The Tactical Playbook for Building a Scaled Customer Support and Success Operation

Hi everyone, thank you so much for being here. Hope you all enjoyed the keynote. That's a tough act to follow, but really excited to just walk you through our journey. I'm Kayvon Sadiq, I'm the VP of Customer Operations at Lucid Software.

And I'm Sierra Sagoon, I'm a Skill Account Specialist at Lucid Software. And today we're excited to walk you through our journey from a startup focused on one-to-one support to where we are today as a strong self-service model, and finally to a future centered around zero-click support and where we're going with AI. Our core story is how we've managed to decrease our support ticket volume as our user base has expanded pretty dramatically. All right, so what we're gonna cover real quick.

So introduction to our vision, and then we're gonna walk you through the three pillars of our support strategy, and then we're gonna make sure to talk about some of the challenges we're facing, and then at the last 15 minutes, we're gonna open it up to any questions that you may have. All right, so first I wanna see where you guys are all coming from today. So if you don't mind pulling up the app, our first Slido question is going to be, where are you joining us from today? And so hopefully you guys have done this before, but just submit through the app, and we'll tell you a little bit about where we are coming from.

So I started my journey at Google. So I spent eight years at Google starting in 2007. Initially I was in our California Mountain View offices, and then after four years, I moved to Dublin, Ireland, where I was focused on community support for Google's many products that our customers are facing, products like Gmail, Chrome, Android, Google Maps. And then more recently in 2015, I joined Lucid, and I started in our headquarters in just outside Salt Lake City.

But for the last three years, I'm based out of our Raleigh office, which is our newest office, but also our fastest growing office. And with that, I'll pass it off to Sierra. All right, so I'm currently based in San Diego. I'm remote, but I had the opportunity to start my time at Lucid last year, based in our headquarters at Salt Lake City.

And while I was there, I had the opportunity to train and work across a lot of our teams that are focused on helping customers at scale. So that's our community team, our CSS at scale team, our content team that manages the help center, as well as our sales at scale team. And I'm excited to talk with all of you about some of those today. Perfect, all right, well let's show the results, where you guys are all coming from.

So it looks like Atlanta, Austin, Houston, and Denver are the top destinations where you guys are coming from. And then we also have Ireland, which is super exciting, and some Naples, I see some other international locations, London as well. So again, thank you all for coming and for joining us. So I wanna talk a little bit with our CS strategy in general, right?

Or like how we think about support strategy when it comes to Lucid in our software. But before I do that, I wanna understand what is your familiarity, right? Like we're not a car company, those of you who are here to see Lucid Motors, sorry to disappoint you, but I would love to just hear from you, have you used either Lucid software, any of our products, Lucid Chart, Lucid Spark, both? So again, if you could quickly go into the app and respond there.

And while you're doing that, I'll just talk a little bit about our vision and background. So I always like to give the analogy of driving to a bank during normal business hours. You're faced with a decision. Do I wanna walk in, wait in line, and talk to a teller?

Or do I wanna go to the ATM and help myself, right? I think for the majority of our customers, for the majority of their issues, they actually prefer to self-serve the content that they're looking for, right? And I think that's really the heart of our strategy is in order for them to be able to do that, you're going to need to provide them the resources in order to be successful. And of course, not every issue type is going to be something that can be handled by themselves, right?

If there's a bug or there's a problem with billing side, they are going to need to speak to an agent. And so you also wanna make sure that for those cases that you're setting the customer up with quick responses and the support that they need, right? And I think our product is very unique in many ways in that our users are finding about the product organically. For instance, they'll go to Google and they'll say, I wanna draw an org chart, right?

And then they'll discover our software there, they'll sign up and try it out without talking to a sales rep, right? When they wanna learn about it, they go through our Skill Jar resources, our tutorials. And then when they want support, they're also used to the self-serve model, right? It really does work for them.

I wanna make sure that like, you know, all industries are a little bit different, but this is what's worked for us. And so in the early Lucid days, when I joined in 2015, we were a team of three, very small, but we really didn't have a help center. We didn't have a community. Our team was really focused on email support and telephone support.

And a lot of the same issues and questions were being asked over and over and over again, right? And so that's where we started, but where are we going, right? So before I talk about this, let's just go back and see the results of that Slido. We have a lot of people here who are not familiar with our software, which is great.

And then you can see that Lucidchart is heavily more familiar to you guys than LucidSpark, which is generally the case. Awesome. All right, so going back to the slides, the heart of our strategy here is to really talk about what we've seen on the product side. So when I joined in 2016, there were about nine million customers that we had.

Today we have over a hundred million customers using our products. But when you look at the support ticket volumes over time, it's telling a very different story, right? And this shows just support volumes, the colors, don't mean too much, their categories. The billing tickets are that teal color.

The red shows product and enterprise tickets, right? For those who are curious. But our primary goal has always been the same. Providing a great customer experience, but also allowing them to self-help and seeing that through decreasing ticket volumes.

And you can see the trend, since I think we started pulling this data into Tableau, has been very much going down as we've continued to scale, despite our ever-growing user base. And so the big thing I want you guys to walk away from this conversation with is, what did we do to get there? It's not going to be a silver bullet. We're gonna talk about the pillars, but at a high level, we started with one pillar, kind of gain after gain after gain until we hit a plateau, and then we moved to the next pillar.

And every pillar kind of built on the last one, right? And that's really led to our strategy. But I really wanna focus on, you know, who are the people driving a lot of these innovations? Because I think that largely has not changed.

So, you know, when I think about what makes a successful support and success agent, it's a couple things, right? The non-negotiables are going to be, are you passionate about helping people? And are you a generalist who can adapt to an ever-changing job description every day, right? Because, again, I've been at Lucid for over 10 years now, and things are constantly changing, the industry is changing, and obviously with the advent of AI, that's only going to go faster in terms of the pace changing.

So beyond that, though, when I think about the first hires on a team, right, they're all gonna be doing, you know, the core work, but beyond that, they're all bringing unique skill sets to the team. And so you're gonna see that theme throughout our pillars today, but these are kind of those categories. You have the content specialists, right? The people who are writing the user manual on how to use your software.

You have the community managers, the people going in and finding those passionate users in your user base and incentivizing them to help one another. You have the data analysts, right? Like, how do you measure success of these initiatives? How do you tell the story to leadership and beyond on how the team is scaling?

And then you have the systems admins, right? The ones that are so knowledgeable in Gainsight, in Zendesk, and, you know, building out the AI structures for how you're dealing with some of these top tier issues. So you're gonna see a common theme here regarding the people behind all of this. All right, so first pillar of support I wanna talk about is, again, like each pillar compliments and reinforces the others.

And the first one really is at the heart of the self-serve motion. It's the Help Center and the community, right? So we have over 250 articles in our Help Center. We have over 14,000 community posts in our community.

And there's a wealth of information on almost every topic. And users are able to self-access this. They can also see our Skill Jar resources, which are embedded in our Help Center. And depending on how you wanna learn, whether it's through text, through videos, or what have you, there's a solution for everyone.

And the Help Center really is the core of our self-help resources. It addresses top questions and issues to effectively intercept or deflect some of these issues you're seeing. When we first started the team and tickets started coming in, we found ourselves repeating the same questions and answers over and over again. And so that formed the basis of our first Help Center, right, that content.

And then when you look at the analytics behind the Help Center, users are coming to your Help Center, typing in exactly what they're looking for in your search bar, and you're able to then see, okay, what is the content you're surfacing to them? Is that the right content, or is there missing content? We also have federated search across Skill Jar, our Gainsite community, as well as our Zendesk Help Center that surfaces all of our content at once so they can see sort of that whole picture. On the challenges side, right, so our product and our Help Center is localized into 12 languages.

And so there's a challenge with that, right? There's costs, and not all of the translations are machine translated. We spend a lot of money on human translations because, again, this is our top viewed content and really the manual to success with using our products and our content specialists are pouring into the data and making sure the content is really top notch. Unlike the Help Center, the community is primarily designed to answer long tail content.

An example I love to use is we have Lucidchart and LucidSpark apps for Android. But if a user comes to us and says, hey, I have a Samsung Galaxy S19, I'm having this problem with your app, it's unlikely that our support team is going to have that exact phone and is going to be able to answer their question directly. But chances are one of our 100 million plus users has that exact same phone and may have also come up with that same issue. And the community allows you to connect these users with one another and provide them the help that they need.

So beyond that, we also have our Lucid Legends program built on the Gainsite community. We'll talk a little bit more about that in a little bit. But I really wanna walk through some of the success that we've seen in the community, right? Community really allows you to do more with less.

And the way I like to think about it is you can answer a question in email once and you're helping that one user. But if you answer that same question in the community, it's gonna live there in perpetuity, right? It's gonna be shown through, whether it's Google searches or now AI searches, we'll talk about that in a bit as well. But that's really our philosophy is for almost all questions outside of billing and outside of bugs, the community is the optimal experience for everyone.

And so what does this graph represent? Well, our peer support is climbing quickly, right? Initially, when we set up the community, users ask questions, our team went in and answered them, which was again, far preferable to answering via email. But now with Gainsite and all its platform capabilities, we've incentivized users to help one another.

And the green line here shows you that our user community answered 39% of users for community questions. And there's plenty of room to grow from there. Our Lucid Legends program, which is kind of like our top contributor program at Lucid, has 125 Lucid Legends and counting. And they've shared over 4400 answers.

It exceeds at this point a full-time FTE on our team. And again, these are just volunteers who are passionate and we've been able to kindle that passion into really helping other users. Beyond that, ticket deflection has also tripled, right? So questions that we think are, they're coming in one-to-one support, but we think they're better answered in the community.

We're deflecting those over and providing the answer in the community where everyone can benefit. And that's really allowed us to hit the tremendous scale that we've been able to. But I think the key thing here is, I talked a lot about how we've scaled support. It's actually very easy to do, right?

Like if you just hide the contact us form link so that users don't know how to contact you, you'll quickly see the number of tickets go down. But for us, we've never done that, right? Anyone here can send an email right now, support at lucidchart.com or support at lucid.co. You'll go directly to our support team.

And so as great as Help Center and Community are, it's not going to reach everyone. All those people who choose to email us directly kind of sidestep that. And that's why the second pillar becomes so important. And with that, I'll pass it off to Sierra.

Thanks, Kavon. So let's talk about macros and auto-replies. So this pillar focuses on automation. One more slide.

Oh, my bad. Okay, perfect. This pillar focuses on automation. And I'm going to explain how we use macros and auto-replies to bring down our overall ticket volume.

So macros are designed to standardize agent responses to top user questions. And these aren't just templates, these are multitaskers. So with a single click, you can update the ticket status, you can change the priority, you can assign the ticket to a different person, a different queue, and you can also tag the ticket for better data reporting. And this also gives us a standardized voice.

No matter which agent is helping the customer, they receive a consistent and professional experience. So macros have been instrumental to our team to quickly spot trends and support data. And even if we might not use the writing in a macro, we still are encouraged to use it because it tags the ticket. It will already categorize the ticket and lead to very rich data reporting.

And you can see that in these charts shown here. The chart on the left, this shows our password reset tickets. And we're always going to have a healthy or normal amount of these tickets just because, you know, like emails get sent to your spam or you forget your login information. But sometimes there's actually something more wrong.

And we can use this data to identify it. And we can just quickly send this to our engineering team and fix the issue. The chart on the right, this is similar, but this is for our Visio import into Lucid. We do offer a lot of functionality to import documents from our competitors.

And again, you always will see a healthy or normal amount of these tickets. And that is because we do have to decide between functional fidelity and visual fidelity because of the differences of the two applications. And a lot of the times our engineers will err on the side of functional fidelity so that your shape behaves as originally intended. Well, visually, this may be a little different.

But again, sometimes our import breaks. And we can use this data to escalate this to our engineering team and get it fixed. Okay, let's talk about auto-replies. So we rolled out our auto-replies in March of 2024.

And auto-replies are designed to give a user an immediate response. So on our, a user will submit a contact us form in our help center. They will then be directed to select an issue. This issue will trigger a pre-written response with troubleshooting steps.

The user is then asked to respond to the email if they need further help, and this will be taken care of by an agent. Now this chart shows our results. This is our successful and our failed auto-replies. So in the orange, you're gonna see our successful auto-replies.

And that's when a customer fills out a contact us form, the auto-reply gets sent, and we don't hear back from them, it helps. So maybe they're having storage issues. They select that in the contact us form. We send them an email.

We notice you're having this issue. Here's how we can help. The blue, these are our failed auto-replies. So this is when the contact us form submitted.

We send the auto-reply. It does not help the user and the user response. This can be a response such as thank you or that they need more help, but that is what's captured in the blue. But to really underscore the key takeaway here is that in the first quarter of the calendar year, we deflected 1,487 tickets.

And this is the equivalent to one full-time support agent. But it's important to recognize the limitations. Auto-replies only respond to form tickets. Now this is a majority of our tickets at 64%.

And also by relying on the customer to self-select the issue, this does limit the complexity and the number of auto-replies we can deploy. We don't wanna make our form too complex, so we don't wanna give too many options. All right, so let's look at how our first two pillars contribute to our ticketing landscape, just before we jump into how AI can help us scale further. So this is a thank you diagram.

And this is looking at ticket volumes for the first 10 weeks of the calendar year. You'll notice on one side, you can see our ticket types, our product tickets and our billing tickets. And on the other side, you can see our tools. These are our auto-replies and our community.

And then you can also see our manual tickets in the middle. That's gonna be anything that a support agent has worked on. So you'll notice a trend between the community and the product tickets. And this is just because typically product questions aren't as sensitive as our billing tickets.

So users can post in the community without having to worry about their sensitive data. Whereas billing, these may be a little more sensitive, so that's where our auto-replies are a little more valuable. And you'll notice the help center is not captured here. Our help center is our first line of defense, helping users help themselves.

We receive hundreds and thousands of page views per month at our help center, which really helps deflect ticket volume. However, that is a little difficult to measure, so you're not gonna see it on this diagram. But I do love this diagram because it shows how each of our pillars reinforce and really complement the others. Okay, this leads into our third and newest pillar, AI.

And it would be impossible not to have a section on this day and age. And AI has implications across all team members. So this is content managers, community managers, a system admins, and data analysts. And before I dive in, I did want to note that there is a natural tension in the relationship between the community and AI.

AI both relies on the community for its content, but also incentivizes users to no longer create content. AI depends on the rich and detailed content created in the community to give users a generative answer. However, because AI gives real time answers, this is a compelling alternative to posting and engaging in the community. And it might intercept many of our users, decreasing the amount of useful content in the community.

All right, let's talk about what we're doing with AI. So we are being very cautious. We are setting up guardrails to make sure that our users are still supported. We are doing nothing generative.

And if the tool doesn't know what to send, we have it air on the side of escalating to an agent. We have been building on our strengths. We have a long standing foundation of optimization, scale and automation. And we are also relying on our team to bring collaboration, innovation, and ownership to our experiments.

We are going to continue to build out our career paths, up level our roles, and increase our scope and impact so that our team remains relevant. And the time that we are saving with AI, with the community, with auto replies, this time can go into other areas of the business. Whether it's a new AI project, whether it's customer success at scale, or sales at scale. And these skills are giving our employees marketable skills to bring with them.

All right, so AI can be very divisive, which is why I think it's important we talk about what we aren't doing here. There's a lot of valid fears and concerns with the use of AI. So we are not getting caught up in the hype, and using AI without thinking about its downsides or weaknesses. We are not just changing everything and throwing out our existing support philosophies.

We are not removing roles from our team, or downsizing, and we are not leaving our users without the support that they need. Thank you, Zira. And one thing I wanna stress again, is that tension between community and AI. The holy grail for us is, if you've asked the question that's already been answered in the community, then we want you to get an immediate answer from AI, because we know we have that answer.

But if you're asking a question that we've never seen before, we don't want AI to respond. We want it to be in the community so it can get a good response, and then it can keep our AI up to date with these new posts. So it sounds so easy, right? But that's the holy grail that we're working towards.

All right, so what are we actually doing? We walked you through the auto responses, and that had some limitations. Not everyone is going to our help center and contacting us there. You can only have a certain number of selections from the dropdown before the form becomes unwieldy and users give up.

And so our first advent into AI is extending our auto reply strategy through AI. So with AI, we don't have to worry about the forms or users mis-selecting what their issue type is. We can expand the same concept of figuring out what is this issue type, and then sending a quick auto response to all of our tickets. Because our AI is now going through all tickets, looking for the buckets, which we've now expanded to over 30, for things that we think we know how to solve in giving those responses right away.

And then again, if the agent, or if the customer says this isn't helpful, they respond, it goes straight to an agent, and we use that data to make our AI even better. Okay, so talking a little bit, right? Like so we have about 36% of tickets that completely bypass the form, and so now AI has really allowed us to tackle that. And then the other piece around AI is it's not just having an impact on one-to-one support, it's really having an impact across the board.

So when you think about AIO, what we're seeing is that a lot of customers are sidestepping our properties completely. When they do a search for like, I wanna, how do I draw a straight line in Lucidchart? In the Google search results, they'll get their answer right there, right? And more and more customers are doing this.

In fact, we sent out a survey, and our own customer said they are more likely, twice as much more likely to get their questions from like an AI agent versus from searching our help center or doing a web search. And so what does this actually look like? Here's a screenshot of the Google AI summaries. And again, this is lost traffic.

We're seeing it in the help center, we're seeing it in the community. We're getting fewer page views because a lot of these are being deflected from search engines, and this is not even showing you what the experience looks like if you go directly to chat GPT, Gemini, Claude, right? It's a completely different process. And so we're seeing that shift and it's having a pretty big impact on us.

Thankfully, we actually don't need to start from scratch when we think about our strategy here because essentially our community is a goldmine of knowledge that's adapted for AI. It contains unique, highly specific and educated information that our formal help center articles don't cover that we talked about. And it's also driven by real world scenarios and in the natural language of our users, right? These are things that AI is really prioritizing at the moment.

And so, it's been such a great transition for us. But again, we need to keep the content up to date so that the AI remains up to date. All right, well, the whole game is changing, as I mentioned earlier. Some stats that I think are really interesting is first of all, there's so many different AI platforms that users can be on.

And each of them has a different methodology to ranking, right? I think in the old days with SEO, it was pretty easy. It was about page rank. How important is your site when being cited by other credible sources?

Well, a lot of that's changed now and every company's kind of doing their own thing. It's a little bit like the Wild West. And on average, companies are updating their models every 17 days. So even if you're building something to adapt to AIO today, in a couple weeks, it's gonna be out of date, right?

And so, I think it's early stages, things are still evolving, we'll see where they land. But again, the great thing about the community is it really allows you to leverage the signals that have always done so well. And then the Help Center as well, as you think about embedding Skill Jar courses and licenses, when you think about embedding YouTube videos or other videos into your Help Center, these are all things that in the early days are doing really well with AIO. So what are we exploring next?

We talked a little bit about generative AI, right? We're very sensitive about that for legal reasons, for security reasons. But that is going to be that next frontier that we wanna tackle, which is, okay, beyond that auto response expansion, how can we actually have our Help Center and our community as the source of truth, but allow AI to generate the answers to questions that may have not been asked in that same form, while making sure all new questions are keeping our content corpus up to date. And so, what I like to think is, in the future, you're gonna have the AI talking to the AI, you might be a user, you notice an overcharge from a Lucid account, you tell your AI, "Hey, I wanna refund for this." That AI talks to our Lucid AI, confirms all the information, works with the Zenith AI to get that response out to the user, works with our internal AI to make sure the refund is processed.

So I think that's the world that we're working towards, but we have some work cut out for us before we get there. So with that, I wanna thank you all for coming. And we have another, oh, or I slide, we have another question just for you guys to another Slido link for you guys to submit any questions. And with that, I think we will, yeah, open it up.

(audience applauding) Hello, I'm back, hello everyone. You guys killed it, I loved it. I thought it was really cool to see how much your team relies on community, especially extending that support out to the long-tail customers. And with SEO, how much more important it is now than ever to have a great community presence and have all those sources available there for it to be found online.

So thank you guys, I learned a lot. Starting with some questions here. So what incentives have most effectively led customers to contribute to your community? This is a great question.

And due to time, I had actually skipped a slide. I don't know if you guys noticed, but I saw this question come up and I wanna go back to that slide and talk to you about it. The thing I'll say with incentivizing customers is that there's so many different reasons why customers choose to volunteer their time to helping other customers. And you have to have a very scaffolded approach, right?

Because what motivates someone may completely be meaningless for someone else. And so that's what we've learned from our use cases, is constantly expanding what motivates these customers and how can we sort of bridge that gap. So some of the perks we have, you'll see here, the big one I would say for us is, GainSight's platform allows little things like badging, leaderboards, right? We never had that on our old platform and it was really, really hard.

Like people want to know that if they're putting this time, they can see themselves getting points and moving up our leaderboard. They wanna see that when they give a response, they have a badge that shows like, hey, I'm not some random user like posting them for the first time. Like there's authority to, and credibility to what I'm saying, right? So that's been really, really good.

And I think a lot of people are driven by this metrics approach. Alternatively, every year we do a hackathon and the whole company almost shuts down for about a week and works on like these crazy ideas. And at the last day of the hackathon, we actually have a voting process internally where Lucidites can vote on their favorite ideas and those make it into the product. Well, one of our Lucid legends actually heard about this and said, hey, I wanna be there for this.

I wanna contribute. And so we've actually invited our Lucid legends now to the hackathon where not only can they like hear all the ideas that we're considering, but there's an actual award that we've devoted to them. Like their votes work towards the Lucid legend award for the hackathon theme that has the project that resonates the most with our Lucid legends. Another example I'll give you is that, these users are so passionate about our products that they wanna have a voice in the product sort of room when it comes to what are the feature requests we're working on and things like that.

And so we will set up time with product managers and our Lucid legends for them to show a little bit of a preview of what they're working on and get their input. Our trusted tester program is largely focused on our Lucid legends, where before we release new versions of our software to the masses, we have a dedicated group that can help us like find bugs and sort of test the waters before the broader release. So again, there's no silver bullet here. We're still learning.

We have about 115, but I think the potential is so much higher. Just to give you a little bit of perspective, when I was at Google, they had eight or nine products with over a billion user, things like Gmail, Maps, Chrome, Android, and they had no employees responding to community posts. It was completely driven by the community. That's something that we aspire to, but we also don't have a billion users.

So it's a little bit harder, and we have to be a little bit more creative in how we incentivize. That's very cool. Let's see. In our industry, I see trends of clients resisting the self-serve model.

Did you see something similar, and what active advice would you give us as we move closer to the self-serve model? This is a great question, and I think, again, I started off by saying that so many of our customers are finding us organically. They're paying for the product without ever talking to a sales rep. They're going through our skill jar coursework on their own.

And so I think in our industry, we've actually seen the opposite. For instance, when we launched LucidSpark, we had a chat bot because we felt like, if you're working, for those that don't know, LucidSpark is like a visual whiteboard where your team can brainstorm ideas together, and it's meant to be really collaborative, like 20, 30 people brainstorming in the same LucidSpark board at the same time. And we thought, okay, well, if something breaks then and there, customers are not gonna wanna submit an email. They want support right then and there.

So we had a live chat bot upon launch, and then we found out that no one was using it. Everyone was choosing to go to the help center, and we stopped staffing that, right? Because that's not what the customers were looking for. So having said that, though, I do think we benefit from our customers really being at the forefront of the self-help movement.

However, you're always gonna get the customers who, no matter what, they wanna talk to someone, right? And we've approached that in different ways. We have a premium support offering where you can get on a phone with us at any point. You'll have regular check-ins with our support team, and they'll go over trends on what they're seeing in the data.

It's a very proactive, high-touch, white-glove treatment. We also, like I said, we will let anyone, doesn't matter if you're paying us or not, you can just send us an email and bypass all the self-serve pieces. And that's what really keeps us honest, I would say, is knowing that anyone can skip our self-serve model at any point allows us to always think, okay, is our strategy working? Or when we make a strategic change, are we sending more people to bypass it because they don't like what our strategy is, right?

And I think that really does keep us honest. Awesome. Are you using the failed auto-replies to help with support content creation in your community? Yeah, this is a great question.

So when it comes to failed auto-replies, and I just wanna stress that a failed auto-reply means if there's any agent interaction with that ticket, it's a failure. So if a user's like, hey, I lost my document, and we're like, hey, we think it's actually on another email address, or this is where you can find your document, and then they respond and say, oh, that's really great. How do I change the admin on my account? Completely new question.

Well, guess what, that's a failed auto-reply, right? Because it forced agent interaction. So when you're saying a failed auto-reply, I think what you're meaning is like, in this case, they self-selected, this is a lost document, we sent our macro around that, and they're like, yeah, I've tried all these things, that didn't really work, right? And so as we investigate with this customer and we figure out, okay, how do we get you your document?

And then we take that learning and we say, okay, is this something that's a one-off that's unlikely to be experienced by other users, or do we then need to go and change our auto-reply strategy and maybe add an extra bullet for this use case? And then the most successful auto-replies, again, the Help Center is kind of like that core to our self-serve model. Any auto-reply that's getting a lot of traffic, we ultimately wanna ask ourselves, what is the Help Center article that can be associated with that auto-reply? Because just as a sense of scale, we get about 4,000 support tickets per month.

We get over 300,000 Help Center page views per month, right? So the scale is vastly different by orders of magnitude. And so we know that if we can address something in the Help Center, we're going to be deflecting this. And so that is our strategy.

When it comes to failed auto-responses, it informs our AI, it informs our Help Center strategy, it even informs our macro strategy on the one-to-one side. So these pillars are very interconnected. Quick one here. What tool are you using for your federated search?

Great question. So again, when you think about our tech stack, we're using Zendesk for the Help Center, we're using GainSight for our community, and then we're using Skilljar for our courses and webinars and things like that. So across those three, we're using GainSight's federated search because it's the most powerful. It allows us to search across those platforms.

It normally knows when to surface Help Center versus community versus some of our courses. And it also allows for a lot of customization. So that's been by far the best solution. We've tried using Zendesk, but it wasn't as good.

Heck yes. I'm just kidding, we love that. Okay, probably last question here. While AIO makes it easier for customers to reach the community or Help Center, doesn't making it public also lead to the leakage of your own technology?

How do you perceive this area? I was concerned because the service I was in charge of was particularly competitive. I'm trying to make sure I understand this question. So AIO is surfacing our community and Help Center content within the search results page or within-- It's our community public.

Our community, oh, sorry, yeah. Our community is completely public. So you don't even have to be a user. You can log in and you have to log in if you wanna upvote, if you wanna ask a question.

But if you're just trying to view content, you can be an incognito, not log in and view all of our content. And in terms of the leakage of your own technology, I'm assuming that you don't put like, hi, whatever it's called. Like nothing that's top secret is going on your community. Ah, great question.

So GainSight has an incredible functionality where you can have parts of your community be private. So one of the things that we're working on, a little preview, is a visual refresh of the editor. And like I said, we have a trusted tester program where some of our users have access to this new experience before we roll it out more broadly. The challenge is I don't wanna have Help Center content showing the new experience for multiple reasons.

Number one, almost every article is gonna change because it's a visual AI. And I don't wanna have an old experience and a new one. And then maybe people on the old experience say, I like the new one better and they wanna like, how do I sign up? Like it has to be very controlled.

So what we've actually done is we've leveraged GainSight where we can link it to our own product data and say, okay, this user is on the new experience. Once they go to the community, they will have access to a part of the community that is completely private. You need to sign in. It's inaccessible to AIO.

And it's only for those users who have that new experience, right? So this is one of the most compelling use cases, in my opinion, because now we can really start thinking of community, not just for all use cases, but also, you know, like, do I want a community for our trusted user groups, for instance, right? Or all the different experiments we're running, we could have a separate experience in the community for that. And I think the platform is really, really powerful.

And we also don't want these things like screenshots of our new UI to leak, right? So with our Lucid Legends, they're under NDA, right? So now I get the gist of this question. Yes, we are able to sort of gatekeep our content in a way that only those users, who we want access to have access.

The secret sauce, as someone else put in their question. Okay, well, that is all the time that we have today. Thank you guys so much for being engaged, asking those great questions, and thank you both for your time. Thank you so much.

I appreciate it. Thanks, everyone. (audience applauding)