bringing our total share repurchase to $425 million out of our $500 million authorization.
Our strong balance sheet gives us the financial flexibility to continue to invest in growth, both organic and through acquisition, and to opportunistically return capital to shareholders. Next, I want to talk about our levers for current and future growth of our business. I think about the opportunity in two broad buckets. We have a set of current levers that have been delivering results for our business for the last few years. We also have an emerging set of levers enabled by AI and new monetization that will support our growth into the future. Let's start with these current levers, which should be familiar to many of you. We take a platform-oriented approach to address our market opportunity, and we believe that our platform is the key driver of our upmarket and downmarket momentum, as well as our strong customer retention.
Our upmarket momentum continues to accelerate, and we're seeing more large deals and more platform consolidation. The share of our new Pro+ ARR that is multi-hub has grown steadily over the last four years. In the first half of 2025, nearly 80% of customer ARR came from customers who landed with multiple products, and almost half is from customers who bought three or more HubSpot products. New and existing customers are consolidating their go-to-market technology stack on HubSpot. Our strong platform traction with new customers, as well as our continued success in cross-sell, has led to an increasing adoption of our platform across our install base. The share of our install base ARR that has on a single hub is now just 12%.
More than 60% of Pro+ ARR is on three or more hubs, and customer ARR on five or more hubs has grown from zero to 28% in Q2. This is resulting in more large deals, which we define internally as more than $3,000 a month in recurring revenue. A growing portion of our Pro+ install base is large deals. About half of our Pro+ install base is in that 3K-plus bucket by the end of Q2, and 3K-plus MRR is growing quickly, 30% in the first half of 2025. 5K-plus MRR is growing even faster at 37%. Turning to our downmarket segment, our starter suite continues to deliver a compelling value proposition at the low end of our platform and fuels strong and consistent customer growth. Starter customers represent about half of our total customers, which is up 10 points over the last four years.
Starter customers total 130,000 at the end of Q2, representing a nice upsell opportunity for HubSpot and an attractive ecosystem opportunity for third-party developers. Turning to retention, customer dollar retention continues to be strong and stable, improving slightly year over year to 88%. As I've said in the past, customer dollar retention is the metric that we believe is most closely aligned to the value that our customers see in our product. Platform consolidation, along with our focus on activation and customer usage, continues to drive really nice and strong customer dollar retention. Over the past year, we have also seen an improvement in net revenue retention from 101.5% to just below 103%. While the customer dollar retention is the foundation, the improvement in net revenue retention over this period is a result of increased seat upgrade rates following our pricing change last year.
This has been somewhat offset by continued headwinds in other upgrade motions, like contact tiers. As I shared in our earnings last quarter, we expect net revenue retention to be flat in Q3 and then up two to three points in Q4 with some help from the seat migration renewals. This platform-driven momentum has translated into net new ARR growth that has steadily improved over the last two years and has grown faster than revenue since the second half of 2024. When we report revenue growth every quarter, we typically share ARR metrics on a half-year basis in our annual proxy statement. And our intention is to continue with that practice. That said, Yamini started her presentation by saying that clarity is important in times of change and uncertainty. And so I wanted to be very clear about the recent business performance.
As you can see, net new ARR was challenged beginning in 2022 as customers focused on optimizing their software spend and hit a low point in the first half of 2023. Over the last two years, we have seen building momentum and a nice step up in net new ARR growth in the back half of 2024. Since then, net new ARR growth has outpaced revenue growth. We are executing well, and our growth is beginning to accelerate. So our current levers are working. We are also seeing a number of new levers for growth emerge. Let's take a look. We're in the midst of a pricing change, which is a multi-year tailwind. And we are adding significant AI and platform value that we will monetize via our Core Seat and through Credits. These are new and increasingly important levers for our business.
Now, before we look at the new way to look at the business, for the sake of completeness, I will share the old way that we have broken down our growth one last time, which is through the lens of our individual hubs. Marketing Hub ARR is north of $1.5 billion, growing 11% year over year as of the end of Q2. Sales Hub is growing 24% year over year, and Service Hub growth sustained in the high 20s. And finally, Content Hub and our newly renamed and relaunched Data Hub are growing in the mid-50s and high 20s, respectively. Overall, we're happy with the performance of our individual hubs. That said, and we talked about this last year, the individual hub lens does not reflect how we think about the value that we're delivering to our customers and how they are leveraging our platform.
Let me share a few common buying patterns to show you what I mean. Our new marketing playbook, it's not just about Marketing Hub. It's about marketing plus data plus Content Hub, and that buying motion is growing fast. Customers are not buying a sales or service product in isolation. They are looking for solutions to connect sales and service teams with data and now CPQ to eliminate the friction of customer handoffs. And the full power of our AI-first customer platform is resonating. Our mix of five-plus hub customers is up five points versus last year. It's these multi-hub buying patterns along with core platform seats and credits that will drive our growth in the future. Speaking of seats, I want to give you a progress update on the impact of the pricing change. Now, this is a slide that I shared with you at last year's Analyst Day.
When we talked about the pricing change last year, I shared that we expected it to have two large impacts on our business. First, we would add more customers, and then the new model would result in better net retention. Let's see how we're doing against the first of those, which is adding more customers. Our pricing model eliminated the seat minimums, making it easier to get started with HubSpot. Post-launch, we have seen the average number of Pro+ portals added every quarter increase by 16%. In addition, customers under the new model are healthier because they buy only what they need upfront, and we have seen consistently better seat assignment rates. Second, we said that the new model would result in better retention, and that is exactly what we are seeing.
The share of customers who upgrade their sales or service seat is about three points higher under the new model. We also introduced the Core Seat as a way to monetize the value of our Smart CRM, and this has resulted in higher paid seat ratios. Approximately 20% of users required a paid seat in our legacy model. Today, about 60% of users who have purchased under the new model take a paid seat. Finally, the migration of our existing customers is well underway. We expect that about 90% of our ARR will be through their technical migration this year, and about half of our ARR will be through their first renewal by the end of 2025.
When we step back and look at the overall impact of the pricing change, we continue to expect that it is a two-point tailwind to net revenue retention in 2025 and a continued benefit in 2026. Let's dive a bit deeper on the Core Seat. As I just shared, we introduced the Core Seat in March of 2024 as a way to monetize our Smart CRM. In just 18 months, Core Seat ARR has surpassed $100 million and is growing quickly, validating our thesis that there was a ton of value in our Smart CRM platform. As a refresher, the Core Seat provides CRM edit access and admin capabilities, allowing our customers to customize and extend the CRM to fit their business. The Core Seat is typically used by marketing and sales operations teams and go-to-market leadership. With today's product announcements, the Core Seat becomes significantly more powerful and valuable.
We've added meaningful AI and data capabilities, including Breeze Assistant, Note Taker, Smart Data Enrichment, and Project Management capabilities. Everything that we're doing to expand the value of the Core Seat should result in more people having a seat on HubSpot. Our goal is to drive adoption of the Core Seat for every front office employee, significantly expanding our opportunity over time. The other way we're monetizing AI is through Credits, a unified way to purchase any consumption-based HubSpot product. We introduced Credits at INBOUND last year, and we made the Customer Agent available for new purchases starting in June. Today, we added Prospecting Agent, buyer intent, and data syncs to the credit system. We'll use Credits when value scales with work rather than with people. And finally, all portals will have some Credits included to make it easy for customers to experiment with agents and other features.
Customer Agent was our first direct AI monetization via this credits model. While we're still incredibly early, Customer Agent activation and usage continues to scale. As of the end of August, we had more than 5,000 customers with resolution rates now over 65% and more than 1 million tickets resolved by Customer Agent. Okay, before we finish this section, I wanted to talk about how seats and credits monetization live together. In particular, I wanted to address the concern that the consumption models will negatively impact seats-based revenue over time. We believe that our hybrid model, this combination of seats and credits, isn't just sustainable, it's what is best for our customers and what is best for HubSpot. While we're still in early days, I thought it would be helpful to share some early trends through a couple of customer examples.
The first one is an upmarket AI services company that adopted Customer Agent in December. They are resolving about 35% of their total support volume via Customer Agent, and their goal was to significantly scale up their customer support without adding any headcount. They've retained all 18 of their Service Seats, and they've grown their overall MRR with HubSpot by 80%. The second is a small business that gives higher education financing to companies. They adopted Customer Agent in March, and they wanted to use Customer Agent to resolve their WhatsApp ticket flow. Customer Agent is successfully resolving about 25% of their total inbound support volume, and this allowed them to go from five Service Seats to four Service Seats. The efficiency gains allowed them to grow their business, and they've since added more Core Seats and more Credits.
Despite the reduction in their service seats, their total MRR with HubSpot is up about 10%. I want to acknowledge that these are just two examples and that we are still very early, but we have not seen any abnormal downgrade behavior from customers who have adopted Customer Agent. Okay, I want to wrap up this section by summarizing our levers for growth. Our current levers are delivering results. Our platform is working upmarket and downmarket, and customers are growing and retaining on HubSpot, and our emerging levers are gaining traction. We're in the midst of a pricing change that is a tailwind to growth, as we expected, and finally, we're delivering a ton of value in AI and data, and we will monetize this via Core seats and Credits over time. Okay, now I want to shift gears and talk about how we are balancing growth and profitability.
We're building a healthy and balanced business, and we have delivered consistent leverage throughout the history of HubSpot. Since 2021, we delivered an average of 240 basis points per year of operating margin expansion. We expect full year 2025 margin of 18.5%. Now, you know we have medium and long-term targets for operating profitability: 20%-22% in 2027 and 25% long-term. We are confident that we're on the right path to hit these targets and that we can continue to deliver strong innovation and leverage. Okay, before I dive into the shape of the P&L, I want to talk about the role of AI in helping to create incremental financial leverage for HubSpot. I think about the role of AI in two buckets.
There's a set of functions or areas where we're going to continue to invest, like engineering and sales, where AI is going to drive faster innovation and higher productivity. Yamini talked about the meaningful increases we're seeing in code generation and in productivity and win rates on the sales side. We're going to continue to lean in here. In other places, like support, marketing, and G&A, we will use AI to drive efficiency. Support is a great example. We have been using AI to resolve customer inquiries for a couple of years now in support, and our agent handled about half of our customer tickets in the first half of 2025. As a result, we have been able to hold support headcount flat for over two years.
In marketing, we are heavy users of AI for content generation and personalization, and we have seen both efficiency gains and also higher conversion rates as a result of AI, and finally, we have started to deploy AI tools across our employee lifecycle that we believe will allow us to continue to scale our best-in-class G&A efficiency. The combination of these approaches will allow us to both grow and deliver leverage. Let's talk about how this translates into our reported P&L, beginning with gross margins. As we shared last quarter, our gross margins this year are expected to be down slightly as a result of data center investments and higher R&D capitalization. Looking ahead to 2026, we expect some incremental COGS pressure from new product investments in unstructured data and from scaling AI and commerce. Turning to OpEx.
In R&D, we will continue to drive a fast pace of innovation, but AI productivity gains will allow us to grow R&D spend at or below revenue growth. For sales and marketing, AI tooling and improved rep productivity upmarket, along with better conversion efficiency at the low end, will be key drivers of S&M leverage, and finally, we will realize modest additional gains in G&A by leaning into AI and automation. As I've said in the past, we may see a bit more or less leverage in any given year depending on the opportunities we see, but we will stay on track to hit our interim and long-term margin targets. Let me close with my three key takeaways from today. HubSpot is the leading customer platform for scaling companies. Our business is accelerating, and we have a big opportunity to continue to drive momentum.
AI creates big new opportunities for HubSpot to deliver work and to extend the value of our platform to every front office employee. And finally, we have delivered and expect to continue to deliver consistent leverage enabled by internal AI adoption. Thank you for your time today and for your continued support of HubSpot.
Great job, Yamini and Kate.
Thank you.
Yeah. So we're going to have some mic runners. I can see hands already going up here in lots of hands. Might need to get an extra night at the hotel. So yeah, we're going to be running around with mics here. I think we're going to have about 35 minutes just given the length of the presentations that we let Dharmesh get to where he needs to get to while chasing down there. But yeah, why don't we go ahead and get started? Alex.
Guys, congratulations.
That was an excellent presentation, even more excellent in San Francisco.
Yes, I know.
Self-serving.
So if you think about the message and the metrics that you guys presented, I think it's pretty clear you're helping customers with the real pain point, particularly with the playbook. And it sounds like the future is a hybrid monetization model. You're seeing an AI product, customer agent specifically, already have an impact, and you're saying that new ARR growth is re-accelerating. When you think about the concept about durability, what are you trying to tell us about, A, how much of that acceleration is tied to the consumption monetization that you're already seeing? And what is the confidence level that we can sustain kind of that good, I'm not going to put a number on it, but good net new ARR growth cadence that is ahead of revenue?
And kind of when should we see that revenue growth come closer to that net new ARR growth?
That's such a typical Alex Zukin question. 18 questions in one. I'll take it away.
14 questions in there.
Okay, I'm going to start with the crux of what you're asking. How confident are you in the durability of the growth and what is going to drive that? And then Kate will answer any guidance questions. We feel really confident. We feel really confident in terms of the growth levers and the ingredients that we have in order to drive durable, sustainable growth, not just in the next two years, but in the next five to ten years. And I think Kate did a really nice job of laying this out, but I'll provide my perspective on it. We are in the middle of incredible change within this industry.
And I think the biggest change that's happening is within the marketing landscape. Change with search also is like the biggest opportunity with AI. And that is exactly what we are focused on. What we launched today with the playbook is exactly what our customers have been wanting. I've talked to so many CMOs in the last few months. They want a playbook. They want a step-by-step process to figure out how to drive growth within the marketing era. And that is exactly what we delivered. And now we have the platform and the entire products to drive that. That is a multi-hub opportunity. It is not a single-hub opportunity. Because in order for you to execute the Loop, you need a Marketing Hub. You need to start with the data foundation with Data Hub, and you need the Content Hub. So we think that is a multi-hub opportunity.
I think it's the same thing with sales. Sales is going through a fairly big transformation as well, where prospecting is all about data and how you reach customers at the moment that they want to buy. And that is a multi-hub opportunity. So both of those are really big. I talked about Core Seats. This is super interesting for us because one of the key things that we are doing, and I want to maybe emphasize this, is our AI monetization is not just Credits monetization. It is actually going to be part of the embedded Hubs. It's going to be part of the Core Seat. What we did this year and what we did just now is by unbundling the Core Seat and putting in AI value, data value, and platform value, we've now made it important for every go-to-market employee to use it.
That's a pretty big, massive opportunity, and we're going to take full advantage of that opportunity. And then it is, of course, credits where, as we deliver more work, then that provides a whole opportunity. So I would say marketing and multi-hub, sales and multi-hub with CPQ involved, and core seats as well as credits. All of that means we deliver more value to customers. That is what we get excited about, and that's what you started your question with. We want to drive more value, and when we drive more value, we feel much more confident in the durable growth levers. Plus, we have the current levers that are kind of working. And so pretty excited.
Thanks. I just wanted to double-click on the core seat. Sorry. There you are. A nd I guess. Thank you.
So I just wanted to double-click on the Core Seat opportunity. That $100 million number is very impressive. But how do we think about the size of the employee base at your customers? And what is maybe a long-term penetration rate there that is maybe a base case scenario for us to think about?
Yeah. I think we really like the whole Core Seat opportunity. And just the simplest way to think about it is that we used to embed Smart CRM within every Hub way back. Then last year, we separated out the Core Seat because there was enough platform value in it. And customers buy it because of admin capabilities, creating a new customer object. So anybody that needed to edit the platform, they bought the Core Seat. And that is now a $100 million business for us. Plus, it's a strong seat upgrade motion.
At earnings, I mentioned that we're seeing an upgrade motion of 25% for that core seat alone. And what we've done today is we've unlocked it with AI data as well as platform. And you remember Clearbit, we acquired this company, all of that company data and contact enrichment data now within that core seat. So we're super excited about it. The way to think about it is that today, admins, RevOps, FP&A folks, those are the ones that need it. And tomorrow, as Kate laid it out, every go-to-market employee needs will have it. And typically, in a scaling company, a mid-market company, there's between 30%-40% of the whole employee base that is in go-to-market. So you can think about single digit to 30%-40% of the employee base. Now, this is not going to happen in one year.
We're not talking about all of this happening in one year. But I think we're setting the foundation for a durable lever that pays off in the next few.
And just to be clear, Brian, it's really about where we were was purely persona seats. So paid seats were for sales reps and customer support reps. And now where we are, you have admin, so operations teams and sales leadership buying the core seat. What Yamini is describing as the future is that sort of 30% to 40% penetration of every front office employee.
Thank you, Mark Murphy with JPMorgan. I know you're packing the town in pretty well when I try to get a Waymo in the morning. And it's going to be, instead of a two-minute wait, it's a 17-minute wait.
Sorry, Mark.
No, we love having you here. Made it so easy.
I actually want to ask you about AI-powered email. The 82% increase in conversion rate, I mean, I think 8% would be a lot. Can you explain how individualized are you making these outbound emails? I mean, that's a tremendous change on a major channel. And how do you control the cost of that? I mean, if someone's sending, if we send a million emails and each one is being scrubbed for what did the customer do and what did they buy and what are they interested in, it sounds a little LLM intensive. And I'm just wondering if you can help us understand that dynamic, how you manage that cost.
I'll start with the front end of the question, and then you can talk about the second part. And Mark, thanks a lot. We like being here in San Francisco as well.
So on the email personalization, a lot has been said about just search disruption and what AI is doing to search. But the flip side of it is even more interesting that it actually drives better conversion. And specifically on that email conversion, what we are doing is now we know how to capture the intent signal of each buyer. Then we are using basically RAG to take that buyer and map them to the specific information that they need to get. So for example, it might be an academy class that they want to attend or a very specific demo that they need to see. So we're using that matching between the intent signal and LLMs to be able to send much better personalization. We talked about it last year at INBOUND, and we've improved that email personalization at scale.
Now, that's working for us, but today you heard that it's also working for our customers. That's exactly what Kelly Services said in terms of driving better conversions as well as engagement. That's how we are doing it, and I think that's the new playbook. The playbook is not just waiting for search traffic to return. It's actually to do better personalization and segmentation, and then maybe you can talk about the COGS impact.
Yes, so Mark, we haven't seen material COGS impact from the personalization of emails. We have a, well, we call them the COGS squad. It is a cross-functional team across, well, some of them are in the back. You could raise your hand, COGS squad. Across finance, engineering, our product operations teams to really deep down understand sort of the drivers of the AI-oriented costs as well as costs across our total platform.
Where we have been seeing some pressure on margins, and I talked about this in my talk track, is really we opened a couple or three new data centers in Q1. We have been investing in R&D outpaced with revenue over the last three to four years. And that's showing up in our R&D capitalization in COGS. That's the sort of pressure we're seeing. Next year, there will be some scaling of AI. I think the good news here is that we have revenue monetization associated with the things that are going to drive scaled-up costs for AI. And we're watching it closely and continuing to monitor.
Just one quick comment on the kind of inference cost. We get very excited, including me, about the newer models and deep reasoning and deep research and all these kind of high-end features that are in Opus and GPT-5 Pro.
Email is a good example. The baseline capability that we need for a vast majority of go-to-market use cases are handled by something like a GPT-3.5 or a 4.0, right? So we don't need the frontier model capability for a vast majority of these use cases. And the cost of GPT-4.0 has gone down 250x from the time it was released, right? So that slope still goes down. Yes, the advanced models are getting more expensive, but most of the models we need for most of the work that we do, including email, is not those frontier models.
Brad Sills from Bank of America. Thanks so much for hosting another great event. Great to see you all. I wanted to ask a question about some of those early adopter customers, that 5,000 customer base that's on Customer Agent. What did it take to get those customers live?
Was there a certain amount of effort involved at the data layer with data prep, or was it more plug and play? HubSpot's known for simplicity and plug and play here. Or was it simply just, these are features, and these customers are turning on these features in a seamless manner? I would imagine the answer is different depending on the market segment that you're going after, but would love to get some color on what it took to get those customers live. Thank you.
Yeah. Good to see you, Brad. And so I think it's a spectrum in terms of driving adoption. We have customers where they have really good knowledge base to begin with. And the knowledge base, plus the history of support tickets that they have already resolved, is like a super easy starting point.
We've had customers where they've plugged the knowledge base, they've plugged the history of support tickets, and in less than a week, they begin to see clear AI resolution of support tickets. The setup sometimes is like 30 minutes. It is super easy. Now, that's the best-case scenario when you have all of the documentation that is needed. On the flip side, we've had customers where they don't have knowledge base, they don't have a support history with us, and they're beginning to kind of build that type of history where it takes a little while for them to start seeing the impact. We've seen some customers that start with 20%-30% in terms of AI resolution rate. This is why we mentioned this last time. We have a knowledge-based agent.
That knowledge-based agent basically recognizes when you don't have a good knowledge base, and it starts writing articles that can then go into the knowledge base with validation from users. Once you do that, you begin to see the resolution rates go up. It is a spectrum of both of these. When you have documentation, it starts working. When you don't have it, then you begin to kind of create that history in order to get there. Overall, that use case has found product-market fit. 5,000 is just the beginning because we think that a customer agent is like the website of the old days, right? Where when you were in the internet era, you needed a website, and you needed to get your digital presence started.
You now need a customer agent, not just to answer support tickets, but to be able to answer any question about your product, your pricing, your marketing, the next event that is happening. You need that. And so we think it's going to be as ubiquitous as a website, and we're pretty excited about it.
Hey, Raimo Lenschow from Barclays. On that note, the one big debate we have in the industry at the moment is about how GenAI and kind of the SaaS model kind of will work together in the future. And there's the discussion about the death of SaaS, etc.
Listening to you now and making a new focus on data and the Data Hub and the context as well, it kind of feels like there is the answer already that there's so much kind of knowledge in your system that you actually kind of you are needed. And kind of that's actually the way you thrive in the future. Can you speak to that a little bit? And if I'm on the right path here, thank you.
Just a couple of quick notes. One is SaaS effectively was a kind of deployment and business model kind of transformation, not that big of a technology transformation. So I think what endures is software. Software is a high margin, high leverage. You can put investment and solve a bunch of customer problems. So I think we have the largest opportunity as an industry in software than we've ever had before.
The business models will change. I think SaaS in its purest form is unlikely to remain the way it is right now. That's why we see the hybrid pricing models. But I'm just super bullish about the opportunity this creates because now we can solve problems with software that we were never able to do before. Before we built tools for humans to use, now we can actually do the work, so the value that software is going to produce over the coming decades is orders of magnitude, I think, higher. The TAMs are just going to be bigger, and we're just starting to see it's the early innings of that game, but I don't think software is dead. SaaS as a pure business model might transform over time, but software as a way to make money and put capital to work, I think, is going to be amazing.
And I have to say, I have a very specific point of view, which is that AI and SaaS are complementary. I'm following all this raging debate about one eating the other. I think it's complementary because if you think about LLMs, they provide insights. And HubSpot, we provide the customer context to drive insights about growth. That is not there in an LLM. And then if you look at what an LLM is great at doing, taking action on behalf of one user, we are really good at maintaining state across thousands of users, their permissions, their roles, and what actions that they can take in a business context. Those two are complementary. And all of the interactions now get started on LLMs. People are beginning to start with asking questions.
And all of the actions and, more importantly, the logic that is necessary to take the action in a business context, that belongs in HubSpot. So I think the combination is powerful. And our viewpoint is that we are going to deliver more value to customers. And look forward to seeing this raging debate continue. But we have a point of view.
Awesome. Someone has to represent the East Coast. I hear Boston's lovely this time of year.
Thank you, Samad. Yes.
All right. So this is either for Yamini or Dharmesh, either one, and maybe both. But if I think about a couple of the exciting things, there's Breeze Studio and the ability to build custom. And then also, I know Agent.ai is not part of the company, but you're building kind of this two million person developer curious or developer army out there.
If I think about that, the ecosystem, how should we think about that allowing you to push more up market to larger customers that were maybe beyond your traditional focus area because of what it'll allow them to build now? Related to that, do both of these make you more attractive future partners for new ISVs that you maybe hadn't worked with before and SIs, and I'll say global SIs, just to go ahead and push it forward a little bit? Will that pull you further up market potentially?
That's like an Alex Zukin question. You should start.
Yeah, I'll start. You can feel free to add commentary. Breeze Studio and the core kind of agent platform within HubSpot is built for HubSpot customers, HubSpot partners. That ecosystem is going to help us move up market. It solves that kind of higher-end agent problem.
So we have our featured agents built by professional developers, built by the HubSpot product team. And that's going to mirror the kind of evolution we've had within HubSpot for a long time. And we're very, very good at that. Think about Agent.ai as the other end of the spectrum. This is citizen builders solving simple tasks, simple use cases, either for themselves or their small teams. The reason we're excited about it, and we think we want to cover this as kind of the bimodal approach to HubSpot, is that we think there's going to be this entire new generation of agencies and software companies that are going to be made possible by AI. And they're going to need the tools to build. They're going to need a market.
We think we can create that ecosystem, which is the next-generation agentic ecosystem, alongside the Breeze Studio and the Breeze Marketplace that we have now, which is for our current core market. We're sort of serving both ends of the spectrum. Just for the record, Agent.ai is actually part of HubSpot, part of our innovation team, the HubSpot Next project. Yeah.
Great. Thank you so much. Elizabeth Porter from Morgan Stanley. Just going back to the idea of addressing more of the front office employee base, anyone that's involved in go-to-market. As you have some of these AI agents being built themselves, how are you able to use that to see what are people interested in, the kind of the workflows that you're building?
And when you combine that with the fact that AI is allowing you to build even faster, can this start to pinpoint new hub opportunities that you'd like to address?
Yeah. Absolutely. All of the above, right? I think one of the core things about HubSpot is that we start with solving for the customer. And that's been our mantra for a very, very long period of time. And we can very easily see the adoption patterns. And I would say the adoption patterns within mid-market scaling companies is quite broad. It's all the way from AI curious to leading with AI. And it is, while we are super excited about AI and the potential for AI, you've got to know that the customers are in this broad spectrum of, "I'm just getting started.
I want to be curious with AI," all the way to, "I'm leading with AI." And our approach has been three things, which is we want to embed AI and remove the friction of people even thinking about AI. Because believe it or not, our customers don't wake up thinking about AI. They wake up thinking about growth. They wake up thinking about how they can drive more leads and close more deals. And we want to embed AI into everything that they do so they can drive growth and focus on their job. And that's why we've had the strategy of not bolting on an AI solution, but building it as part of the customer platform. Then the second thing is all the agents. And Dharmesh just walked through it.
We have everything from featured agents to agents that you can build yourself, all the way up to citizen building agents. But the agents that we are building, we're looking at where customers need value, where there are manual tasks that need to be augmented. And prospecting is like a really big use case. We just launched it. It is now generally available. And we have a waitlist, which shows that doing account research, it's pretty tedious. I started my career as a salesperson, and I would spend hours reading like 10-Ks and 10-Qs. That is a use case. And so that's why we are building the prospecting agent. Data agent, like today, Karen did a wonderful job of showing this. You cannot just go to the web and say, "I want for millions of my records this particular data." Now you do.
And so we found that you can create smart properties. You can go out to the web. You can run LLM searches and then bring that information back. And that just came from working with our customers. And so I think we'll continue on this path of embedding rich AI capabilities into the core and then building agents where we see a set of tasks that need to be augmented, or we were not able to do that before. So super exciting, and it'll continue to go from there.
Siti Panigrahi from Mizuho. I love having INBOUND here. Thanks for bringing to San Francisco. Awesome. I guess we need a vote by the end of the night. I think we can vote. We know where the vote will be. I want to go back to the slide you talked about AI, how it's transforming your business and other business as well.
So, you talked about driving lever. So, help us understand how you are right now using AI. And are you still you talked about productivity improvement. Still, are you hiring? Are any changes to the headcount? And Kate, we see the long-term margin you still kept at 25%. Why shouldn't we think about margin going even higher?
Well, so you and I guess Alex Zukin's questioning method is getting adopted across the board here. But maybe I'll start there, which is taking us back to the first slide that I opened with, which is the balanced principles that we have in sort of our core financial principles, which is around balancing growth and profitability. And our top priority being building a big, profitable, enduring company. And we are going to continue to invest to drive top-line durable growth over the medium and long term.
So I think that's the foundational reason that we are not moving our margin targets. That said, we feel really confident that we can continue down the path towards our interim and long-term margin targets. I think AI is a big part of that. And I think you heard from Yamini and hopefully from me as well that there is real broad adoption of AI across the organization at HubSpot. And everywhere is getting more productive. And I think the difference in the two buckets that I outlined is where are we going to take advantage of that productivity and sort of double down and continue to invest? And that's areas like engineering where we're seeing a lot more code generation. And we can even increase the pace of innovation that we're seeing.
It's areas like sales where if rep productivity goes up, I want more reps, not less reps doing things. Then there's the other bucket, which is where are we just going to focus on driving efficiency and support. First use case with customer agent, first use case for us internally. We've been leveraging AI for support for a bunch of years and have steadily gotten leverage. We haven't added a support rep in two years. So there'll be more and more places across our organization where we'll be able to drive efficiency because we're leveraging AI, and we won't have to add headcount. So I do think you'll see over time our headcount growth will be lower than it has been historically.
We're not stopping hiring in R&D and sales anytime soon. I think that's the point that you made. I completely agree with that.
We're just going to continue there. And there are other places to get leverage.
Tell me to go ahead. Hey, sorry, I didn't know my mic was live. Hey, Brett Huff from Stephens. Thanks for having this event. Appreciate it. And thanks for the clarity on a couple of the big questions that we've been getting. First around durability and second around disintermediation around AI. So appreciate the clarity on those. To dig in a little bit on the AI question, you mentioned a couple of things. And we kind of have a theory as well that inferencing is really good for LLMs. But getting the data in and then getting an action out, we think probably remains the purview of systems like yours. Can you give us more meat on the bone on how do you defend against that sucking up the data or keeping the data?
Because AIs are pretty good at managing structured and unstructured data. And then talk about what specific actions can keep you embedded in that workflow. Thanks for the question.
Just we'll start off with one thing. And then Yamini should chime in as well. Is that when an LLM invokes HubSpot through one of our connectors, that's happening on behalf of a human. That's happening on behalf of a user. That's provisioned in HubSpot. We know the security. We know the permissions. We know all the things that are happening. This is not a bulk data. This is not a content business, right? Where it's like, oh, you can just slurp all the articles over the web and put them in the LLM. And now it can do everything. That's disintermediation. In our world, someone or something is going to need to manage all the things.
It's not just a matter of the data. It's a matter of all the workflows around that data. It's a matter of all those users around that data. So the way I think we should be thinking about AI is think about it as a new operating system. So when we saw the internet come along, it's like, oh, we have this new thing called a browser. And now software is going to plug into that browser. The browser became the interface that users interfaced with. And I think AI is going to become similar. But the browser did not disintermediate web applications. They actually drove a bunch of explosion of web applications. We're going to see the same thing with AI.
I think there's going to still be a need for systems of record for structured data and unstructured data together and be able to kind of manage all those actions, do all the permissioning. I don't think that goes away. I think that's going to just be driven more and more important. I think that's a major change coming up.
I think that's exactly right. And a lot has been said about even our connector strategy. And the question of why do we feel so confident that we're building connectors into LLMs? Because we think that we have the growth context and the customer context. We know the users. We know the actions that they need to take. And we're not just like an article that is now on the internet and therefore the value is gone.
We are actually the logic, millions and millions of lines of logic that is specific to a user, their role, the actions that they can take. That is what a SaaS application is, and it's not just like going to go away. Now, the reason we built connectors is, one, this is a new operating system just like Dharmesh talked about, and within the new operating system, there are new customer behaviors and patterns of asking questions that are developing. We want to be there. If we call ourselves a customer platform or just any platform, then you've got to be where everybody else is, and if there's someone who's asking a question, we want to be there. We want to bring that customer context into that conversation.
And then when we do that, I hope you saw this in the spotlight. We light up, which means LLM now recognizes HubSpot. AI referrals get back from LLMs into HubSpot. Leads flow from LLMs back into HubSpot. And actions, the logic of the action still remains within HubSpot. And so we're learning so much from being one of the first out there and going through with this. So it's such an exciting time to be. Net of all of this is that this allows us to add more value to our customers, which means we will do that. And we will focus on just that.
OK. Probably have time for one more question so we can keep Dharmesh on time. And then we're going to have another half an hour or so to connect with all of you.
Yeah. Hi, Taylor McGinnis with UBS.
Thank you all for taking the time today. I'd love to press on the evolution of SEO. I think that was a big part of the keynote. So Yamini, maybe you can talk about what that means for the runway left and opportunity for Marketing Hub and Content Hub as we look ahead. And then Kate, for you. I think you had a slide up. And it looks like Marketing Hub has been growing in the low teens. And I think you made a comment earlier about how one of the headwinds to NRR has been contacts and some right-sizing there. So how do you think about those growth drivers evolving as we look ahead and especially given some of the new features announced today?
Yeah. I'll take the first part. Absolutely clear that there's a huge shift within marketing.
What is changing? We've talked about it really at length: traffic to websites is going down. But AI is making it possible to get to know your customer, drive better capture of intent, and therefore drive conversion better. That is exactly what our reimagined Marketing Hub, Data Hub, Content Hub are going to enable our customers to do. Our customers are asking us for how do you drive leads in this whole shift that is happening. We now have a clear answer and a clear playbook. Yesterday, we had our partner day. We activated the whole partner ecosystem. They are excited to talk about this playbook with our customers. I want to address one part of the question that you asked, which is, what's the remaining runway for Marketing Hub? Maybe I want to clarify it. Our Marketing Hub has been a full funnel solution.
And SEO is a top-of-funnel tactic. So top-of-the-funnel tactic in SEO is changing. But we've always had one of the best email automation and sending engines out there in the market. We've always had diverse channels that we have supported from social channel to ads to a whole bunch of other channels. And now we are adding AEO into marketing as well as driving diversification within content. And so I think you should think about Marketing Hub plus Content Hub as a full funnel platform that we are now reimagining to support the new playbook, which is Loop. And that is exciting. It's a whole nother set of opportunities for us that I'm super excited about.
Yeah. Me too.
Yeah.
And Taylor, I think your other question is really getting at what is really happening with net revenue retention. When are we going to see it inflect?
I think the, I would say the long and short of it is that the components of net revenue retention really haven't changed. We've been talking about the same components over the last few quarters. We have this really nice and strong foundation of customer dollar retention. We are benefiting from the tailwind of the pricing change we made in sort of the middle of 2024. That is going to continue to be the case. What we have seen is that outside of that seats tailwind, we have seen headwinds in sort of our net upgrade drivers. I would say the good news is that the downgrade portion of that has been stable for about, I would say, eight quarters or so. It's really that reinflection of the upgrade momentum that has been lagging here.
And I would say, I think it will, we have the room to continue to increase net revenue retention over time. It's going to be a combination of buyer behavior continuing to loosen up, but also some of the newer levers for growth that we've talked about over the course of today starting to really gain scale over time.
Awesome. Well, great questions, everybody. Appreciate the support. I think we're going to break here, grab some lunch, come back into the room. Yamini, Kate, and I will stick around and have an opportunity to catch up with you all separately. And Dharmesh, I think we're going to let you run.
I'm going to run.
Good luck, Dharmesh.
Good luck, Dharmesh.
And thank you so much for all the support.