Q4 2023 Chegg Inc Earnings Call
Participants
Dan Rosensweig; Co-Chairperson & CEO; Chegg Inc
Andy Brown; CFO; Chegg Inc
Josh Baer; Analyst; Morgan Stanley
Ryan MacDonald; Analyst; Needham & Company
Brian Peterson; Analyst; Raymond James
Alex Fuhrman; Analyst; Craig-Hallum Capital Group
Eric Sheridan; Analyst; Goldman Sachs
Brent Thill; Analyst; Jefferies
Presentation
Operator
Greetings, and welcome to Chegg Inc's fourth quarter 2023 earnings conference call. (Operator Instructions) As a reminder, this conference is being recorded. It is now my pleasure to introduce Tracey Ford, Vice President of Investor Relations and ESG for Chegg. Thank you. You may begin.
Good afternoon, and thank you for joining Chegg's fourth quarter 2023 conference call. On today's call are Dan Rosensweig, Co-Chairperson and CEO, and Andy Brown, Chief Financial Officer. A copy of our earnings press release, along with our investor presentation, is available on our Investor Relations website, investor.chegg.com.
A replay of this call will also be available on our website. We routinely post information on our website and intend to make important announcements on our media center website at chegg.com/media center. We encourage you to make use of these resources.
Before we begin, I would like to point out that during the course of this call, we will make forward-looking statements regarding future events, including the future financial and operating performance of the company.
These forward-looking statements are subject to material risks and uncertainties and could cause actual results to differ materially from those in the forward-looking statements. We caution you to consider the important factors that could cause actual results to differ materially from those in the forward-looking statements.
In particular, we refer you to the cautionary language included in today's earnings release and the risk factors described in Chegg's annual report on Form 10-K filed with the Security and Exchange Commission on February 21, 2023, as well as our other filings with the SEC.
Any forward-looking statements that we make today are based on assumptions that we believe to be reasonable as of this date. We undertake no obligation to update these statements as a result of new information or future events.
During this call, we will present both GAAP and non-GAAP financial measures. Our GAAP results and GAAP to non-GAAP reconciliations can be found in our earnings press release and the investor slide deck found on our IR website, investor dot chegg.com. We also recommend you review the investor data sheet, which is also posted on our website.
Now I will turn the call over to Dan.
Dan Rosensweig
Thank you, Tracey, and welcome, everyone, to our 2023 Q4 earnings call. To start, I am pleased to announce the appointment of David Longo as our new Chief Financial Officer effective February 21. As Andy announced on the last call that he will be retiring.
David has been our Chief Accounting Officer and Corporate Controller since coming to Chegg in 2021. And we look forward to his continued leadership in this new role. He's joining us on this call today. So welcome, David.
Now back to the business at hand, Chegg had a good quarter and exceeded our expectations. The last few years to have seen real challenges as we navigate the post-COVID world. Despite those challenges, it's actually an exciting time at Chegg, and I'm proud of the team and how they are navigating the complete reinvention of our company, leveraging the advancements in artificial intelligence and making it core to everything that we do.
In less than a year, we reduced -- we redesigned our entire user experience developed our own large language models, launched automated, answering built proprietary algorithm and algorithm to optimize the quality and accuracy of our exclusive content.
And we began to compete more aggressively for new customers around the world. While early, our packaging pricing and product strategy are yielding encouraging results for both students and our business.
The process of embedding AI into every facet of six platform is ongoing and iterative as we build a truly personalized learning assistant, a service that anticipates the student needs, adapt to their strengths and weaknesses and supports them academically professionally.
And personally, there are numerous ways we intend to aggressively market our new product experience because the data tells us that once the student tries us, they love us internationally, we focused our biggest effort on testing promotional pricing to convert the millions of students who have entered the funnel but did not yet subscribe.
Additionally, we are building sharing into our service to increase word of mouth, expanding our presence on TikTok and enhancing our SEO with increased questions from automated answers. Our business model benefits for more students asking more questions as we index those questions into search and other platforms to drive even more customers.
Let me provide a little context since introducing automated answers. In late December, we've seen a significant increase in the number of students asking new questions as well as the number of questions per student.
This is because our new automated services delivering quality and accuracy almost immediately which is a huge benefit to students by building our own language models, along with our algorithms to check for quality students can feel confident in what they are learning on Chegg and get support in real time.
The impact has been immediate and significant. In January, Chegg's automated answers, delivered more than 2.2 million solutions to students, which is three times the number of new questions asked and answered at the same time last year.
Importantly, as we scale to ensure we meet our standards of accuracy and quality, we expect to launch the rest of our proprietary models by the end of Q1. These models are being trained on Chegg's data, and we are leveraging our 150,000 subject matter experts to optimize our solutions for learning in education, students cannot afford the illusion of accuracy to learn they needed to be correct, immediate and personalized.
We believe this is what Chegg can uniquely do for students, and it's a huge competitive advantage over generic AI models. The overall benefit of our new service to students is enormous, and there's also significant benefits to check as the hype of AI dies down leaders in their verticals like Chegg are taking control of their own destiny by building their own models, which allows for higher quality and lower costs.
As an example, the cost to answer new questions using our own AI models is already more than 75% less expensive, and we believe it will continue to decline over time, this means we'll be able to serve more students at a lower cost per student faster and in more subjects and languages, we are confident in the value of our new product.
And because of that confidence and to be more competitive, we began testing promotional pricing in international markets in the middle of last year. We believe that if we could introduce our offerings to more global learners, they would find the value and benefit of Chegg and continue to choose us and stay with us.
In Q4, we saw year over year new customer growth outside the US for the first time in two years. And just as important for our business model, more of these users are taking the Chegg Study Pack, which is our higher priced subscription and remaining paying customers for longer periods of time.
We developed this pricing and packaging to be revenue neutral this year. While we expand new account growth substantially. While it is still early, we are seeing encouraging results. Given the success of what we've seen internationally. We are now testing promotional pricing for new accounts in the U.S., which began in mid-January.
As we have said, online learning support and skills-based learning are a huge market and they are only getting bigger. AI is still in its infancy and our product road map is ambitious and exciting. Throughout 2024, we are reduced.
We are introducing more AI-driven capabilities such as conversational chat, which continues to layer in personalization and interactivity for our learners. We also plan to integrate personalized learning tools such as practice questions, flash cards and study guides to our conversational learning experience.
Looking beyond 2024, as AI automated translation gets better and cheaper. We plan to expand the localization of our offerings to non-English-speaking users. We also plan to build out more AI capabilities within Chegg's skills and integrate pathways for students with assessments and other tools.
We are already seeing a reduction in the time it takes to launch new skills programs by approximately 40%, which allows us to offer new courses at greater speeds and will significantly reduce our costs. And the importance of skills-based training has never been more critical.
In fact, half of recent graduates are questioning how prepared they are to enter the workforce, given the disruption of artificial intelligence and employers agree as 79% say that workers need more training to work with AI more effectively.
So the opportunity for Chegg skills has never been greater or more important. There are a number of exciting opportunities ahead of us in 2024, and we remain focused on the following priorities, returning to new accounts growth globally, maintaining strong margins and cash flow rolling out the next phase of Chegg's enhanced AI services and leveraging our momentum and skills for continued growth.
Every decade or so, the pace of technological innovation accelerates and new growth opportunities open up the history of the Internet has shown us that vertical players who know their customer, have reach proprietary content and can provide a personalized user experience will win and win pay.
Given the strength of our brand. With over 90% of our customers reporting they're satisfied with Chegg service. We believe we are well positioned to do just that in our sector.
Before I turn it over to Andy, I want to again thank him for all he has done for Chegg during his 12.5 year tenure under his guidance, Chegg grew from a physical textbook rental business to a global online learning platform.
When Andy took the job, Chegg was in debt, unprofitable, and we had a single business model renting textbooks, and he guided us through our transition to a fully digital business. And in doing so grew our digital revenue from zero to over $700 million annually. In its final full year as our CFO. Chegg generated $222 million in adjusted EBITDA and $173 million in free cash flow.
Thank you isn't enough to acknowledge the impact to Andy has had on this company and me personally, Andy really quite a legacy of Chegg, and you will truly be missed. And with that, I will turn it over to you, Andy?
Andy Brown
Thanks, Dan, for those kind words, but more importantly, congratulations David on a well-deserved promotion, and I look forward to working with you as you transition into your new role over the next few weeks. Today, I will discuss our financial performance for the fourth quarter and full year 2023, as well as our outlook for the first quarter of 2024.
As Dan mentioned, we ended the year on a positive note with total revenue, adjusted EBITDA and free cash flow all coming in above the high end of our expectations. While the year had its challenges, we executed well on our plan to reinvent the way we help students navigate their living experience by leveraging AI.
And we continue to see strong profitability and cash flows. This, along with the strength of our balance sheet gave us the confidence to extinguish a significant amount of our debt at a discount and repurchase shares, which we believe have and will continue to enhance shareholder value.
Looking more specifically at our 2023 performance, total revenue was $716 million, with subscription services declining 5% to $641 million. Total subscribers was 7.7 million, of which international subscribers were 2 million.
Since 2021, international has increased from 11% of total revenue to 14% in 2023 or $100 million. And over time, we expect international to be even more significant skills and other revenue of $76 million declined 20% year over year.
While skills grew 55%, this was offset by the impact of exiting the textbook business in 2022. We continued to take a prudent approach with expense management and we were very pleased that we were able to deliver adjusted EBITDA margin of 31% or $222 million and free cash flow margin of 24% or $173 million, which represented 78% of adjusted EBITDA.
We expect interest income to contribute less in 2024 from a combination of lower interest rates and a lower cash balance as a result of the aforementioned repurchases.
Looking at Q4, total revenue came in above the high end of our guidance at $188 million, which drove better than expected adjusted EBITDA of $66 million. Subscription services revenue of $166 million declined 6% year over year, driven by a decline in subscribers, which was partially offset by Chegg Study Pack take rate and a continued increase in retention.
Skills and other revenue of $22 million declined 22% as growth in skills was offset by the impact of exiting the textbook business.
Looking at the balance sheet, we ended the year with cash and investments of $580 million and net debt of $20 million. This is the result of repurchasing $597 million of outstanding convertible notes during the year at a $92 million discount to par and initiated an accelerated share repurchase or ASR in Q4 of $150 million, which reduces our outstanding shares by approximately 12%.
We believe this prudent capital management will enhance shareholder value. We exit the year with 103 million shares outstanding, including the majority of benefit from our most recent ASR. This represents a 19% reduction in shares outstanding versus 2022.
We believe our company is undervalued. As such, we will continue to look for opportunities to return value to our shareholders. Our business is somewhat unique given our subscription model and the life cycle of a student.
While we are seeing encouraging signs in the business, it is too early to predict when we will return to revenue and margin growth. The green shoots and engagement, acquisitions and retention will take time to build our renewal base before we see a positive impact on total subscribers and revenue.
In the meantime, we will continue to be prudent with expense management and prioritization while we continue to drive strong profitability and cash flows with respect to Q1 guidance, we expect total revenue to be between $173 million and $175 million, with subscription services revenue between $155 million and $157 million, gross margin to be in the range of 73% and 74%, and adjusted EBITDA between $43 million and $45 million.
In closing, I am proud of what we have accomplished during my 12.5 years at Chegg. This is by a large measure, the best company I've worked for during my career. The mission, the culture and especially the team are second to none.
And I want to thank everyone who is with me on this journey. In particular, a special thanks to Dan for your leadership, mentorship, and especially the friendship we have developed. It means more than words, concerts. Thank you.
I can also say with confidence that the future is bright for Chegg and as a long-term shareholder, I look forward to seeing the many future successes the team accomplishes.
And with that, I'll turn the call over to the operator for your questions.
Question and Answer Session
Operator
Thank you. (Operator Instructions)
Jeff Silber, BMO Capital Markets.
Thanks so much. Just wanted to confirm something you said in your remarks about rolling out. I think it was the new product experience on to other subscribers. And forgive me if I misheard you. I thought you said Q1 that would happen. Can you just confirm that and is that included in your guidance and how so?
Dan Rosensweig
Yeah, this is Dan. I'm not I'm not. Yes, we are we are constantly rolling out the new capabilities that AI presents. But the big one that we wanted for the quarter. We did roll out already, which is automated answers.
And what automated answers does is it allows us to in almost immediately answer questions that either we used to have to send to a human at a much higher price or answer questions immediately that students wouldn't ask because they didn't want to have to wait until we actually gave them the answer so we have seen a significant increase in the number of students asking questions and the number of questions per students.
So we I think we answered over 2 million additional questions than we other than we got last year at the same time this year. So that was the big move for them for this semester. The other ones will roll out over the course of the year. But this is the one that we think has the most meaningful impact in the near term because not only will it improve retention, but it will improve traffic and overall conversion and growth rate.
So it's the flywheel, the Chegg built itself on, but this is a dramatic acceleration of that. So I think where you may be thinking is when we said we were going to roll out promotional and pricing in Q1 in the US as a test.
That's probably what I was referring to. And that's actually my second question, can you talk about the order of magnitude of the promotional pricing that you've already rolled out internationally and give us some framework of what we should be expecting in the US?
Dan Rosensweig
Yes. So internationally, we started testing this last June and our ambition, our belief was that as more students tried check that they would that they would love it and retain because for the first time we're competing against a large named competitor, ChatGPT, which is turned out, is less of a competitor than we were concerned with and that's really good news.
And that is because they give an illusion of accuracy, but it's not accurate and students cannot afford inaccurate or incorrect information plus they don't go there to learn, they actually come to Chegg to learn. And in order to actually Matthew subject, you have to learn it. So we're built specifically for this.
They are not. And so the more students that came to us and tried us the longer they started to stay. So we rolled out promotional pricing there because we wanted to compete more aggressively for students that have never had a chance to before.
And that resulted in substantially higher conversion rate of people that were already in the funnel of which we have to come in. And register but have yet to convert. And so this is beginning to pick up a number of them.
Second, on average, they're staying substantially longer than they did before. So they're retaining longer. So all the things that we wanted to see plus or taking higher take rate of the more expensive product. So for outside the US, we've already returned to new account growth.
And I and we expect to return to revenue growth much faster there because it just the volume of new accounts that we're able to bring in. So that gave us the confidence in the U.S. to want to test something similar to this.
Now the percentage of discounts outside the US are higher than they are in the US. The US discounts that we're testing or promotional pricing, we'll range they range about 25% to 35% and on and we have the opportunity to raise them back to full price whenever we think it's appropriate if that makes the most sense to do.
So we're testing that to a percentage of our new customer audience in the US only. This does not affect any of the existing business, any of the existing revenue. We have blade ways to block cannibalization. This is a new customer promotion only.
And so the tests ranged from 25% to 35%, and it's only been three weeks. But the results so far in terms of what's measurable, which is conversion is up and take rate is, but the next big metric that we want to look at before we determine which of the test makes the most sense because we're running three different prices in five different variants because we'd like to do things and know the answer to it before we roll it out to the full audience. So right now, so far so good.
But we have a couple of weeks more to go to see how retention works in month, one month, two, month three. And then depending on what that yields we can determine if when and how much we want to roll out in new promotional pricing for the second half of the year. But so far, really good result and outside the US really good results.
Thanks so much for the color. And Andy, again, thanks for all your help and best of luck.
Andy Brown
Thanks Jeff.
Operator
[Doug Enmos], JP Morgan.
It's Bryan Smilek on for Doug. Thanks for taking my questions. Just to start on international, can you just talk a bit about more what drove the inflection back to growth there and what the next steps in international? And I guess just on a market by market basis, did anything stand out in particular it was it just broad-based recovery?
Dan Rosensweig
Yes, so what drove it? There was what drove it and what's driving it are going to be a combination of the brand, new product experience. So the way our business grew initially and so large, you got remember only a few years ago in 2019, total revenue for our subscription services was in the $200 million range. Now it's close to $700 million range. So we have grown quite substantially.
The challenge was we grew so much during COVID. And now we're coming back down to the realization of where we're the bases that we can grow from. And we believe that that's what we're experiencing this year. So outside the US, the new pricing promotion allowed people who were nervous about trying Chegg because we're not as well-known there to cross the chasm and then ultimately subscribe.
And because the price is promotional prices lower, they're actually staying on longer, which is exactly what we hoped for on top of that, the long term flywheel is the way Chegg works is the more customers we have and the more new questions that they ask, not ones that are in their existing database of 100 million of ratings, but new ones that they ask.
They then get indexed into search around the world, depending on which country they came from and then other students search and they find them and then they come into check.
So we're seeing an improvement in traffic, traffic from new questions and then holding our conversion rate so that those two things together, which is promotional pricing and the ability to rebuild the flywheel, those are driving the new growth and they're just driving a faster outside the US because it was a smaller base in which to start from.
But we are anticipating that we'll see the same kind of success inside the US. The combination of the new products when students are asking last January, a year ago, January, we had about 500,000 new questions that we were able to answer on behalf of students, all of them went out to humans this year.
We had the same number that went out to humans, but we had another two plus two million that went out that we were able to answer through automated answering, which is our big AI move to start with the high quality, and that's driving more traffic and more conversion to check. So that's a bigger hole that we need to fill, but we're on our way to filling.
Got it. And did any markets stand out in particular? Or was it just broad-based recovery?
Dan Rosensweig
Yeah. No, it's everybody had done a lot better outside the US, but we're starting on very different bases. So the larger bases are the English-speaking countries because everything we initially did was in English. But one of the other real benefits of AITAAI. is moving from a headwind two, a tailwind for us because it's allowing us to build a better product allows students to ask more questions to answer and faster index them get more traffic and convert more traffic.
And because our quality is unique and we work with our own large language models rather than external models, the cost of it is substantially less so as more countries come online and more students can ask in more languages because translation will really be towards the end of the year and into next year because AI will allow for automated translation.
So in terms of growth rates, you see countries like India Finally, taking off for us. But you see the largest countries that Chegg has historically had like Canada, Australia and the UK. Those are still our largest countries and they are also growing again.
Great. Thank you.
Dan Rosensweig
Yep.
Operator
Kunal Madhukar, UBS.
Thank you for taking my questions. One on the on the new product rollout of the new user experience that rolled out in 3Q of last year. So when you were talking about this new rollout in 1Q of this year, what exactly has changed. That's one.
And second is in terms of the promotional pricing, how long is the promotional pricing for the 25% to 35% discount, is it just for the initial month, or is it for a few months? Thank you.
Dan Rosensweig
Yes, good questions. I'll answer the first one. Let me answer the second first, which is we outside the US, the promotional pricing we expect to continue for the rest of this year and determine whether or not we want to raise the promotional pricing to the normal pricing and when we want to do that.
That's all things that we are testing now. So right now, it's the promotional pricing, only four new accounts and I and it will go on as long as we determine that it makes sense for it to go on. So we don't really have a final answer on that yet.
However, outside the US, the combination of the number of new accounts that we're getting versus what we were getting before, the higher take rate of the more expensive package. And the length of time means that we expect revenue to grow despite the fact that we're getting new accounts, this promotional discount inside the US, we are really just at the beginning and of testing it.
We've got about five weeks of testing, as I said, in five different variants of three different price points and all of them are performing as we would have hoped. And so we'll be able to determine if we want to roll it out to the full audience and if so, how long we want it and promotional pricing, those are all testable things where the goal, obviously is to grow as fast as we can from a revenue perspective and new account perspective.
So we'll look for that balance, but we have time to figure that out and because it's really early on the question over product. No, the answer is we did not roll anything out of Q2 last year. We rolled out we rolled out in the second half of last year was a very, very small beta test where we tested the new interface and we tested search and the search results and the quality of those search results.
And then we made the determination to expand that from 1% to 10%, 10% to 25%, 25% to 50%. And then at the very beginning of this year, rolled it out to 100% of our audience, the new user interface. It was not rolled out in Q2.
We've tested a bunch of things since the inception, you got to remember, we only really made this decision last February. It's only been a year since we did all of this. So we announced things like our relationship with scale that allows us to build our own funds language roles, and we'd move off to OpenEye so that we can build our own and it's a lot cheaper and all those things a lot has happened between now.
And then the big move started really in that the testing of automated answers target towards the end of Q4. And then really rolled out in the first quarter of this year, really just in January now. And it's just been a spectacular success in terms of utilization of the quality of it.
And we have a lot more to go. And the hope is that by the end of 2024, the overwhelming majority of questions will go through automated answering rather than humans and that will continue to substantially lower our cost per question and reduce our overall CapEx over time on an annual basis.
Thank you.
Operator
Josh Baer, Morgan Stanley.
Josh Baer
Great. Thanks for the question. And Andy, congrats again on your retirement. I'm wanted to ask about the trajectory of revenue per subscriber. I was just hoping you could kind of unpack some of the different of factors.
I mean, we've talked about some of them, but basically, how are you thinking about the trajectory of revenue between the growing mix of international and the promotional pricing broadly, is there still runway for the Study Pack bundle attach? And then also just thinking about the amount of time that a student has their subscription turned on in a given year? Thank you.
Dan Rosensweig
Yes, all terrific questions. Exactly why we test so that when we're ready to give an actual answer we can. What I can tell you is outside the US where the test is being done well, not a test now. It's run it's been running now for about seven months from what we look for is first thing we look for is conversion increase, and that happens or doesn't happen right away.
In this case, it happens depending on the rate of discount in the country and where we are price before. So different conversion rates, all of them up nicely. It's very hard to make up their promotional pricing discount just on conversion so that the next variable you look at is take rate, which is also measurable right away.
And yes, there's room to grow on the take rate, particularly as we start to differentiate the capabilities inside of Chegg and Chegg Study Pack more based on AI capabilities and other kinds of things that we can offer in the more expensive one that it's that's had not started yet and probably will not start the rest of this year.
But yes, there is room to move because we still have more than one out of one out of two that don't take it, that the second variable is the third one that has the greatest amount of impact because it starts from your net paying customers, your net subscribers, which have to do with your renewal base.
We are seeing a very nice uptick in the length of time that students are saying. So in a perfect world, we would make up the revenue per customer get as neutral as we could to revenue per customer and over a multiyear period.
We believe that's exactly what's going to happen but in the short term, particularly outside the US, the volume of new customers and those other variables together will generate revenue growth outside the US, even though on a per customer basis.
It's still not equal yet, but it's very early. We have multiple wind retention periods to go through a number of years to go through. And so that's what's been going on outside the U.S. has been really positive inside the US. The plan.
What we are doing is we are testing and the first measurement. So it's not to 100% of the new accounts. It's only to a percentage of them, but it's only to new accounts. So it doesn't affect any of the existing revenue. Then as I said, there's five variants with three different price points. And that has to do with presentation of why there's five.
And then what we do is we narrowed down to the ones where we see the desired conversion, the desired take rate. And in a couple of weeks, we'll actually see how it improves retention, depending on how month one retention improves that month to month three and month for all of those, hopefully get us closer and closer and closer to a breakeven on revenue per customer with a lot more customers. That's the objective.
So we have lots of room to grow, lots of optimization to do. And again, the test is now three weeks old, but the first two hurdles are really good, which is driving increased number of overall customers versus what we would have done had we not done because we believe that in a competitive market we want to go.
We want to go fight for market share because frankly, nobody else is anywhere near as good as we are and once students realize that they stay on and they use us more. And then when you launch the new product stuff which we have with automated answers.
As an example, this is the highest engagement on a per-customer level we've ever had. And if it works, as we hope which is to regenerate the flywheel, more customers, more new questions, more new questions, go into search, more students that haven't used us go to search, find it or TikTok or wherever we put it and driving customers.
All of those things so far have been positively impacted, but the overall objective is to be as close as you can to revenue breakeven on a per customer basis because you've got those variables then in year two for whenever we choose to get rid of promotional pricing, we raise the rates on those existing customers.
So the revenue on those customers goes way up. And that will be testing that to determine cancellation rates and all those things. But the business model is very, very sound. And but we have to do it one brick at a time. And we've done the first two bricks, which is conversion and take rate.
Next one is retention. And then what can you raise it to when you want to comment our promotional pricing for the customers that took it. And how long do you want to keep promotional pricing for new customers, but right now better to fight for market share and make sure people don't come away with the conclusion that ChatGPT. is a better solution than us because it's not.
Josh Baer
Great, thanks, Dan.
Dan Rosensweig
Okay.
Operator
Ryan MacDonald, Needham & Company.
Ryan MacDonald
Thanks for taking my questions and best of luck in the future. And David, congrats on the promotion on.
Let me just start, and I'm can you talk about sort of it's great. It sounds like there's some real benefits here on the automated answers in terms of cost savings and how you're answering questions. Can you talk about the mix of, I guess, the overall questions that are being answered today by automated answers?
And then perhaps what we need to see that get to as a percentage of the total questions for that to start to translate into some margin expansion relative to sort of what the first quarter guidance implies?
Dan Rosensweig
Yeah, that's exactly what we're looking at. So what you should be looking for what we should be talking about and what we will be talking about over the course of the year. We didn't really know where to start talking as we didn't know how popular automated answers would be and how quickly it became this popular.
But as we articulated in the prepared remarks already, the cost of answering a new question is 75% less simply based on that the tailwind that we're getting from a I think we call it automated answers, but it's a result of us taking our own large language models and developing our own answers.
We have a quality algorithm. So right now you won't see much in the way of savings in the first part of the year because we haven't fully rolled out all of the different large language models. The expectation is over the course of this year.
And by the end of this year, the overwhelming majority of all new questions will go through automated answering and there will be fewer and fewer going through Q matching. So you should begin to see margin expansion towards the end of the year.
As a result of that, we really see a CapEx reduction first, right, which is the amount of money that we spend on content has been coming down, but it's going to come down again this year and then substantially in 25. So we can answer 10 times as many questions for the cost of one question is the way to think about it.
But we just we have to get all the large language models out so that we can end the quality at the level that students expect from check that they don't get from ChatGPT. or Bard or anybody else. And that will take up the rest of this year to do it and done that.
And we're very excited about it because it will be a meaningful improvement in our ability to pay for content and have super-high quality and as I said, also generate the flywheel. The other real benefit is the more questions that get indexed more people find us.
We don't have to pay for that traffic and it comes. So that's the way Chegg was originally built when we first built QX. for Q&A, and it's just going to be a lot faster now because just the volume of questions that we can answer, I don't think we've ever answer to your questions in a in a month, and that's just the month of January which really doesn't get started until mid-January.
Ryan MacDonald
Super helpful color. I appreciate that, Dan. And then you talked about '24, the growth priorities are one of the priorities is obviously that pathway and returning back to growth, sounds like the promotional pricing is going to obviously help with that.
But as you think about the macro, how important is sort of enrollment within colleges and universities to driving that return to growth algorithm for Chegg? And sort of how do you how are you sort of building that into your sort of assumptions for this year?
Dan Rosensweig
Yeah, look, of course, it hasn't. It has an effect. It's just not the biggest one. And I know people think that it is and I can understand why they do, frankly. But let me just let me just sort of size to you just US because outside the US, it is just it's wild, right?
We jumped during COVID, then sort of stop when things got out of COVID, then Chad CPT. launched ahead of it beginning of an effect, but the new product and the new automated answers and the new indexing of the question to the promotional pricing is reverse that course outside the outside the US.
And so yes, so when you think about enrollments last year or less on average in a given year, Chegg will have over 10 million students in the U.S. that will register for Chegg but not convert. And so we have that whole universe to go after that has nothing to do with enrollment.
So we think that near term, the goal is to get those people to cross the chasm into becoming loyal tech customers because our retention rate is so high because our quality is so good. And now with the immediacy of the responses at high quality, which no one else can give them, we think that's the key to growth. So enrollment will matter over time. But right now, there's a big group of people that are right in front of us.
Andy Brown
Brian Peterson, Raymond James.
Brian Peterson
Thanks for taking the question and a well-deserved congrats to Andy and David. So just I want to follow up on Ryan's question. You've done a lot with the platform this year and you've done that with R&D dollars actually being down year over year.
Actually the opportunity for automation, but also mentioned scaling the different languages and time to roll out the alums. I'm just curious how we should think about the puts and takes to R&D intensity in 2024, maybe how that arc should look longer term basis?
Dan Rosensweig
Yeah, really, really good question. So hopefully, for those people who like you that have been tracking check for a while you see, we tried to be very, very efficient, where's our capital. And so if you take a look at OpEx, OpEx has not grown substantially in the last couple of years and it's been relatively flat.
We expect it to be relatively flat again for this year. And yet we have 250 people now working on AR and that is because we have a history of reallocating resources against the things that have a priority and either stop doing the things that no longer matter as much or where they're at scale and can survive with the number of people that we have on them.
Same thing goes through with our sort of capital expenditure, which is the overwhelming amount of capital that we spent it was on content. Of course, we're not like there's I think there's some confusion because we said last April that we have a partnership with OpenEye. And so it's easy to have made.
The conclusion of is really on us that we have big API costs. We don't where we have moved mostly often will almost be completely off OpenEye shortly. Because it's a lot of different alternatives and they're a lot cheaper.
But more importantly, since the real cost is sort of GPU costs and other things like that it's not. And so the overwhelming cost that Chegg has And CapEx is always going to be content. And since the cost of content is dropping significantly, it's already down 75% based on automated answers versus human answers to the next big steps to get off the more human answers, which we hope to do closely by the end of the year.
So the way to think about it is you can assume our capital this year will be much better, much more efficiently spent because AI is allowing us to spend it a lot cheaper on a per question basis than we ever could before. So we are not holding back investment. It's just all we can get a lot more for our money now thanks to AI. It's one of the one of the great benefits of having moved as quickly as we moved.
Yes, operator had another question.
Operator
Alex Fuhrman, Craig-Hallum.
Alex Fuhrman
Hey, guys, thanks for taking my question on. Can you talk a little bit about your core customer today versus a few years ago. It sounds like the biggest change on given everything going on with AI. Is that maybe some of your more casual users at the margin who might pop in for a month or two around final or midterm exams have kind of maybe looking at other AI powered options.
Can you talk about the types of students who have stuck with Chegg. Are there any major call out and you're now versus a few years ago in terms of your core student in terms of either age or full-time or part-time status or field of study or two versus four year schools. Anything like that that you're focusing on a little bit differently versus a couple of years ago?
Dan Rosensweig
Yeah, it's an awesome questions. So for those who having spent as much time as you have on this, the overwhelming percentage of students in the US go to state schools. So we hear about that. We hear about the schools in ag maybe have lost down their prestige, and that's a couple of hundred thousand people versus 22 million. So they're not a big percentage.
The overwhelming number of students in the school go to state schools and a very large percentage of them go to community colleges. As an example, 10% of all students in this country go to just one community college system, which is California. So over 2 million students go to community college.
The average age of a student is about 25 years old and about 25%, 26% of them already have children the biggest growth areas for where students are going. They are they are shifting into three or four very large buckets in very big numbers. So one of them is online.
And so our focus on making ourselves more visible to online students has been a focus. The last few years and one that we want to accelerate. And we really do believe that automated answers will help us with that because we'll be able to have a lot more things that people can be searching for in search.
So and we're hoping for a boost to that. So I think the largest school in this country that's not for profit is an online school. I think it's Southern New Hampshire University, which are which Dr. Paul, the bankers on our Board. So we know a lot of information about that. So they've actually grown their enrollment substantially year over year because more people are not going to get degrees, they're going to get particular courses.
The second big growth area that the country is seeing is a real move to giant state schools with incredible sports programs. So students are determining the value of graduating from school A. versus Scooby is not significant in their career path.
So they rather go to a school where they can enjoy themselves. And that's exactly what they tell us by the way. So we're not we're not speculating. I would give the numbers shown as what they have told us.
The third big area is some students of color are gravitating, as you might imagine, to HBCUs. And the reason is they finally got funded by the government and those and they're very much developing their programs and they feel more purchase culturally aligned there. So those are three big areas now where we're spending more of our time trying to figure out how to reach them.
The other thing is even though more students every day are taking STEM the classes, one of the other benefit of a eye for us is you can ask a question increasingly so on Chegg that is outside of SMB. And as we build that out, we'll be able to expand our US TAM even a little bit more.
And then, of course, the giant TAM, it's outside the US, which is not what you asked about, but that is it's hard to imagine countries that are sort of bigger growth opportunities in India, the Philippines, Mexico, places like that.
And we already are seeing really good results as a result of that promotion in those places. And we expect over the next several years to really trying to optimize around growing in those countries. So that that's what we're focusing on in terms of our messaging and where we're putting our messages. Hopefully, that helps answer it.
Operator
Eric Sheridan, Goldman Sachs.
Eric Sheridan
Thanks so much for taking the questions. Maybe two, if I could, on your own LLM., can you talk a little bit about what you're most excited about in terms of how the LLM. will evolve and what it means for product and roadmap on a multiyear view as opposed to just this sort of the one of the LLM and sort of how it might get smarter, how might change out might result in different elements of pricing dynamics on the platform on that would be number one.
And then in terms of generating the flywheel effect that you talked about and referenced in the slides, can you just go into a little more granularity of what we should be watching for of where the incremental investments will be made as you see some of those signals as we look out over the next couple of quarters?
Thanks so much.
Dan Rosensweig
Yeah, those are two very substantive questions. Let me take the second first because I can remember that once So what should you be tracking? So as I said, our expectation and of course, expectations can change. And as we learn more, we always share more. Sometimes we share too much like our concern that Open AI, and we take our customers and it has.
So but in this case, we have fought back or battling back. And we think we've got the things in place to compete quite effectively. And what we look at is how many students are asking questions and how many and what kinds of questions that they are asking and do those questions resonate with other students that go into search or TikTok?
And will they generate more traffic for us at the flywheel. It will be hard for you to look at expenses because they're going to be relatively flat in CapEx and relatively flat in OpEx because of the efficiency in which we move people into different roles or move our capital expenditure to things that work better than the things that we're working before. So it won't be that obvious for there.
But I think you'll hear from us a number of Quest number of students that are asking questions versus the past number of questions per student overall, number of new questions being asked and answered by automated questions and the greater percentage that are answered by automated questions that the better it is for our cost structure on CapEx.
And then we expect as we get more comfortable with the certainty around these things that we're learning that we'll be able to put out a longer-term forecast, which we can be measured against more readily. We're not there yet because we because all of these are so new.
But as you saw from the last three quarters, we have beaten the expectations we put out there and that's building more comfort for us, but I think automated answers and the new pricing and packaging will learn a lot more in the first half of this year.
So you won't you won't be able to see it sort of in where the expenditures are up, but the expenditures over time should be flat to down. So the question on LOMs improve over time. Yes, that is that is an amazing question, and I'm not sure even I understood just how much they can evolve and how quickly they can evolve and how useful they are when they evolve.
So the speed in which we can answer your question, the types of questions that we can answer and the cost of doing it, all of those things is the LOM learns more, get better. The second thing is the personalization that we can do on a per-student basis.
So their data beyond just the question and the answer there is data that gets mined, what school you ask, who is your Professor? How does that Professor prepare? So we have years of history of when the midterm is when the final is. So imagine a scenario on a per student basis is over time. This is not this quarter.
I don't want to confuse anybody, but it is what the LOMs will enable us to do. And what we're working towards will be able to go like a coach and say, to the students say we know your midterms coming up in two weeks.
Do you wanted to build you a schedule of how to study you wanted to build you a unique practice test just for you and we can do it not just based on the subject but based on the individual and the professor themselves, because we have all the data of what that Professor does and when they do it, the kinds of things that they ask. Because we have all these years of history already in our dataset.
Those are the kinds of things that are unique to Chegg that can differentiate. We think those things create value in terms of going back at some point to pricing power and second on retention and expand out to the kinds of things that we can do.
So another example is things like we know you go to this school, you go to University of Arizona, and we know that students that take these classes major in this and we know students that major and this generally tend to work for these 10 companies and these 10 companies are looking for these skills. Would you like us to assess you on whether or not where you are on those skills and then we can upsell them to.
Would you like us to teach you those skills that you can get a certificate that shows that you're actually competent in Asia, things that our own LOMs and datasets will allow us to do that. We believe no one else would be able to do.
Operator
Brent Thill, Jefferies.
Brent Thill
Thanks. It relates to Q1 guide. You're guiding EBITDA down 24% year on year, and I think revenues guided to down mid to high single digit. I think the question we're getting is when do you returned to healthy growth on revenue and EBITDA?
Dan Rosensweig
Well, I'll answer the second part of that because I'll be here going forward and Andy is already in retirement, but I think you can answer why the model is what it is in the first quarter. So we are not prepared today to say when we will return to that, we do feel comfortable in saying we will return to that.
And because of AI going from a potential headwind to a tailwind we think we'll be going to that sooner and than we otherwise would have because of our ability to have more questions ask have them index drive more customers plus promotional pricing so what in the next few quarters, we think we'll be more able to articulate when we think that happens.
But we're absolutely currently on the path to do that. And we our margins, we think we can return back to where our historical margins were. This is a company that should be able to grow and start with the fact that we started with fewer net customers this year than last year because of the people that rolled off in the post COVID and all the other stuff that we were dealing with.
So we have about 9% fewer customers starting this year than we did last year. What you'll be able to track over time because we report on our customers every quarter is you'll see that gap on soon when we do our guidance, you'll see that gap continue to close, and that's how you'll see and be able to estimate when we'll be able to return to growth, but we're on that path.
Operator
Jason Celino, KeyBanc Capital Markets.
Great. This is actually Devin on for Jason today and thanks for taking our question. Just one quick follow-up on Ryan's question also on automated answers. I think in terms of the questions being answered by the answers.
Are these more within the STEM and Business Objects? Or have you seen I mean, Andrew also having the ability to provide quality responses to maybe the non-stent subjects that I believe is an area that's a little bit I'm less penetrated for Chegg.
Dan Rosensweig
Yes, terrific question. So let me start with the initial questions through three parts to that question. The first one is what are students already asking us. Those are because the people that are already with us or SMB students, they're asking deeper questions that AI. allows us to answer in the next phase is going to be conversational nature, which will be really fun and helpful for the students because we'll be able to ask them questions.
In addition to them asking us questions. So first, wondering if they're going deeper into the categories that that perhaps might have been really cost inefficient for us to tried to answer them because they're deep into subjects. So that's the first thing.
The second one is yes, we absolutely are building it to expand beyond DMD as students asset. Third part, though, is we have to be comfortable that we can answer with the same quality and accuracy that we can answer.
Other questions. And that's just going to take the course of this year to get ourselves more comfortable we're focusing now first on the 26 subjects we have in the 52 LLM.s that we're building for each of those subjects, and then we'll move to outside of where we currently are. But that is absolutely part of our roadmap. It's just not part of the roadmap today.
Operator
There are no further questions in the queue. I'd like to hand it back to Dan Rosensweig for closing remarks.
Dan Rosensweig
Thank you, everybody. Just to repeat, it's been a complex last several years with COVID and post-COVID, then with AI and other variables and the company is in extremely strong shape. Despite that, which is we generate a lot of profit. We generate a lot of cash flow and we are we are moving back to net cash positive on our balance sheet.
We've been really efficient with our capital and expect to continue to do so. But the real opportunity is rebuilding the flywheel with the new product, new service, new automated answering and a am moving from potential headwind, two absolute tailwind on our growth and um and costs.
And so it is a very exciting time. It's not easy. These things are moving very fast and the speed in which we're moving assessing than we've ever done. And so I just I very much look forward to the rest of this year in the future because we've got a lot of good momentum ahead of us.
It's just going to take time and we have to rebuild the base of net new accounts. And like I said, you'll see that every quarter if we execute well on. And of course, I want to welcome David Longo's to the role who brought David in a couple of years ago with the hope and expectation that he would fill Andy's role nobody can replace Andy.
And to Andy Brown, I say that I've never had a better partner, a better work trend and Chegg would not be where it is today, had you not joined as you know, I made the decisions that you've made and back to people that we've had and hire the people we have. So your mark is indelible and thank you, my friend, and thanks, everybody. Talk to you next quarter.
Andy Brown
Ladies and gentlemen, this does conclude today's teleconference. Thank you for your participation and you may disconnect your lines at this time and have a wonderful day.
