7 Ways to Boost Retention (Both Pre- and Post-AI)

For consumer products, the rubber hits the road at user retention. In this episode, a16z Partner Bryan Kim covers the seven most effective ways for AI-native consumer companies to improve user retention. From essential strategies to proven methods, Bryan reveals how leading AI-native products are already benefiting from these approaches. Featuring real-world examples from companies like Snapchat, Google, Perplexity, and BeReal, this is a must-watch for founders, product managers, and anyone interested in the latest AI and consumer application trends. Packed with valuable insights, this video will help you enhance your product's retention and success. Timecodes: 03:26 - Obsess Over Delivering the Core Product Value. Fast. 06:11 - Feature-Gated Onboarding 10:07 - Designing Reciprocity 12:10 - Building Smart Notifications 17:12 - Give Them Summaries 19:54 - Keeping Streaks Alive 22:14 - Status for the Power Users Resources: Link to Bryan’s article: https://a16z.com/7-ways-ai-native-companies-can-improve-retention/ Find Bryan on Twitter: https://twitter.com/kirbyman01 Stay Updated: Find a16z on Twitter: https://twitter.com/a16z Find a16z on LinkedIn: https://www.linkedin.com/company/a16z Subscribe on your favorite podcast app: https://a16z.simplecast.com/ Follow our host: https://twitter.com/stephsmithio Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.

• 0:00Speaker 0

  My estimate is that on average, if you employ some of these combination of these methodologies, you're able to move up your what I would I would start with D30. I think you can move that up 10 to 15 percent.

• 0:11Speaker 1

  Whoa.

• 0:12Speaker 0

  Which is which goes from okay on my on my list of like 20 percent D30, that's like okay, that's good, to whooping 35 percent, that's exceptional.

• 0:24Speaker 1

  We are seeing in this wave of AI a lot of questions around retention. And I think something that you've pointed out is that you know maybe things aren't so different from the last wave of consumer applications, or at least that we can learn from them, right? And so you're you're an expert in this. What have you seen in terms of the past retention mechanisms that you feel like in consumer wave one can now be maybe applied to consumer wave two? Aaron

• 0:48Speaker 0

  Powell Yeah, this is uh this is a fun one, fun, very fun topic. And it actually came uh from one of my uh chats with AI native founder who was trying to find ways to say, how do I get my product into more consumers' hands? It's interesting because um I've said it once before that retention is the you know bread and butter of key consumer products, because if you don't get users coming back and again and again, ultimately you have what I call this like burned forest problem where you went and cut all the trees off and there's no more new acquisition to do.

• 1:20Speaker 1

  Yeah.

• 1:20Speaker 0

  So ultimately retention is incredibly important. And it tended to be this metrics that is very difficult to move.

• 1:27Speaker 2

  And I

• 1:27Speaker 0

  remember this because I ran uh gross product at Snap. And when we move, when I tell you one percent under pips, one percent of uh a retention rate of a D30 that we're tracking, we moved it up by a percent. There was a huge party. That that actually led to a very clear uh incremental value to the company in terms of the daily active users that we're gonna keep and the reduction in any pay-dads that we would have to do. And it was very, very clear benefits. So that ingrained in me, oh my God, this this number is very, very difficult to move. Nowadays, because of these uh methods that are available to consumer founders, uh my estimate is that on average, if you employ some of these combinations of these methodologies, you're able to move up your what I would I would start with D30.

• 2:17Speaker 0

  I think you can move that up 10 to 15 percent.

• 2:20Speaker 1

  Whoa.

• 2:21Speaker 0

  Which is which goes from okay on my on my list of like 20 percent D30, that's like okay, that's good to whooping 35 percent. That's exceptional.

• 2:30Speaker 1

  Yes. Yeah.

• 2:31Speaker 0

  So what I've learned in the past two, two and a half years, meeting with excellent founders who are on the front line, thinking about different product mechanisms. There have been seven ways that I've seen that uh you know, very distinctively you can employ to move your retention up. And uh the more what's more, which is really exciting is that I've thought about ways that the new AI native products are actually adopting these. And they are already seeing the benefits, and the set founder that I'm talking about had actually told me that, oh, after that chat, I employed a bunch of mechanisms and and actually my weekly retention shot up by a lot. And that's all the validation I need to have published this. Aaron Powell

• 3:13Speaker 1

  Amazing. Well, let's go through them. So you said you've got seven mechanisms, and as you said, we're kind of seeing examples from the past from the earlier traditional consumer wave, and then you're already seeing implementation, which is kind of crazy because we're so early into this new wave. So let's start with the first one, speed to core product value. What do you mean by that? Aaron Powell

• 3:30Speaker 0

  Retention is made up of two things in my mind. Onboarding matters. If the new user comes on on day zero and didn't find a pro value that they're looking for, they're not going to stay.

• 3:42Speaker 1

  And you

• 3:42Speaker 0

  just like lost that user. And it's very, very unlikely that you recover from a low D0, D1 retention. It's very difficult to recover from that. So onboarding really matters, and then you know, making sure that as you onboard someone, getting them to experience that core product value that you're delivering quickly matters a ton because the less friction, less time-lapse you have between opening or or starting the product, the better chance you have to retain them. You know, if you just look at traditional companies, or traditional is the wrong word, but you know, uh leading companies. Google, when it first started, and Larry Page was very famous for uh really really hammering this down, where speed of search, shaving milliseconds to search result really mattered. And that really mattered in terms of like drop rate that really mattered in terms of people sticking around and coming back and working on the product.

• 4:34Speaker 0

  And sort of that became one of the mantra enough that they started building it into their search ranking. If it the how fast the site loads became a factor in relevance because it's relevant,

• 4:46Speaker 1

  surprised at how much this impacted. Exactly.

• 4:48Speaker 0

  And they wrote this like you know blog that said like speed matters. As they think about the AI products. It's interesting because we talk a lot about the improvements and underlying foundational model and what that means. What I well actually what that means is the results come to you faster, better, more relevant. And there's an example of a com AI native companionship product where they actually did this test where they use GPT-1, GPT2, GPD three, and they actually track the cohort of user retention using each of the models. And it's very clear as a model quality improved, which means you get response faster. The retention uh out to 90th day of user being activated kept on improving. And one of the other sort of ways I think about this, and this is kudos to Arvin at Perplexity, is that whenever the state-of-the-art model comes out, he incorporates that into perplexity on the day of, if not earlier.

• 5:43Speaker 0

  And you know, I was thinking about this, I'm like, oh, that's like incredible velocity and that that's great. But I actually think what it also does, it makes this product perform um much better than anything out there and delivers faster, better, more relevant results. And I think that's part of um how um the speed to core product value is manifesting in these different type of AI native products.

• 6:07Speaker 1

  I love that they can see the difference between, as you said, GPT-1, 2, 3, 4. Let's go to the next one. Feature-gated onboarding. What do you mean by that? Aaron Powell

• 6:15Speaker 0

  So as I said, onboarding matters a lot. Now you can have an onboarding where you gently nudge people, like this is how you use the product, have fun. Or you can actually really force feed them. And I think what we are seeing is that if you actually gate the usage of the product behind something that users need to do to unlock, I think that actually ends up leading to higher retention. You're exchanging lower conversion to actual product initiation, if you will, because some people will not be willing to go through that onboarding process. The thought process there is that they probably weren't gonna stick around anyways. So you're actually like finding out early that they're not willing to do that. But if they do go through it, you actually see a much higher retention for people who are activated.

• 7:02Speaker 0

  So the example from prior non-AI generation would be the company called Laps. It's a camera app that gives you the feel of the instant camera, if you will. And what's interesting is it's essentially a social product.

• 7:15Speaker 1

  Yeah.

• 7:16Speaker 0

  And what they actually gated the product usage behind is not only a pretty lengthy onboarding process of how the product works, but also you can't use it unless you invited five new friends. Now, what does that do? It's famous for social products where if you have more than five friends on the product, you had your chance of sticking around is much, much higher. So what this product is doing is actually force building a friend graph that when you come onto the product, you already have uh value that you're going to find by looking at your friend's content, one out of five, two out of five out of five. So I think that's like a way of building in retention as you go in. I think one of the more recent AI examples would be vigil.ai.

• 8:01Speaker 0

  This is a fun product where you get to uh put a picture of yourself and you all you need is an image. And they have certain set of scenes. And uh one of the more famous scenes was Lil Yachi dancing out in stage. And you can very easily overlay anybody's image on it. There have been very different, you know, viral moments of Jerome Powell doing uh doing a dance after announcing a rate cut. And and those type of examples are so fun. Now, how do you create these? It's magical, right? You layer on your image, like I actually put on my image on one of the famous guardians of the galaxy scene where Chris sings ooh child in front of his nemesis.

• 8:41Speaker 2

  Ooh, child things are getting brighter and bring it down hard.

• 8:47Speaker 0

  That's uh it's one of my favorite scenes. It's so funny. Now what Vigil does is it's only available in Discord

• 8:52Speaker 1

  to create.

• 8:54Speaker 0

  There are three steps you need to go through to be able to create a vigil that includes reading the rules that also involves actually going through the mechanics of how to build the right sequence in the Discord chat so that you can get that result. And only then, after completing those steps, can you use a product.

• 9:12Speaker 1

  Yeah. Now

• 9:13Speaker 0

  what does that do? Well, that also that effectively taught you how to create this forcefully to some extent. And now you have created it. Voila, the magical moment happened.

• 9:23Speaker 1

  Right.

• 9:23Speaker 0

  And now your propensity to come back to the product is a lot higher because it's not foreign to you anymore. And I think that's a very interesting use case of actually putting your core product creation behind uh uh sort of a fairly long use uh rules and understanding how to use a product and actually getting them to create an example.

• 9:44Speaker 1

  A good one too, a good example because they've been trained. Aaron Powell

• 9:47Speaker 0

  Exactly. So it's been it's been a very interesting uh thing to watch, and I think more and more uh companies with somewhat of a uh niche or specific uh creation flow would benefit from.

• 9:59Speaker 1

  Yeah, I also like that. The frame of maybe how can you create a product that is so good that people are willing to sit through the friction, right? So the next one, designing reciprocity.

• 10:10Speaker 0

  So here, this is one of my favorite ones because this actually underpins probably the highest retention gain that you could get. And this is also one of the things that I don't have an AI example yet. The idea is give to get i.e., if if you have a product and and you haven't contributed to the product, you don't you don't get it. The example really would be B Real, which is a social product where you can check up on your friends uh you know authentic moments. And it's a fair value exchange. Hey, you don't get to see your friend's authentic moment until you're willing to share yours. And I think that's a powerful concept. Uh and the reciprocity is very effective because what it does is now as a user, I have to create. I have to contribute to consume.

• 10:57Speaker 0

  The other interesting thing is if I get any sort of reaction or positive reinforcement of people liking that content, whether it's a like button or a comment button, what have you, I get a jolt of uh ajoy or dopamine. And I'm more willing to do that on an ongoing basis, which leads to higher retention, which leads to richer content on the platform and so on and so forth. I think this can actually translate really well into certain um AI products where it becomes important for uh the product to have a rich creation flow. I think we're trained to go to a platform hoping to see C of content, feed, or what have you. But what if you gated that behind creation? What if people needed to create and actually contribute to the you know dynamicness of that product before consuming it won't work for every single product, but I think you can design it into different AI products where creation and this sort of maintaining a dynamic creation consumption flow within the product can really really be helpful.

• 11:59Speaker 0

  Aaron Powell

• 12:00Speaker 1

  Yeah, I know. I mean you've got my wheel spinning, and maybe it's partially because creators get so much value from these AI tools that they don't actually need to consume as much as in the prior era. But looking forward. The next one is building smart notifications.

• 12:13Speaker 0

  Notification is this one thing where it's trying and true. People know that notification can bring people back. Uh it's it's a double double-edged knife, double-age short, and that you send too much, people turn it off, you are not getting them back. I know this empirically that once you turn a notification off, the chance of something turning that back on is extremely low. Now, what's interesting on notification is this. When you think of notification, I think there are three different things to think about, which is who who's sending it.

• 12:40Speaker 1

  Of course the

• 12:41Speaker 0

  company is like sending it, but like who is a perceived sender to the user? That's one. Two is what is in the message, what is a payload. And then three is what is the frequency? And all of all three of these interact differently. But I'll give you one example where let's say uh Snapchat sends happy Earth Day message. Well, don't pay me into villain, but I don't actually care much about Earth State. Like I don't care. Like why are you telling me this? And in that case, the the payload is happy Earth Day.

• 13:10Speaker 1

  The

• 13:10Speaker 0

  frequency was this holiday, if you will. And the sender in this case is a company.

• 13:16Speaker 1

  Yep.

• 13:17Speaker 0

  It's not my friend, because if you snap me and I get a little you know, purple or or or blue, blue square, I'm not thinking, oh, Snap is sending me another notification. I'm thinking Steph wants to talk to me.

• 13:30Speaker 1

  This is

• 13:31Speaker 0

  great. So the perceived sender is very different. Another example would be part of if you think uh you're going to an event and you signed up for it, you RSVP'd, and you get a notification that the venue changed last minute, or it's raining and they're changing the time, or it's so and so is not going. You want to know that and you get a notification on your SMS that says, guys, instead of eight, we're going to seven. Instead of Crown's King, we're going to some other bar. That's really relevant. And the sender is not partiful. The sender is the host. Aaron Powell

• 14:01Speaker 1

  Right. Yes.

• 14:02Speaker 0

  So that's really interesting because the more you can move the user's mindset into, oh, this is really useful to me. Oh, the perceived sender isn't necessarily the company.

• 14:13Speaker 2

  This is

• 14:13Speaker 0

  actually really useful to me. That's like sort of how you shift the mindset and therefore make it a lot less likely for someone to turn off. You know, if you turn on notification on uh on perplexity, that actually delivers uh you very relevant news that you might care about. So during the open AI, you know, reshuffle or saga, like a lot of different movements happen. And when you're able to get that news pushed to you really relevantly quickly, you no longer think, oh, why is this product bothering me with notification? I can't believe that just happened. I am so glad I learned it before I learn it on TikTok or Twitter.

• 14:49Speaker 1

  Yeah.

• 14:50Speaker 0

  Or news. And I think that's sort of a way to build very smart notification that a lot of AI products can actually uh build in. Aaron Ross Powell

• 14:58Speaker 1

  What is your take on giving consumers the ability to customize their notifications as in is it generally best practice to design your own notifications and send those out as you see fit and also iterate? Or there are some products that will let you say, you know what, I only want notifications if I'm tagged here. Or you know, maybe I want everything of everything in this kind of community. Aaron Powell

• 15:20Speaker 0

  I think overall most products won't need to actually go through the granular detail of what you want notification for. Um but of course, like let's think about the commerce company or social products. The commerce company, do you want to notify when I have random sales? Do you want to notify new products? Do you want to get notified for returns or relevant to your order? Right. Well, I just just my orders, please. You know, like maybe that's like the relevant way to think about it. Social products, the same thing as you said. It depends how you're using it, right? Like Instagram, I I want to be notified as someone comment on it if I want to be notified if I have new friends, et cetera. But that type of granular detail initially for AI products, I think is probably less relevant at this moment.

• 16:04Speaker 1

  Yep.

• 16:04Speaker 0

  And really uh probably think more about how do I actually deliver value through this notification. So one interesting that you asked. So let's actually use the example of character. Character AI has a notification toggle button where do you want to be notified by your companion? What does it even mean? Aaron Powell

• 16:24Speaker 1

  I was just trying to process the. What that means is

• 16:26Speaker 0

  that I you know it's a it's a two-way dialogue, actually. But you know it's a bot or companion that is governed by LLM. What does it mean for that to reach out to me first? Aaron Ross Powell

• 16:38Speaker 1

  That's so interesting. It's a friend. Trevor Burrus, Jr.:

• 16:40Speaker 0

  It's a friend, right? So it's like a very interesting uh emotional nudge of like do you think of this being as truly your friend, in which case you would want to hear from them uh first and proactively versus do you want to engage on your own terms or you want to engage with the sort of dig uh communication with a digital being. So those are the types of notification where I do think is very nuanced and fun and impactful. Aaron Powell

• 17:07Speaker 1

  That is so interesting. You're basically coding boundaries in a way. Yeah. Okay. Next one, keeping streaks alive. Aaron Powell

• 17:15Speaker 0

  Streaks. This is a well-known example. One pet peeve I have on Streaks though is that people think of it as oh, gamification.

• 17:23Speaker 1

  I

• 17:23Speaker 0

  understand streaks. Let's just put a fire emoji in a number next to it, and it'll be

• 17:27Speaker 1

  great.

• 17:28Speaker 0

  It's uh the the well-done streaks are a little different. They're very, very effortful.

• 17:33Speaker 1

  What do you mean by that? Aaron Powell

• 17:34Speaker 0

  Let's say you and I have a streak of 1,000. What that meant on Snapchat is I sent you a video, not a text, not a chat, a video, a snap or image, image or video, and you sent me back something in a 24-hour period as well. And streak doesn't appear until three days of doing that.

• 17:56Speaker 1

  The

• 17:56Speaker 0

  fact that there is a fire emoji and 1,000 next to it is that we did that nonstop for nearly three years. That is not a like button. That is not a retweet. That is not a heart. That is really, really expensive friendship badge. And you bet if you break that, I'm not talking to you. So there are ways to think about these streaks as something that is very meaningful that is not just some number behind it. And this is a way to actually really encourage people to coming keep coming back to your product. One good example that I think about is there's a lot of AI journal products like Rosebud or JOT. And it actually has streak mechanism, and that's helpful and another language learning app called Speak AI.

• 18:41Speaker 0

  And it's similar to Duolingo, but it's a language learning uh based on AI. These products are very beneficial when they're habituated. Writing journals is proven to be very a good thing for your mental health. Language learning. We know that consistency matters. So when you bake into these type of products, even if it's AI native product, if it's habituated, helpful to the users to be habituated, I think these are the very, very interesting places to start seeing it. And bonus points if you know uh founders can start thinking about streaks as digital signature or display of something effortful and expensive, not a like button, not a retweet button, which I tend to think of as cheaper digital signs.

• 19:26Speaker 0

  Trevor Burrus, Jr.:

• 19:27Speaker 1

  Yeah. I mean, I almost think it's identity-based in a way. There are some people, by the way, you can look this up who have been doing the duolingual streak for like a decade. And so for those people, there's there there has to be an identity piece of that where they're like, yes, it is trying to keep up the streak, but it's also like I'm a language learner or I'm dedicated, or as you said, I'm a good friend.

• 19:46Speaker 0

  I'm a good friend.

• 19:47Speaker 1

  Yeah.

• 19:47Speaker 0

  Exactly. And he or she is a good friend of mine.

• 19:50Speaker 1

  Yeah. Exactly.

• 19:52Speaker 0

  Trevor

• 19:52Speaker 1

  Burrus, Jr.: Yeah. Awesome. Okay, we've got two more. The next one is the wraps. Aaron Powell

• 19:56Speaker 0

  This is fun. Raps popularized by Spotify year and rap. My number one uh musician is Han Zimmer. Um I I was proud of that. I think

• 20:05Speaker 1

  you've got one you can share. Mine are all embarrassing. I'm sure

• 20:08Speaker 0

  I've shared that out quite widely. Reps is interesting because we know AI or LLMs are good for a few things. And that they also include summarization. They're very good for that. So I think what we're seeing now is this unique technology that uh can actually sift through your behavior, your engagement, what you consume, et cetera, and be able to tease out summarization.

• 20:32Speaker 1

  Yeah.

• 20:33Speaker 0

  Excellent for that. And so I think we're starting to see very interesting behavior because if I know if I'm gonna have a wrap, like I'm excited, I'm gonna continue to do whatever that thing is, or whatever that product is, so that I have a pretty cool summary. Or that summarization really benefits me. And you know, I I think back to the old example when I got my first American Express card. Year end they sent like a 20-page document of like what you did, like all the spending that you did. That was not that was not good. That was not helpful. But hey, in the era of LLM, we have companies like Oops Finance where you're able to actually tease out the themes of what's happening. You could very easily say, hey, like you've been eating out a lot.

• 21:14Speaker 1

  Yeah.

• 21:14Speaker 0

  You've been uh spending a lot of entertainment. You spend on a lot of trips, or you should probably do XYZ. So I think um this is one of the other methodologies that I can very easily see being useful to a lot of new AI native products.

• 21:31Speaker 1

  Totally. And I think about the equivalent um in Spotify, they'll tell you, right? Like if you're a Han Zimmer like top one percent fan, you feel really good about that. We need Hans

• 21:40Speaker 0

  in our B Roll.

• 21:41Speaker 1

  Yeah. But also you can imagine the same thing like for finance, right? You're in the top, you know, five percent of savers, you're in that you're the top 10 percent who buys from this brand. Like there's some cool stuff you can do. How fun would that be with that?

• 21:53Speaker 0

  And we're already seeing that in products, like uh there's a product called Airbuds, and they sort of use uh the technology to figure out you're the top one percent listener of Taylor Swift. Yeah, you're you're sharing that.

• 22:04Speaker 1

  Yeah, totally. And and I guess like as you're saying, AI actually helps us enable that because we no longer need you no longer need the Spotify data set.

• 22:12Speaker 0

  Exactly.

• 22:13Speaker 1

  That's really interesting. The final one is status for power users.

• 22:17Speaker 0

  I know we talk about retention as sort of a broad base, overall user-based uh you know thing. But you know, if you actually go pretty granular, there usually is this like bulwark or uh extremely healthy users in any given leading product. Power users. And I'll just give you one example. Let's say MAU, monthly active user. Whether someone showed up every single day of 30-day period. Or showed up once, they're both one moo. But clearly, surely one is far more valuable than the other.

• 22:48Speaker 1

  Yeah.

• 22:48Speaker 0

  And so you start thinking through how do I actually think about power users? How do I really build that in? This is interesting, and this is helpful to build retention because by giving people and bestowing some kind of status, if you will, for power users, the more likely you're able to incent them to continue that status. So old school examples, again, one way to do that on Reddit type of the world is to be a moderator. That's that's an honor for that community. For the subreddit, it's an honor. Strava's local legend. That's an honor. You bet your ass like they would want to stay there.

• 23:22Speaker 1

  Yep.

• 23:22Speaker 0

  And that's very important. And I think we're already seeing uh very cool AI examples as well. One of the companies uh that sort of focuses on uh fine-tuned image models and acting as a hub for that. There's a company called Civit AI. It's important for them to have a large density of fine-tuned image models. And so how do you actually engender that? How do you actually get creators of these models to want to come back and do more is to start bestowing them status and leaderboards. And I think we are seeing uh that type of activity really continue to encourage the power user to keep coming back and keep contributing to the platform, therefore making it healthy. Another funny example, there's a product called ReadWise, where it sort of shows how many books and how many sort of passages you read.

• 24:11Speaker 0

  And being on top of that list or being uh you know a top X percent user on those tend to be uh a really important as you said, identity for me.

• 24:20Speaker 1

  Yeah.

• 24:20Speaker 0

  And I think uh those are a very interesting ways to keep people from coming back and back to the product. There's a famous article, Status of the Service that was written by Eugene Wei, where he says like people are a status syncing monkeys. And to some extent that's true, and I think uh you know, a lot of the AI product, especially where creativity and contribution matters a lot. It's a it's a very unique thing to be able to bestow and and and share with their uh most exciting user excited user base. Aaron Powell

• 24:49Speaker 1

  Yeah. I mean, if you use many of these examples like the Strava local legends, they would rather be up there than be paid to do the things that they're doing. And I think maybe if there's one key takeaway here is as I look down your list of seven different mechanisms, they're all free. You know, you need the cycles to actually implement this, but this is not something that um is rocket science either, and it's not something that you need a huge budget to implement. Like these are really things that you can bake into almost any product. And hopefully we'll get an example of it was designing reciprocity, right? Reciprocity

• 25:18Speaker 0

  example, and I'm excited to see one too. And you know, I think as you said, these are not expensive mechanisms, but I think what's interesting is like you know, you you get to pick what actually really uh boosts the attractiveness of your core product value and really serve that.

• 25:35Speaker 1

  Yep. Absolutely.

• 25:36Speaker 0

  Cool.

• 25:36Speaker 1

  Amazing.

• 25:37Speaker 0

  Thank you. Thanks for having me.

• 25:38Speaker 1

  Thank you. As a reminder, the content here is for informational purposes only. Should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security. And is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see A16c.com/slash disclosures.