Appian Corp. Q3 FY2024 earnings call

• Cloud subscription revenue grew 22% YOY to $94.1 million, subscriptions revenue grew 19% to $123.1 million, total revenue grew 12% to $154.1 million, and adjusted EBITDA was positive $10.8 million.
• Pivoted to a more efficient cost structure with minimal short-term impact, focusing on high-value implementations in core verticals and use cases, seeing rising advisory service attachment rates, existing customer renewal uplifts, and higher pricing.
• Positioned as a leader in multiple markets like process automation, orchestration, etc., and is a process company where process is a frame for AI, with examples like a Latin American bank modernizing with Appian and 15 US cabinet-level agencies being customers.
• Personnel updates: Welcomed Mark Dorsey as Chief Revenue Officer, elected Boe Hartman and Michael Beckley to the Board, and Mark Matheos is leaving with Mark Lynch as Interim CFO.

In the third quarter of 2024, Appian's cloud subscription revenue grew 22% year-over-year to $94.1 million. Subscriptions revenue grew 19% to $123.1 million. Total revenue grew 12% to $154.1 million. Our cloud subscriptions revenue retention rate was 117% as of September 30th. Adjusted EBITDA was positive $10.8 million. Subscription revenue represented 80% of total revenue. Our international operations contributed 36% of total revenue. Non-GAAP gross margin was 77%, subscriptions non-GAAP gross profit margin was 89%, and professional services non-GAAP gross margin was 30%.

• Fourth quarter 2024: Cloud subscription revenue expected $95 - $97 million (14%-17% YOY growth), total revenue expected $163.5 - $165.5 million (13%-14% YOY growth), adjusted EBITDA expected $6 - $8 million positive.
• Full year 2024: Cloud subscription revenue expected $364 - $366 million (20% YOY growth), total revenue expected $613 - $615 million (12%-13% YOY growth), adjusted EBITDA expected $5 - $7 million positive. Adjusted EBITDA for full year 2024 is an improvement from the break-even forecast last quarter.
• Guidance assumptions include professional services revenue flat to down sequentially in Q4, on-prem license revenue increasing sequentially, other income and interest expenses Q4 $4 - $5 million and full year $20 - $21 million, capital expenditures Q4 $1 million and full year $4 - $5 million, and FX rates as of November 4, 2024.

• Actual results may differ materially from expectations due to risks and uncertainties described in our SEC filings. For example, FX movements are a risk but not considered in guidance as we don't forecast FX rates.

Q: Hey, this is Keith Weiss actually filling in for Sanjit Singh this morning. One for Matt just in terms of the discussion of like process versus agents. Is it really process versus agents or is agents sort of like a user interface and execution layer on top of process? So are you talking to us about an ability to create better agents or are you saying that just utilizing process kit? Basically obviate the need to use agents, if you will.

A: Okay, I see agents not just as a means to an end but more importantly as an end. The most important thing about the Agentic AI movement in my opinion is the interest in having AI that takes actions and how you get there is up for debate. And I'm proposing that there be a different a better way to get there to get toward action taking agents. If you empower those agents with information from a data fabric, structured, collaborators with whom to work and a set of pre-coded output levers that it can pull and utilize. We've just made our agents bionic is basically what I'm saying. I think process empowers and agents to be stronger. So I'm proposing process as a better way to achieve a superior version of the same intent.

Q: Derrick Wood with TD Cowen touching back on Agentic AI, what would you highlight are some of your key differentiators in tackling this market compared to other approaches?

A: Yes, well, first of all, we can source data better because we have access, we can quickly provision information from anywhere in the enterprise using our data fabric, unlike our competitors that have to gather the data into their own proprietary data center before they can properly interact with it. We can read and write from that data fabric, and I want to clarify that that is, our functionality is quite different from what's passing as data fabric in the market today. So that gives us an informidness advantage and then given that we believe that Agentic should happen within the context of a frame of a process, the actions are also far more auditable and specifiable. And in certain industries and for high-level tasks, certain industries like the government, they shy away from improvisatory behaviors and they want to see something they can truly predict, audit, and improve. So this slightly more structured approach to agents is more appealing to our customers.

Q: Derrick Wood with TD Cowen on pricing, how is the pricing on core tiers rollout going in terms of customer adoption and any meaningful impact to revenue growth at this point? And update on pricing around consumption with your AI solutions?

A: Yes, we are pricing by consumption with AI. I don't have much to say about it yet, other than that that's our pricing framework. As for our pricing system that we rolled out this year with our three tiers, really two so far, just know that the top one is being filled out over time with more features, but the advanced tier, we've seen substantial interest in the advanced tier, which is higher priced and higher functionality than the standard tier, which is more on a line with what we offered in our full package in years in the past. The advanced tier has been particularly popular as a start point for new customers who tend to buy into that level almost as a default. We lead with it, we sell advanced first, we focus on the specific features that you get from advanced and of course we tried very hard to make those essential and we find that most of our incoming customers have sprung for it. So we're getting the reaction we wanted and now we're also going back to our existing customer base and proposing upgrades to advanced and we're beginning to get those as well.

Q: Steve Enders with Citi on the macro landscape and changes on the sales and go-to-market side?

A: Okay, I don't have any insight into a Q4 budget flush. We're seeing a macro environment where we can succeed. Of course, there's certain tightness and variables we have to deal with, but macro is not a factor in our decision-making right now. It's not a factor in budget setting. It's not a factor in target setting. We believe we can operate in this environment. Yes, well, we come in with a couple of key sales priorities with our new leader, Mark. We're going to shift our energy ever more to larger opportunities. We're going to focus more on our existing and very happy customer base and expect to sell more back to them. And we want to bring forward our personalities, the human side, the partnership of Appian to differentiate us from our large bureaucratic competitors. We should be, and we should appear to be, the anti-big tech when customers do business with us. They should feel like they have more of a personal connection, and we're going to put that forward. So those are the main ways that we want to sell or continue to evolve our sales process.

Q: Jake Roberge with William Blair on data fabric and its differentiation?

A: Yes, absolutely. Thank you for asking. Our data fabric is a world apart from anybody else's, and it's one of the most important features that we've ever written. It's one of the best adopted features we've ever written. It is broadly used across our whole user base. Our data fabric allows you to treat the whole enterprise worth of data sources as if they were local data sources, addressable in a programmatic way as if they were local. You can read and write, and it's performance tuned. So we create indexes, we tune it so you can get instant response. Our competitors, by and large, are putting the data fabric label on a product that repairs broken integrations between their own products internally to their own siloed ecosystem. As I say, it's a world apart. Data fabric is more important than it's ever been, because AI is effective when and only when you provision it with data. In order to answer a question well, you need to provide AI the context. There’s really two ways to do that. One is you preload the AI with all the data it could need and expect it to be ready for any question. That's fine for some. It's expensive. It might be a bad thing for privacy to hand out over your data. You might feel like you're beholden or in a bad negotiating position against your AI provider, but for some, they're willing to just train AI beforehand. We're not. That's not our approach. Our approach is the other way that you can inform AI, and that is to wait until the question arrives and then quickly fetch the data that's pertinent to the question and send the query with the data all at once. That's retrieval augmented generation, or RAG, and it allows you to be far more private in what you share with AI and far less beholden to your AI provider. It's also more auditable. It's more changeable. You can train it. There’re so many advantages to it, but it all depends on one key factor, one non-negotiable technology you simply must have in order to use model number two, and that is you have to be able to fetch pertinent data within a second. From across your enterprise, you have to find the data that makes sense related to the question that was just asked. No matter where it exists across your enterprise, and that requires a data fabric that is both comprehensive and performant. That's the data fabric we have, and you see it's actually the key to the best model of AI. The best approach to private AI actually depends upon the kind of data fabric that we provide and others don't. That's why I'm so dialed into this, why I think it's such an important thing, because it's actually the foundation to our approach to AI.