Good afternoon. My name is Jenny, and I will be your conference operator today. At this time, I would like to welcome everyone to the Coveo Q2 fiscal 2024 financial results conference call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question and answer session. If you would like to ask a question during this time, simply press Star, then the number one on your telephone keypad. If you would like to withdraw your question, please press the star, then the number two. Thank you. Mr. Moon, you may begin your conference.
Good afternoon, and thank you for joining us today. With me on the call are Louis Têtu, Coveo's Chairman and Chief Executive Officer, and Brandon Nussey, Coveo's Chief Financial Officer. Before we get started, I would like to note that certain statements made during this conference call are forward-looking statements within the meaning of applicable securities laws, including those regarding our plans, objectives, expected performance, and our outlook for the Q3 and fiscal year 2024. These forward-looking statements are given as of the date of this call, and while we believe any statements we make are reasonable, they are based on current expectations, which are subject to risks and uncertainties, and actual results could differ materially. We do not undertake or expressly disclaim any obligation to update our forward-looking statements, whether because of new information, future events, or otherwise.
Further information on factors that could affect the company's financial results is included in filings we make with Canadian securities regulators, including under the section titled Risk Factors in the company's most recently filed AIF, and the section Key Factors Affecting Our Performance in the company's most recently filed MD&A, which are available under our SEDAR+ profile at www.sedarplus.ca. Additionally, some of the financial measures and ratios discussed on this call are either non-IFRS measures or operating metrics used in our industry. A discussion on why we use these results, and where applicable, a reconciliation schedule showing IFRS versus non-IFRS results, are available in our press release and our MD&A issued today, which may be found on our investor relations website at ir.coveo.com and our SEDAR+ profile. Please note that unless otherwise stated, all references to financial figures are in US dollars.
Lastly, slides accompanying this conference call are available for viewing and accessible on our IR website under the News and Events section. I will now turn the call over to Louis to begin. Louis?
Thank you, Paul, and thank you all for joining us today. Starting off with the financial results, I am pleased to report that Q2 SaaS subscription revenue, excluding the impact of Qubit churn, grew 19% year-over-year. Additionally, our adjusted operating loss for the quarter was $1 million, a significant year-over-year improvement from $4.7 million for the same quarter last year, and was ahead of our guidance for the quarter. We are incredibly excited by the promising early results and customer interest from our generative AI rollout program, and we're thrilled to announce Xero as our first customer to go live with Coveo Relevance Generative Answering on their Xero Central customer service portals globally.
We believe this, combined with the building momentum of our partnership with SAP and overall interest in our AI platform, creates a substantial opportunity for us in the mid to long term and expect a positive impact on bookings in the coming quarters. Generative AI has been, for Coveo, a natural extension of the capabilities we have built over the last decade in our platform. Having said that, following the release of ChatGPT, we have observed so far this year a landscape where enterprises have been grappling with the hype surrounding generative AI and have been carefully assessing and evaluating its practical applications, along with its many risks, including security, factuality, accuracy, and cost. With that hype came an onslaught of marketing messages and pitches that enterprises have been forced to sift through to ascertain what is reality and what is actually worth buying that will drive real business outcomes.
As a result, many customers have been in evaluation mode as they further refine their strategies around generative AI and have been cautious about entering into new deals, especially in light of the ongoing macroeconomic headwinds that have more broadly impacted customer spending on software. While this has, of course, been disruptive to sales cycles, I believe this ultimately plays to our advantage, as Coveo is one of the few companies on the planet with a decade of experience in AI, helping large enterprises realize superior outcomes at scale in a secure, reliable, and trusted way. We believe we're starting to see the hype cycle around generative AI subsiding and believe the market has started to coalesce around four main areas where tangible ROI, even in the current macroeconomic environment, can be achieved. These areas are customer service, knowledge management, commerce, content creation, and co-piloting.
In the first three, we believe we are extremely well positioned to be successful through our unique ability to solve the many challenges of leveraging generative AI in enterprise use cases. Our approach to creating an enterprise-grade generative AI solution has always been guided by our last to hype, first to results principle. And as I'll talk about shortly, we now have the results to justify the hype. So in summary, we are very bullish on the underlying long-term growth drivers of Coveo, in spite of what we believe are near-term macroeconomic headwinds. As I mentioned, we're extremely excited about the early progress and rollout of our generative AI program, and I'm pleased to announce that we're on track for general availability of Coveo Relevance Generative Answering for service, website, and workplace use cases in December 2023, a major accomplishment by all measures.
We also recently announced early access availability of Coveo Relevance Generative Answering for B2B and B2C commerce, offering customers an enterprise-scale generative answering solution for commerce. This innovation empowers customer experiences with AI-driven question answering and advisory capabilities, fostering customer engagement and product knowledge discovery in e-commerce. Coveo's generative answering solution is designed to enhance the digital storefront experience by providing a conversational way to share product information with customers. We signed our first 5 generative AI transactions in our fiscal Q2, including one in commerce that I'll talk about later, with a substantial number of additional deals expected to close in our fiscal Q3 and Q4. This significant customer interest in generative AI and our platform more broadly, resulted in another good quarter for pipeline generation, with commerce and our partnership with SAP in particular, continuing to represent a significant portion of our growing pipeline.
Most excitingly, while we have been articulating the value that we believe Relevance Generative Answering can bring to our customers, we finally were able to see the benefits via A/B testing. I am excited to share that the early results were far better than our expectations, with customers seeing case deflection improvements in excess of 20%. This would equate to significant cost savings, especially when considering the scale of most of our enterprise customers. Importantly, these savings are in addition to the cost savings already realized by our customers through the use of our AI platform. Being able to show concrete and tangible improvements in business outcomes, like case deflection and customer service, at the end of the day, is what our customers care about. It was the one question we still had as we began providing product access to our design partners and advisory group customers.
Now, with this data, we have the answer. This is the most significant and massive advancement in self-service we have seen that can significantly improve self-service, case deflection, resulting in lower costs and higher customer satisfaction and loyalty. We believe this will result in a tectonic shift of value from the contact center into intelligent self-service, an area where Coveo is also uniquely positioned. As I mentioned, we had our first customer, Xero, go live with Coveo Relevance Generative Answering to enhance its customer service for its 3.7 million business subscribers. We believe this is one of the first examples in the market of a company deploying generative answering for self-service use cases at scale globally.
This enhanced solution aligns with Xero's vision of delivering real-time conversational answers, leveraging Coveo's unified index for secure and controlled access to information, combined with our AI relevance stack that understands a user's context, complete with citations for traceability to sources of truth. By offering personalized and generative self-service journeys, Xero helps their subscribers to independently manage complex tasks, freeing up their contact center's capacity while delighting more customers. In the coming weeks, Xero intends to implement generative AI for internal employee support and knowledge management. Xero's vision is to streamline and enhance the overall digital experience across their organization, which encompasses approximately 4,000 employees worldwide. This comprehensive approach underscores Xero's ongoing commitment to improving interactions with both customers and employees, and we at Coveo are thrilled to play a central part in this journey.
Another compelling example of the potential of generative AI for Coveo within our commerce line of business is our collaboration with a major U.S. home improvement retail chain. We successfully created a proof of concept that can connect their extensive internal product documentation to their product catalog, resulting in a rich, guided discovery and advisory experience for their customers. This groundbreaking capability showcases how shoppers can engage with in-depth product exploration much earlier in their buying journey while maintaining their focus on the retailer's website, removing the need to navigate to external sites. ... This achievement not only exceeds our initial expectations within commerce, but also shows our ability to rapidly implement new capabilities and applications of generative AI, thanks to our single AI platform. We are thrilled to continue innovating with this customer to further optimize results.
Additional generative AI wins with our customers included a leader in high-performance silicon chip design, verification, IP integration, and software security, a leader in spatial data solutions, and a global enterprise software company focused on project management for customer self-service and agent proficiency use cases, respectively. We believe these early wins are a great start as we continue to penetrate what we believe is a very significant opportunity, both within our installed base and with net new customers. In addition to our generative AI successes thus far, we also won a new commerce customer that is a leading global B2B distributor of connector and sensor products. This customer shortlisted Coveo and one other competitor to replace their existing homegrown search solution, ultimately decided to partner with Coveo because of stronger product capabilities and our ability to handle large and complex catalogs.
Their goal is to make significant improvements in the digital experiences they provide to their customers to maintain their leading market position. We also secured a significant new customer win with a major Australian retailer for their next-generation employee portal, marking another important milestone in the ANZ region. The customer was looking to address critical gaps in their usage of Salesforce to improve employee satisfaction and loyalty as part of a key C-level initiative. Our belief is this is the first of many other opportunities we will have to expand our relationship with this large customer. As shown on the slide, we made several announcements during the quarter that highlight the strength of our AI platform, developed over more than a decade, demonstrating our commitment to ongoing and high-velocity innovation.
Subsequent to the end of the quarter, we announced more than 15 additional innovations to our AI search and generative experience platform to further enhance digital experience across our 4 lines of business: website, commerce, service, and workplace. These advancements solidify Coveo's leadership position in powering billions of individualized, trusted, and connected digital experience interactions on a global scale, enabling enterprises to excel in an ever-evolving digital landscape. I'm especially pleased with the remarkable speed at which, in collaboration with our design and advisory partners and the strength of our R&D team, we have successfully established a truly enterprise-ready generative AI solution.
It has been approximately seven months since our announcement of Relevance Generative Answering, to now having our first customer live with the solution, and Coveo is one of very select few vendors to have any customer live with a generative AI solution that we believe generates a substantial and tangible ROI. This achievement, built on our robust AI platform, enriched and refined over a dozen years, offers customers a powerful technology to drive the new business-to-person imperative through semantic search, AI recommendations, and now generative experiences that are fully secure and accurate. We know what it takes for enterprises to gain a trusted AI experience advantage, and this places us at the forefront of this journey, where search, recommendations, generative answering, chat, conversations, and personalization all converge.
This completely redefines a new, much more powerful digital experience paradigm, and we are proud to see Coveo at the forefront of that innovation, already delivering results at scale with some of the leading brands in the world. With that, I will now hand the call over to Brandon to discuss our quarterly results in more detail. Brandon?
Thanks, Louis. It's obviously a very interesting time for the business, with several moving parts to it at the moment. So to try and succinctly summarize the most salient points, first, we delivered 19% SaaS subscription revenue growth for the quarter, excluding the impact of Qubit churn, which was slightly ahead of our guidance. Customer retention and expansion remained solid, with NER for the quarter at 111% outside of Qubit-related churn, indicating customers continue to buy more from us. We delivered our second consecutive quarter of positive cash flow from operations and are well ahead of our plans on adjusted operating loss and for ongoing cash flow positivity. This, combined with our strong balance sheet, keeps us well-positioned in this fast-evolving market. And importantly, we continue to believe that we have multiple opportunities on hand to re-accelerate our revenue growth rates.
The first of these, of course, is our Relevance Generative Answering solution. We continue to see customer interest building for this product and believe we are one of a small number of companies that are actually beginning to monetize generative AI within the enterprise segment. We remain positive in our outlook for this offering, and in the quarter, signed our first five customer order forms, with further signings in the current quarter. As a reminder, we've always said this will be primarily a fiscal 2025 growth driver as customers evaluate the offering and finalize their purchasing decisions. Second is our opportunity inside of commerce, and more specifically with SAP as an endorsed partner.
We are similarly seeing a lot of promising signs of progress, but again, I've said this will be a fiscal 2025 growth driver, given the work that goes into enabling their sales team and overall sales cycle lengths. Hopefully, many of you will attend our capital markets day next week, where SAP CX's Chief Operating Officer will be present to highlight to you firsthand how they're thinking about this partnership. While these growth drivers give us optimism, in the immediate term, you've heard from Louis that customers are being forced to dissect the noise in the market around generative AI while making their purchasing decisions. This affected sales cycles in the near term, and we've seen our typical pipeline conversion metrics and deal cycle lengths elongate.
As we look into the back half of the year, we are seeing signs of customers resuming and finalizing their purchasing decisions after this period of disruption, and our pipeline of deal activity is encouraging. However, we will remain cautious in our outlook, as I'll get into shortly. We've also discussed Qubit churn in the past, and as expected, this churn is materializing. In aggregate, it is unchanged in estimated amount, but a portion of it did occur earlier in the year than previously estimated. And in the meantime, you can count on us to be efficient capital allocators. We have retired almost 5 million shares via share buybacks, and as mentioned, are tracking well ahead of our targets on reaching cash flow positivity. We are improving our AOL guidance for the year, even with a cautious stance on revenue.
The leverage we're seeing in the business model is encouraging and gives us confidence that when macro conditions recover, we'll have plenty of levers to pull to lean into our growth areas. Now, let's get into some of the details. For the Q2, SaaS subscription revenue was $29.4 million, an increase of 15% year-over-year, and total revenue was $31.2 million, growing 12% year-over-year. Excluding the impact of the legacy Qubit churn I mentioned, SaaS subscription revenue grew 19% year-over-year. In Q2, we observed both land and expand bookings across all of our lines of business, and consistent with previous quarters, all four of our lines of business continued to grow double digits year-over-year on an ARR basis.
Overall, ARR growth was largely in line with our SaaS subscription revenue growth, both including and excluding the Qubit churn. Generative AI bookings are off to an encouraging start with our first 5 customer order forms signed in the quarter. While we will not start to recognize revenue from these bookings until the earlier of general availability, which is expected in December 2023, or when customers go live with our solution, I view it as a positive endorsement that these customers are choosing to sign order forms with Coveo in this hot market segment. Our net expansion rate as of September thirtieth was 106%, compared to 109% we reported in our Q1 results.
When excluding the legacy Qubit churn, our NER was 111% for the Q2, a positive sign that more of our existing customers are finding success with Coveo and are continuing to buy more from us. Our Q2 gross margin improved 78%, compared to 76% for the same period last year. Adjusting for share-based and other payments, adjusted gross margin was 79% for the Q2, an increase of 1% compared to a year ago. Product gross margin was 82% in the quarter, consistent with the prior year, and our adjusted product gross margin was 83% in the quarter, also consistent with the year ago period. Continue to view this as a very important leverage point in our business model that sets us up well for strong profit margins in the mid to long term.
Adjusted operating loss for the quarter was $1.0 million, a significant improvement compared to $4.7 million a year ago and well ahead of our previous guidance due to ongoing strong expense management. We generated approximately $800,000 in cash from operations in the quarter and $1.8 million for the six-month period, which is also well ahead of our plans. The non-adjusted basis operating loss for the quarter includes a $3.2 million non-cash impairment charge for the customer relationships recognized at the time of the Qubit acquisition. With the customer attrition realized in the quarter as a result of our decision to deprioritize this non-core capability, we concluded a write-down was appropriate based on updated valuation models. We ended the quarter with $168 million in cash and no debt.
With ongoing operating cash flow positivity clearly in sight, we are well positioned to navigate the macro uncertainty and capitalize on our market opportunity. So finishing with guidance, we continue to see positive indicators that enterprises are viewing our space as a priority, given the enthusiasm for AI and its potential impact on business outcomes. However, as you've heard, customers have been forced to dissect the immense hype around Gen AI and have had to take the time to diligence their purchases and select their AI partners. Of course, while our market is a priority for many enterprises, we are also not immune to the macro environment and its overall effect on enterprise software spending. This has delayed deal cycles in the first half of our year.
As we look into the back half of the year, we're encouraged by the deal activity we are seeing, and we're seeing signs of customers and prospects finalizing their decision-making and moving through their purchasing decisions. We're optimistic with our position in the market and take the early order form sign for our generative answering solution as a positive indicator of this. However, these are early signs, and we would like to see this materialize further. Our revenue guidance also reflects the updated timing of the Qubit-related churn, which occurred earlier in the year than previously expected and impacts the amount of recognized revenue for the year. Further, we are seeing a greater portion of our bookings coming via expansion in this market, which typically carry lower professional services revenue than new customer bookings.
So with that all said, we're now expecting to land at the bottom end of our previously issued guidance for SaaS revenue for the year, and with a conservative stance, are updating our SaaS revenue guidance to $117 million-$118 million. For the Q3, we expect SaaS subscription revenue to be between $29.1 million and $29.6 million. Approximately 60% of the expected Qubit churn for our fiscal 2024 occurred in the Q2, which will have a negative impact on our sequential quarterly revenue growth for our Q3. Excluding the legacy Qubit churn, we expect our core growth rate on SaaS subscription revenue to be approximately 15% for the Q3.
Reflecting the lower estimated services revenue for the reasons mentioned earlier, total revenue guidance is being updated to $124.5 million-$125.5 million. For the Q3, we expect total revenue in the range of $30.9 million-$31.4 million. Due to our ongoing efforts on efficiency, we are improving our guidance range for adjusted operating loss to $9.5 million-$10.5 million, compared to the previous range of $11.5 million-$13.5 million. For the Q3, we expect adjusted operating loss of $2.5 million-$3.5 million.
We now expect cash used in operations of less than $5 million in fiscal 2024, compared to the $10 million stated previously, and reiterate our expectation of achieving ongoing positive cash flow from operations for the full year of fiscal 2025. Finally, both our Q3 and fiscal 2024 guidance assumes FX rates roughly in line with where they are today for SaaS subscription and total revenue growth, as well as adjusted operating loss. With that, operator, you may now open the line for questions.
Thank you. Ladies and gentlemen, as a reminder, should you have a question, please press the star followed by the one on your touchtone phone. You will hear a three-tone prompt acknowledging your request. Questions will be taken in the order received. Should you wish to cancel your request, please press the star followed by the two. If you are using a speakerphone, please flip the handset before pressing any keys. Once again, that is star one should you wish to ask a question. Your first question is from Doug Taylor from Canaccord Genuity. Please ask your question.
Yeah, thank you. Good evening. Congratulations on securing the first orders for your generative AI product. With that, I'd like to hopefully revisit the discussion on the pricing dynamics that you're establishing around that module, whether there's anything more you can share with us at this time about how that's priced relative to your core enterprise search solution.
Right. This is Louis. Hi, Doug. Thanks for the comments. The pricing, we went out initially with pricing that was in line with the industry, the first movers charging a percentage. And so that percentage, and we published it, was 40%, which was obviously well accepted, given that we've been able so far to land already 5 orders. And that pricing is 40% currently with a 6-figure minimum. So it's considered, you know, for early orders to be pretty substantial. That being said, Doug, we are looking at further refining pricing as we move into production from our early beta adoption program.
Especially given, you know, both the cost and the value, especially in particular that we generate, we expect that pricing to go up.
Okay, that's helpful. Perhaps, similar or along that same vein, given you've now got a customer live with your Generative Answering function, can you tell us a bit more about the onboarding process? Anything that you have learned about how, you know, customers, once they've signed, what the adoption curve looks like relative to your core search solution?
I would say, Doug, that the adoption and both the adoption and the speed and the results were beyond our expectations.
Obviously, we've marketed the solution initially. The early adopters were customers of Coveo, which already had the Coveo infrastructure, the secure index, and all the relevance AI stack in place to feed generative AI. And we've been clear in prior calls that this obviously puts Coveo in a very unique position to take advantage and for customers to take advantage of that technology. But we've, you know, while we had initially anticipated months in terms of of adoption, the customers themselves have been the ones probably putting more pressure on us to accelerate the rollouts, and in some instances, the global rollout of the solution. And the results, as you know, in our space, are done through AB testing.
So we direct part of the traffic to the legacy solution, and we direct part of the traffic to generative answering, in this case, generative AI. The initial use cases were predominantly in customer service and self-service. So we've been able to achieve that and close the loop within a matter of weeks. Closing the loop, meaning being able to actually measure consistent results, and gains. And as I said earlier, and as Brandon mentioned, these were beyond our expectations.
Maybe one last question for me then. Having, you know, the return on investment from your customers being higher than you'd previously thought, would suggest that, you know, the usage and the utilization of the generative large language model function is higher. Yeah, can you speak to any observations you'd have about the cost to Coveo for providing that solution versus your earlier expectations now that you've got the customer live?
Yeah, that's a very good question, and we think will put us at a competitive advantage as well from a cost perspective. So not only we are in a position to deploy generative AI, that is, you know, that generates based on current data, thanks to the index that is secure, and that is accurate and well-grounded and fully traceable. But also, the way we engineer this obviously is goes a long way in terms of optimizing the cost. Generative AI, and the cost of generating an answer can be as much as 100 to even 1,000 times higher than running a query engine, if not engineered properly.
The ability of Coveo to provide, you know, the stack of software that isolates the secure, current and relevant corpus of information in order to feed the large language models goes a long way in terms of optimizing the cost. And right now, you know, although we don't have any completely accurate model of the cost as we're learning a lot from the use cases, we already know that the cost part is well under control and within the range, the zip code of the type of gross margin that we normally produce.
All right. Thank you for confirming that. I'll pass the line.
Okay.
Thank you. Your next question is from Thanos Moschopoulos from BMO Capital Markets. Please ask your question.
Hi, good afternoon. Louis, with respect to the two factors you referenced prompting a slowdown, you know, the macro, and then customers also pausing to evaluate all the various options for gen AI. May be a hard thing to answer, but, you know, would you say one of those factors is having more of an impact than the other? Are they kind of similar? What would be your thoughts on that?
I think for us, thanks for the question. I think for us it's really been... You know, the macro is the macro, and to the extent, however, that we can provide solutions that generate a high ROI, that we can demonstrate, there are ways around it. I mean, you know, the main effect of the macro is more scrutiny in the deal conversions and so on. Sometimes, you know, we say we don't lose deals, they kind of get delayed. The other one, however, was more interesting, is that, you know, there's no question that generative AI and, you know, when ChatGPT came out, and I always joke by saying, you know, when people started writing poems on their iPhone in December and January and got all excited about generative AI.
You know, when that got into the enterprise, enterprises, it took everyone by storm. It took the world by storm and everyone by surprise, and enterprises had to distill essentially how that would be applied in which use cases, and really understand the implications from an IT and data governance perspective, security, and etc. Frankly, also the cost and really understand the true benefit. Coveo is one of the first companies that has closed the cycle, and while this has sort of stalled a lot of decisions, you know, since the beginning of the year.
We're now seeing as we can demonstrate results, obviously, and we can demonstrate to CIOs that, you know, we can run that in a way that is enterprise-grade and fully compliant and secure, we're seeing that level of activity, you know, these companies reengage and convert, especially as they see the kind of benefits that they can get. And so basically, you know, we expect the bookings. We've seen that. We're seeing that in the pipeline activity, and we expect the bookings to reaccelerate and with an impact on the back half of the year and obviously fiscal 2025.
That's great color. Thanks. And then on the SAP channel, you alluded to the fact that, you know, it's kind of in the process of ramping. Obviously, a lot of, you know, training, I imagine that has to be done with respect to the sales force. But maybe could you update us in terms of where that process is?
Yeah. So we announced the global endorsed deal, which is a big deal, obviously, with SAP globally, if you recall, at our earnings call for last fiscal year, basically in April, beginning of May. And this deal, this deal happened really in April, so we started rolling it out and so on. You know, the sales cycles, as you know, and that's mainly on the commerce side, right? The sales cycle really are no less than six-to-nine months.
So, as we went through that period in the summer period, and we're now rolling out more and more enablement, we've seen, you know, that pipeline continue to increase, and we're obviously in deal activity, you know, that is increasing as well right now. So for us, it's sort of a the normal cycle, as you pointed out, of deploying the enablement and, you know, given that it's a global deal, it's a rather large operation that spans through, you know, Europe and North America, mainly for now, and parts of Australia, New Zealand.
So we're into it right now, but we're clearly seeing the growth and commerce as a solution continues to be our biggest growth opportunity, but also now with Relevance Generative Answering as well. But obviously, SAP is a big part of the commerce site.
Great. Thanks, Louie. I'll pass the line.
Yep.
Thank you. Your next question is from Suthan Sukumar from Stifel. Please ask your question.
Good evening, Louis, Brandon.
Good evening.
Congrats on some of the early initial progress here on the year, Gen AI trial. Just kind of looking ahead to the anticipated commercial rollout, just curious, can you share some details on what that might look like, you know, in terms of if you'll be focused on existing customers beyond some of the early adopters in your base, or will you be looking at driving more net new business? Just curious what your priorities are from a go-to-market perspective.
Right. So initially we announced, obviously, that we would come out with the product, if you recall, sometime around the April timeframe. We actually demoed the product initially to our customers in June in California, and then announced our beta program that started later that summer, and are gradually rolling out. We previously announced that we have approximately 45 companies. There's actually more than that aligned right now. And initially, our initial rollout was with existing customers because those are customers who already have, you know, used the large Coveo indexes across their enterprise, already have the data and et cetera already.
And for us, rolling out generative answering as an extension of that infrastructure was a natural, and so this is why we went there. There's no question now that Relevance Generative Answering is a part of our platform. You know, our platform is a platform that's made of numerous AI models and behavioral machine learning, deep learning, large language models, and now generative AI. And so that's an addition that obviously we also promote now to new customers as well. And we expect to see some of those orders also from new customers from this type of functionality in the quarters to come.
Okay. Okay, great. And just to follow on to that, you know, when you look at your pipeline today, how much of that growth can you attribute from Gen AI and from net new customers specifically? And we think about, you know, call it the LTV, the CAC opportunity, you know, for new customers. How does that compare to, you know, what you've seen today with your existing base?
...Right. So we don't see any pressure, you know, given the increase in value and functionality, we don't see, and we don't anticipate any downward pressure despite the economy on the deal size, first of all. So the value of our solution keeps increasing and the breadth of the functionality keeps increasing, and we can connect that functionality to the value. So that's number one. What we're seeing is obviously more competitiveness for the overall platform.
You know, Gen AI is not a separate thing, and if there's anything that we're trying to tell the market, and that frankly, customers are realizing now, as many of them rush to Gen AI as a separate thing, they have realized, and we're here to show that, that in fact, Gen AI is a natural extension. You cannot activate generative AI, at least not in the types of use cases that we do, which are digital experiences, commerce, websites, customer service, you know, workplace type of applications, and mostly, you know, real-time digital experiences.
You cannot activate generative AI, unless you master the science of data connectivity, unless you master the science of relevance, in fact, and understanding users at the other end in real time, and their intent, and their context, and their behavior, which has always been the forte of Coveo. So what I want to say here is that relevance generative AI is a natural extension to that platform. And so the way to think about it is not so much generative AI only as an incremental product, which today it is for our existing customers, but really as an integral part of our platform that becomes suddenly more valuable and more, more differentiated, more competitive in our markets.
And so we expect more and higher bookings as you know, you know, bigger deals, more deals, as a result of that, I guess.
Okay. Okay, good. Thank you, thanks for the color, Louis, and congrats on all the early progress. I'll pass the line.
Thank you very much for your kind comments.
Thank you. Your next question is from Kevin Krishnaratne, from Scotiabank. Please ask your question.
Hi, thanks for taking my question. Maybe just in terms of the change to your full year guidance, I just want to clarify if there was any increase to the Qubit churn that was previously indicated in the prior quarter, if that has remained consistent?
The aggregate amount is consistent, just the timing of it. We did see it start to come in a little earlier in the year than we otherwise had planned. So the impact on revenue recognized, we lost a couple of months in some cases, but an aggregate amount remains the same.
Okay, that's great. And so then you've mentioned obviously the five orders for Relevance Generative Answering in your release. And then maybe in terms of the guidance for fiscal 2023 revenues, just what's implied for Relevance Generative Answering in that guidance? Is it those five deals or is it looking ahead a little bit as well?
Yeah, I mentioned in, in my comments there that of these five initial orders, and there will be more, that the way we structured these agreements is that revenue recognition will begin once general availability is met. And as you heard from Louis, his prepared comments, that's expected in December. So, good news is we're seeing order forms flow, you know, for enterprise to put a signature on an order form. You know, they've done their diligence and run it through their processes. So that's the positive news, but it's not gonna really contribute a whole lot into our current, our current year revenue. It's more of a fiscal 25 driver for us.
Okay, no, that's great. And then you'd mentioned last call, some large software firms that were asking for about a 50% uplift in pricing for copilot type of GenAI, AI solutions. And I think you'd also mentioned, on the order of a 1,000 times the cost of a query to run an LLM answer, over your search index or over an index search. So you'd provided some several strategies on how Coveo would work with customers to help on cost efficiencies for these firms, including using different LLMs and several other strategies. Could you maybe help us to get a sense of like, the order of magnitude of cost savings that your solution might offer against these other competing solutions on a rough level?
Well, I don't know that we can comment on the competing solutions, because we don't know of many that are live. So that's number one. You know, we do not see in the use cases that we run the scale, anything that's at scale like Coveo that is in full production in enterprises globally. That's number one. So it's hard to compare. However, let me give you some color on this. What we said before and what we repeat here is that generative AI can be very costly, if not engineered properly.
So if you're trying to generate, you know, an answer from an extremely large corpus of data, and an extremely complex prompt with extremely complex prompt engineering, there's a very high cost to that. So what we're able to do, given the platform that we already have, that again basically is based on large-scale, high-performance enterprise indexing, combined with the ability to, which we call relevance, the ability to understand in near real-time who's at the other end, we're able to narrow the problem in a way that is extremely cost-effective. So relative to the gains, we've been able to measure in excess of 20%, as we said in the earlier call, in excess of 20% call deflection, with some customers.
And so what that means essentially is if you run a contact center, by applying generative AI, you know, and that 20% was on top of Coveo, by the way. You know, if you run a contact center by using generative AI, you suddenly have 20% less cases. So think about the economic benefits of that. So we think the costs are gonna be not negligible, but certainly reasonably low relative to the value, the economic value that we can generate. And again, as I said earlier, very much in the vicinity, zip code or the range of the COGS that we're used to. As you know, we currently run at about, what, I don't know, last quarter off the top of my head, but roughly around 82% product gross margin.
And, so we think we can, we can pretty much stay in that neighborhood.
Okay, that's great. I'll pass the line. Thanks.
All right. Thank you.
Thank you. Your next question is from David Kwan from TD Securities. Please ask your question.
Hey, guys. So I was just curious, you guys talked about, I guess from the A/B testing, it took probably about a few weeks. So when you look at when the customers, I guess, started the beta testing to moving ahead to at least deciding to go ahead with live production, it sounds like, you know, probably about a month or so. Is that about right?
It's, yeah, it's in that range, a little more than that, but it's in a matter of a few weeks. Yeah.
Do you think-
Single-digit weeks.
Single digit weeks. Perfect. And I guess going forward, though, as you maybe get more data on whether it's A/B testing or, you know, maybe get some actual kind of tangible ROI, do you expect those timelines to actually, improve?
Yeah, they'll, they'll probably be about the same. What happens is... Well, it, it depends whether we're, we're deploying a new customer or an existing customer, obviously. But it'll probably be about the same. What we need to understand here is that Coveo customers are all large. We deal with large, large-scale enterprises, and the common characteristic of our customers is they have a massive amount of content and variety of content and- or- and/or products that they need to deliver to a very, very large and very diversified audience, sometimes across the world. And so we're seeing volumes here that basically converge very, very quickly.
You know, when you have hundreds of, hundreds of thousands of visitors a day, that in some, in some instances, you know, you can very, very quickly understand through A/B testing, you know, as you run 20, 30, 50% of the traffic through a technology and, and you can compare. So you don't need weeks. Now, what's fun, what's interesting and, and fun for us right now, is that as customers have kept using the solution over the weeks, we're seeing the same, the exact same pattern. But the... I guess the timeline, you know, in scientific terms, David, you know, the timeline to gain a statistically representative sample and, an A/B test is not, is not long.
No, that's helpful. Thanks, Louis.
Right.
On commerce then, are you guys also applying to, you know, a beta program like you did with Service Workplace and Website? I know you've obviously you talked about one customer that's already made a purchase order here, so I wasn't sure if you were looking to doing a broader, you know, beta program for commerce as well, for that B2B, B2C offering.
Well, we have not launched one yet for commerce. We might actually be able to go into early production. But I mean, you know, to a degree, we always launch large new features in a form or another of a beta process. But you know, what we're doing in commerce with generative AI has some commonalities with what we're doing elsewhere, obviously. You know, except that it's applied to product discovery and constrained, you know, by the data of the catalog and things of that nature.
So the short answer is, we haven't announced a beta program in commerce, but we've already signed, as we said, one order, and we expect more because the solution is quite innovative and will, you know, should prove to be extremely valuable very quickly.
Great. One last question here: Are you finding, I know it's a tough, you know, enterprise spending environment here, but are you finding that CRGA is maybe helping you progress conversations with, with new customers? You know, is it, is it making a difference here in terms of what's going on in the current climate?
Definitely. Definitely. And I think you got to look at it both ways. If you're in our space, you know, look, we're an enterprise search vendor. I don't think of many industries that have been more disrupted by generative AI than search when you think about it, right? So this is sort of what Brandon and I explained, you know, during the call, that customers, you know, in the earlier part of this year had to sort of distill the hype from the reality and the use cases, and how they could apply that and without getting in trouble.
But the reality nowadays is if you're in search, if you're in the business we're in, which is, which is essentially digital experience, personalization, semantic search, AI recommendations, personalization engines, and that, you know, you just cannot have a conversation now in November of 2023 without talking about generative AI. It would be like if you didn't live in the same era. And, and so now the good news is that, you know, generative AI is, at the same time, a catalyst for, I would say, a new wave of interest for search, for topics like knowledge management within enterprises, because you're talking about, you know, self-service proficiency. Think about a bank, for instance, and how financial advisors will use generative AI, you know, to gain proficiency on their own so much faster.
You know, with generative AI, you can literally ask a question such as: How do I activate a mortgage in Fort McMurray for commercial property that has environmental issues? And get an answer on that. Or what is the difference between a TFSA and an RRSP, and can I have one for my 18-year-old job? And what's the tax impact? So, you know, think about the impact of that, even in the workforce and so on. So, you know, it's very topical. And of course, it's driving more and better conversations, and especially if, like Coveo, you can show it and show that you can deploy it and generate results and quantify those results. It, we think that's promising.
That's great. Thanks, Louis.
Thank you.
Thank you. Your next question is from Koji Ikeda, from Bank of America Securities. Please ask your question.
Hey, Koji.
Hey, how's it going? This is George on for Koji.
Okay. Hi, George.
Hey, how's it going? You know, with you know, some of the early adopters of the you know, generative answering feature, have you noticed any commonalities among them, or is this kind of more broad-based interest among the customer base here?
I think the commonality is within the spectrum of a certain set of use cases, right? You know what, what's really interesting, you know, if you think about generative AI, think about it in three buckets, right? The whole and two of them we don't do, by the way. So the whole copilot code generation area, right? That's one area, we're not in there. Number two is content creation. So whether it's images or symphonies or for the most part, text, you know, summarization, things of that nature, you know, account summarization and, you know, generation of marketing content and things of that nature, that's content creation. That's not something we do.
The third area is in the digital experience, and, you know, in real time, you know, when you have someone online, on a website, in a commerce experience, on a customer self-service experience, you know, or an agent in a contact center, or, you know, plain and simple, you know, as I said earlier, a financial advisor or an employee, you know, in the workplace. So this is the area where we're in. So it's really the commonality is among the use case, but I would say the common attribute is: How do you make, how do you increase the self-service proficiency? How do you make that user smarter, faster, about what is it that they're doing, what is it that they wanna buy, or what is it that they wanna fix?
From that perspective, it is a revolution. And in my personal career, I've never seen anything like that in terms of the magnitude of the quantum leap that it creates for a digital experience. So I hope that answers your question.
Very helpful.
Thank you. Your next question is from Paul Treiber from RBC Capital Markets. Please ask your question.
Thanks very much. Good afternoon. Just a couple quick questions. Just on the pilots that you're conducting with the customers, and it's good to see the number of them have converted the, and the data on the AB testing. Could you speak to, you know, any alternatives that they were evaluating your solution against? You know, is it other competitors? Is it homegrown? You know, and, you know, how did that sort of comparison go?
It's Paul, it's really initially build. Every company went out and, you know, was looking to build something. And doing early tests in the area of generative with technology or technology of a method that we call Retrieval-A ugmented Generation or RAG. Which is essentially, you isolate a corpus of data, you generate from that data, from that corpus, what's called embeddings, which you store in a vector DB, and then you engineer a prompt to a large language model. What companies realized in doing that is that, you know, they needed to integrate that with data that is current, data that is secure, and in a way that is across their enterprise.
So it's a little more complex than working from a single knowledge base. And that also search channels, conversations, chats and navigation, disambiguation of results, et cetera, they weren't going away. So would you imagine a user going to a page where you would have a search box, a question answering box, and then a chat bot? So ultimately, you know, obviously that becomes very convoluted and clogged. And so users realize that it caused some... It posed some very significant security, currency, issues with veracity. Big brands can't hallucinate. So for all those reasons, a lot of these projects didn't turn out. They were experiments, but they didn't turn out to be very successful.
Against a build, obviously, Coveo comes with a solution that integrates the data component, the security component, the veracity, traceability component, and the data currency component. And so that we think that becomes very, very attractive versus to build your own solution. Especially that we can do it large language model agnostic, although today we use GPT, OpenAI GPT service on Azure pretty much, but we have side by side other technologies as well in the lab, that we think down the road, we can interchange depending on the verticals and the application and so on. So we're keeping it very flexible.
Then just one last one for me. You know, it sounds like generative AI is expanding the addressable market just given the potential productivity savings. What do you think happens to traditional enterprise search? Does it become merged or intertwined with generative AI, or do you see them as-
Totally.
- separate use cases?
No, totally. Search results, you know, you can think about commerce or content. People don't just want an answer, they want to be able to navigate a corpus of content. They wanna be able to disambiguate, they wanna be able to narrow down the set of results or the corpus of data within which they wanna generate an answer. And frankly, they want an integrated experience.
You know, as I said earlier, you can't have a digital interface, whatever it is, whether it's an intranet or an employee or a dealer portal or a patient portal or a customer portal. You can't have a page where you would have a search box on the left-hand side and a question answering box on the right-hand side. And you can't just deliver an answer, you know, people still want to control the experience and so on. So the reality is, you just said it, and many companies are realizing that, is the worlds of search, navigation, disambiguation of results, and answering, and frankly, ultimately, conversations and chats and personalization and recommendations are all converging.
All converging right now, and this is in fact the platform that we're creating is to bring that all together.
Thanks for taking the questions.
Hey, that was a very good question.
Thank you. Your next question is from Adhiraj Walia from Eight Capital. Please ask your question.
Hi, good evening, guys. Thanks for taking my question. Maybe just one on the commerce use case versus customer service use case for Relevance Generative Answering. Are you finding or do you think that, you know, a lot of the heavy lifting that you've already done on self-service will kind of transfer over, given that, Louis, you mentioned that commerce could potentially go straight to production as opposed to going through a, maybe a beta test?
Right. So there are differences, obviously, but you're in the world, what generative AI really does in our space, you know, again, we're not in content creation or the other things I described. But in our space, which is real-time digital experiences, what generative AI does is it moves the experience from a search-based experience and results-oriented and recommendation-oriented experience that, you know, if sophisticated enough, using AI, deep learning and behavioral-based machine learning is prescriptive in nature and anticipates and all of that, but still is limited to that. It moves it to an experience that becomes advisory. Should I use this or should I use that? You know, these are answers. These are questions that can be answered by generative AI.
You know, what is the difference between product A, product B, product C, and in what circumstances should I use product C, given that, you know, I live in the north and, I don't like this, right? You know, these are the types of questions that in the context of commerce, you know, and or workplace or so on, can be answered. Now, in commerce, there's a couple of subtle things, which is, you know, you wanna if you're, if you're, a large, retailer, let's say, you obviously wanna provide advice, but you generally would prefer to provide advice across products that you can sell and deliver.
So, you know, you wanna, you wanna be able to limit or manage, or control or tame, essentially, the answers and the large language model to be constrained by the catalog and all the specificities around that, and so on. So you wanna facilitate the consumer discovery towards the product that you wanna sell. You know, nobody wants to serve advice for you to go to Amazon and buy something, right? And so these are. But we already, we already are very competent in that. We already understand that. We understand recommendation engines, and we understand relevance, and we understand, you know, how to create a great consumer experience that will yield to stuff that you will sell, you know, bigger cart sizes and etc.
So from that perspective, I know it's a long answer, but I think it deserves it. You know, it's from that perspective, you know, we're applying generative AI to that specific problem, and we think that's probably one of the best, one of the most important problems in commerce to create a much richer experience.
Understood. Thanks. And then one quick one, and then I'll pass the line. Just on the SAP partnership, you know, it's been live for, call it, six or seven months now. Just on the pipeline that you're generating from the SAP endorsed partnership, are you finding that the deal sizes are changing from, when you first introduced it to now, as you kind of educate the SAP sales force? But just any changes in the pipeline from SAP, any color on that would be super helpful. Thank you.
Well, I'll give you an answer because, as you said, it's been going on for six months and so on, and you know, we're building the pipeline and seeing the deals. But until we see a larger number of deals, we'll see. But the general trend or pattern that we see is in line with SAP as a corporation. You know, SAP in commerce, if you think about Shopify, for instance, Shopify has, you know, a very large number of customers, but they tend to be... You know, and they have larger ones, but they tend to be, on average, fairly small.
Meaning that they have a large number of customers that, you know, on average, probably sell about $100,000 of gross merchandise value every year. And at an aggregate, Shopify probably processes today, I haven't checked, you know, last month, but in the range of about $150 billion a year of gross merchandise value across its customers. SAP has only a few thousand customers, but they process close to, you know, between $1.5 trillion and $2 trillion. And so by definition, by design, those customers are very large global corporations, often that operate in many countries, often that have hundreds of thousands of SKUs, and in some cases, you know, we've seen as much as 1 million consumers a day. And so I guess that answers your question.
So normally, you know, as this kind of technology generates more value, it commands a bigger price. Also because the consumption is higher. Yeah, sorry about that. Also because the volumes are higher, obviously.
Awesome. Thank you, guys.
Yeah.
Thank you. Your next question is Itai Kidron from Oppenheimer. Please ask your question.
Hi, this is Herschel on for Itai. Can you guys hear me?
Yes, Herschel. Hi.
Hey, Louis. So I, I just wanted to ask, in the early conversations that you've had with customers around generative AI, how are you finding them able to balance finding budget for these projects, given the tough macro environment? You know, are you having to see them maybe reallocate, budget from existing projects, or is this mostly a, you know, new budget category that customers are creating? And then just as a quick follow-up-
Yeah, go ahead.
Yeah, just as a follow-up, given the opportunity you're seeing, are you planning to make any changes to your go-to-market hiring for next year?
... Right. So, we're growing the hiring, obviously, in line with the pipeline. And so we're, as you heard Brandon talk about, our Q2, cash flow positive and all that. So we're sort of looking to balance everything, but we're really focused on growth. So as the pipeline grows, we'll continue to hire and to upgrade the team accordingly. As it relates to your first question regarding budgets, I would say it's pretty much all over the map, but in three categories. There are companies that just won't spend money right now because of budgets. Even if you hand them a bag of dimes, they basically won't give you a nickel.
You know, not always logic that just dictates that, but that's the reality. In our case, I think the more important story is we can demonstrate the ROI, and the ROI that can be sometimes very fast. So in generative AI for self-service, I mean, look, if I go to you and you have thousands of agents in a contact center, and I can say I can reduce, you know, the volume of cases by 20%, you will be able to translate the math. And by the way, it doesn't mean that you're gonna necessarily slash your agents by 20%. It means that, you know, because it might be the tougher questions, actually, that get to your contact center.
So there's a bit of a math there, but it's nevertheless very high ROI. In commerce, if you can do A/B testing that shows, you know, higher cart sizes and higher conversions, you know, you generally can justify these projects. So it's more in the justification. I would say, think about it as a two-by-two matrix. On the one axis, it's the ability to spend, which in some companies is just none, whatever, you know, you would show them the teleporter, they wouldn't buy it. But if they have the ability to spend, then the second axis is really the justification, the scrutiny of the justification. Only the high ROI projects typically make it through. And, you know, the good news is we're geared for that part.
Got it. Thank you. I'll jump back in the queue.
Great. Thank you, Herschel.
Thank you. Your next question is from Clara Gerdes from UBS. Please ask your question.
Oh, hi. Thank you. Yeah, this is Clara on for Taylor. So just a quick, quick one on the guide. I think if you look at it, it implies that the 4Q sub guide would be a step up in sequential growth, but the 3Q guide would be flat. So, yeah, is there anything you can share on what that step up might be, those drivers? I know you called out a couple of things that are impacting this quarter, but, yeah, anything you can share there?
Yeah. So for the Q3 or the upcoming quarter, the flat sequential guide that you're referring to, it's mainly driven by that Qubit churn. As I said in my prepared remarks, we digested the bulk of that churn, the majority of that churn in the Q2, which of course then flows through to rev rec into the Q3. And then thereafter, you know, we're through the heaviest part of the churn, and we can get back to growing again. Our focus is, of course, on making sure that bookings new bookings are leading the way. Revenue will always lag those, but so our focus is on that, on the new bookings growth. But that's the dynamic, that's at play on the reported revenue.
Okay, thank you. And then real quick on NRR, going forward, I know the adjusted, on the adjusted basis, it was 111%. How has that been trending? And is there anything, you know, you can give for us to, you know, look at that for the rest of the year expectations? Thanks.
Yeah, it continues to trend well for us. It's really a really positive indicator in a market like this, that our NRRs have held up the way they have. As we look forward, of course, you know, we'll be mindful of just the broader economy and things that are impacting enterprises everywhere. But we see opportunity to grow our customer base. You know, we've talked the bulk of this call about the generative offering and how that's increasingly applicable across the customer base, and that's certainly a wonderful opportunity for us to grow our NRRs. But to this point, yeah, they've held stable and strong through this economy, which is a great sign.
Awesome. Thank you.
Thank you. There are no more further questions at this time. You may proceed.
Okay. Well, thank you very much, everyone, for joining our call today. As you can see, we remain very positive around the positioning of Coveo for future growth. And with that, we would like to invite those of you who are interested to our Capital Markets Day next Thursday at the Toronto Stock Exchange. We hope to see you there. We'll have the Chief Operating Officer of SAP CX with us, talking about what we're doing together, as well as we'll go into much more details and present the customer use cases and so on. So we hope to see you there. And with that, thank you very much for being with us, and with that, operator, you can proceed and close the call.
Thank you. Ladies and gentlemen, the conference has now ended. Thank you all for joining. You may all disconnect.