SaaStr AI Day: Your Product-Led Growth is Too Slow: Getting Marketing, Engineering, and Product t…

Perfect. We've got one East Coaster joining us. Hello, Brian. Oh, we've got some Texans. Oh, a few Texans. Even a Londoner. Wow. Ooh, Mexico. Okay. Yeah, Bay Area. I'm also in the Bay Area, so a little boring. Sweden. Okay, wow, quite the variety of places. Canada, amazing. Another Texan, love it. Awesome. Okay, I'm going to go ahead and kick us off. So like I said, thank you everyone so much for joining us for our very first Sastr.ai day. This morning we are so— or afternoon depending on where you are— we are so thrilled to have Brian Thomas joining us. He is the CPO at Customer Customer.io, and he will be discussing why your product-led growth is too slow and how to get marketing, engineering, and product teams all to collaborate. So a quick, uh, just a reminder for everyone, we want this to be as interactive as possible. So feel free to keep your cameras on if you're comfortable. Um, and if you have any questions throughout, uh, the session that pop up, you can drop them in the chat. Um, if Brian has time at the end to answer them. So thank you everyone again for joining and Brian, take it away.

Right, yeah. Hi, I'm Brian Thomas out of Seattle, Washington. I'm the Chief Product Officer at Customer.io where I lead our design, engineering, and product management teams. And most importantly though, I really love product-led growth. I joined Customer.io after living in the painful world where there was this giant wall between product and marketing. They existed in completely different separate organizations, completely separate tools, had completely separate schedules, and I fell in love with product-led growth because it really just bust that barrier. I believe great products and great marketing are one intertwined story. So I've devoted my career to helping folks tear down those barriers and build great product-led growth motions. And that's what led me to Customer.io. We're a customer engagement platform used by many great SaaS companies to use, to turn their stream of real-time usage data into personalized messages. In a nutshell, you plug in your analytics data, warehouse data and whatever else you have. And then you can create sophisticated automated workflows that drive emails, in-app messages, push notifications, SMS, or whatever else you want to send. We're built for PLG and are PLG ourselves. And that's what gives me the confidence to come here today and say that, like, there's a pattern that we don't like to talk about in PLG, which is that in many cases it's still way, way too slow. While we busted down many of the walls between product and marketing, there's still this massive barrier that we treat as fixed. I can capture that barrier of what's making it so slow in one dreaded phrase I'm sure you all know, which is, it's on the backlog. This wall sets boundaries around where PLG is allowed to be quick. If it's onboarding, we're happy to quickly iterate on activation. Otherwise, it's on the roadmap, part of the strategy, waiting. We have great intent, So why are we doing this to ourselves? Well, tell me if you recognize this story. If you decide it's time to weave message sending in to drive some key metrics, usually you're talking like activation or conversion, perhaps some nurture or win-back. You craft a perfectly customer-oriented onboarding process or quickly iterating on it, optimizing it, and sure enough, you get some activation boost. Excited about the success, you keep trying to move forward in the lifecycle. How do you start influencing key actions to happen after first use? How do you deepen engagement? And that's when you start hearing that terrible phrase, it's on the backlog, let's make some trade-offs, or sometimes even worse. That's a transactional message. We own that, not you. Instead of continuing that pace of iteration, you're stuck in planning cycles and waiting. Collaboration is broken down. We're back in that old world. It's moving way too slow or too siloed. Missing that fast collaboration that characterizes product-led growth. And I think this breakdown happens largely actually because of the tools we're using. It's a, it's a martech problem. Martech models your business as a total mess because it was built to mirror analytics, the same API spec that's been used for Google Analytics or Mixpanel, Amplitude for ages. This spec isn't how any of us imagine our businesses. Marketing tools identify people and their actions. You know, person A clicked button B, scale that up to thousands of unique actions and you get a big mess that's pretty hard to understand. But in reality, people aren't just completing actions, they're interacting with things in a business, the things that actually create value for them, creating a new project, booking a reservation, insuring a pet, taking a class, whatever it is that your SaaS does. And unfortunately, marketing tools don't understand these things or how they relate to people. And that's the thing that pushes all this core logic out of the world of PLG into the really slow world of planning in, in your backlogs. But what if we could take these essential functions out of code and really speed up the pace there? What if marketers, product managers, engineers could work together on these core things in your business. And I'm gonna show you how Customer.io makes this possible with this concept of objects. To help me out with this, I'm gonna introduce a fictional SaaS company called EduLang. It's a language school and edtech SaaS platform that helps folks learn languages. And let's jump into their Customer.io account. Now, the quick lay of the land here is that you've got that stream of real-time activity coming in with what all the students are doing. And what this company is doing is turning that into a set of automated campaigns to trigger email notifications, push notifications, and such. And it's got really the universal set of SaaS campaigns in here. You've got your initial onboarding drip trying to encourage people to sign up for the next language course. You've got an onboarding campaign when someone first starts, and this type of stuff has been the domain of the marketing team, and it's super valuable to the basics of driving activation and conversion. This language learning company, though, does something unique. They have these exceptional teachers who provide regular office hours that allow these students to go and practice the language in real time instead of just working with the software. And there's a big problem in the business, which is that the notification is buried in some weekly course update, and the students just aren't noticing or showing up for these highly valuable office hours. I'm convinced that if students knew they were scheduled, we can get more folks to come, folks to come. It would transform their experience with the platform, and they'll complete courses, take more courses. Problem is it doesn't fit that just like martech mess of streaming events. I need to know a list of students who are in the course. I need to know what course they're taking. I need to know who their teacher for that course is. I need to know when the teacher has office hours scheduled for, and somehow I need to link all that up. And that definitely fits the, the, the characteristics of the situation where you say, It's going on the backload, backlog. It's going to be months or weeks before we can make time for that. And that tends to be just the end of the story, right? But not if you decide to model this stuff in software instead of code. So check this out. In CustomerIO, all the things in your business live right here alongside the rest of your data. This works with any type of thing your business has, companies, accounts, reservations, insurance policies, playlists, anything at all. In this case, it's the language learning courses. And most importantly, if I go and look at the relationships here, I can see how this German course is related to all the people, all the students and the teachers. And hopefully you're starting to see it, which is that the business model existing in software means I can start to move faster. I have the list of students in the course. I know who the teacher is. I know what course that teacher is teaching, and I know when the teacher is scheduling office hours for. I see how it's all linked up without the dev team. So let's send some follow-ups and get them to attend. This is the, a campaign who, that's notifying students of those office hours. If I look at what's triggering it, it's not based on some random event. It's actually based on the business relationship. It's about having a course with a next office hours date and then targeting the people who are students. And then what that does is triggering these automated workflows. And the beautiful part about this is once you got that logic here, it's an experimentation surface for you. I can add a multi-split branch and perhaps go and say, hey, if you're struggling, you really need to be in these office hours, try different content for it. To give you a quick idea of how this works in practice though, I'll go over and pick this, this relationship, go and say like, so normally this is something that would happen in API. I'm going to simulate this just to show you how it works. You know, you go and you say, when's the next office hours? The next office hours, plug in a time there. Just by making that change to say, hey, there's a new attribute on this relationship. What's happening is I can go and see the activity and see that a message has been sent with that notification, including a whole bunch of relationship data. All that complex business logic got boiled into one message. And once you do this, you get this really beautiful world where you can start to imagine whatever you want with your core business logic and go and start to create campaigns around it. So another one we'll look at is this idea of re-engagement. The one challenge in the business is getting courses or students to come back and engage in the next course. The thing that's triggering this again is something that's based on business logic. When someone, of course, is allowed to register for the course and to all the people who are on the waiting list, you can go and start to get that experimentation surface to see what's going to work best for converting that customer. Now, one thing you might be realizing is I'm talking a lot about moving this logic out of code and directly into a rapid experimentation service, but you're stuck still writing a little bit of code, and that's intimidating for a lot of the people who are trying to move really quickly. Well, the beautiful thing about moving this logic out of code and into software is that we can start to plug this data into an LLM that can understand the data, understand the context, and then be able to start generating content, writing the code for you, and then start to have a dialogue with it. You know, in this case, I might say something like, remind them they only have 3 days left. Because it understands that structured data, because it understands the structure of Liquid code, it can start to go and generate messages that would have taken a long time or a lot of coding knowledge before in order to get going. And really, that's my, my parting thought for you here is by modeling the core business logic and software, it not open— only opens the door to this, faster iteration with humans, but it's just the tip of the iceberg of what's possible when you model that logic, when you feed it into an LLM that can take that structured data, can take the knowledge around your business and around how to generate code in order to generate content and segmentation that's most relevant to your specific value prop. So finally, as you saw in this demo, By bringing this, the core business logic directly into your messaging surface, you can start to extend PLG principles beyond their typical areas of strength, start to iterate them on much faster, and get your PLG motion working across the lifecycle at the speed that you want it to. When you use the right tools, teams can accelerate their growth and improve communication. See if how I did there on time. If there's any questions.

Yeah, you did amazing on time. So we appreciate it and appreciate all those insights. I think we have time for a couple of questions. So how often do those teams that you mentioned, how often do they meet and then how do they align like offline to stay connected?

Yeah, so this is just the typical like development cycle that you see like You know, you have, I mean, for us, we have our product marketers, our product managers, we have our engineers all trying to work together to create strategies, lifecycle strategies around the things that we've shipped. And the big barrier I'm calling out is like when you talk to an engineer and you have to go put that in the backlog, wait for it to be scheduled across everything else in the business, and you're unable to follow just like that daily or weekly routine where marketers, product managers are meeting up together and then make that change immediately in a visual canvas, right? Not having to wait for it to be QA'd, run through the actual deployment process and software. So that, that cadence really is about moving at the speed that product marketers can move versus getting caught up in the development cycle.

Amazing. And then one final question. In your opinion, what have you seen already change in B2B SaaS with AI?

Oh man, I mean, the biggest change we see is just the way that we're all working every single day. I mean, the speed that we can do research, speed that we can synthesize information. Like one of, one of the favorite things that I, we've been playing with here is the Google AI Studio where, I mean, we love visual workflow builders, canvases like you just saw, like you can just drag and drop to go and experiment with different models, drag and drop to experiment with different flows based on input. It's just like the speed that we can get a lot of synthesis happening with a lot of information outside of, of our core. And so been excited to both use that in terms of our day-to-day operations, but also as you saw in terms of building it into our own software.

Great, amazing. Um, and actually we did just get one last question in the chat. This is from Ryan. He would like to know what stage is a good time to integrate your product in terms of cost versus ROI? $1 million ARR, $5 million, etc.?

Yeah. Yeah. So one of the principles of Customer.io, we started as a bootstrapped company, and so we've always put a lot of energy into making sure that whether you're bootstrapped or whether you're Series A, Series B, Series C, you can pick the best software out there and we will grow with you. We have a startup program. If you're a bootstrap company, you can go and use this for a year for free. And then it's a very approachable— you start at $100 a month in order to, to get started using this. And so I think the moment, the earliest moments of your software is when this type of experimentation starts being valuable and it scales all the way, all the way up. And, you know, we've differentiated ourselves in the market by trying to make sure that we can span that entire lifecycle.

Amazing. Awesome. Well, we are officially out of time, but thank you so much, Brian, for all of your insights and joining us for AI Day. And thank you to all of the attendees. Just a note that that this session was recorded, so you can reference it in the future on all of our social channels if you'd like. And you can also reference Brian's slides as well. So thank you again, Brian, and I hope to see everyone in future AI Day sessions.