Subscribe
Share
Share
Embed
10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.
Follow the Wallet
===
[00:00:00] Kiri Masters: Venture capital firm Andreessen Horowitz just dropped some major insights on their blog about the adoption of consumer AI technology and the implications of that for the rest of us. [00:00:15] The rapid deployment and adoption of this technology over just the past few weeks has had everyone talking. So today I'm gonna share some of the findings that were the most new and novel to me, and those with the [00:00:30] most obvious through line to the world of retail and advertising.
[00:00:36] What they just shared is the seventh edition of their top one hundred Gen AI consumer apps report, and it's the first [00:00:45] time they have included consumer spending data
[00:00:49] Since I know you like to listen to podcasts, 'cause you're listening to this one, you might wanna listen to their recent podcast recap where they had two of, uh, a16z's [00:01:00] partners on. Uh, one was Olivia Moore, who authored that report, and the other is partner Josh Elman, and they talked about it on the a16z podcast.
[00:01:13] So you might wanna go over [00:01:15] and check that one out to listen to the full insights. But let's jump in to some of my personal highlights.
[00:01:23]
[00:01:24] Kiri Masters: So number one, AI is an essential daily utility [00:01:30] that barely anyone actually wants to pay for. The share of Americans who pay for AI has doubled in a year, but it is still a very small percentage of overall users.
[00:01:44] About [00:01:45] half of Americans say that they use AI, and about a quarter think that they use it close to daily
[00:01:51] That's just wild when you think about the amount of time that we've had this technology available. Now, here's what's interesting. Only about four and a half [00:02:00] percent of US consumers pay for an AI subscription themselves. They got that data from a Yippit data panel from August, and this counts paid personal [00:02:15] subscriptions to ChatGPT, Gemini, or Claude.
[00:02:18] Now, there are some other sources out there that put that estimate more at the two to two and a half percent range. Either way, according to any of these sources that you look at, the number of [00:02:30] people actually paying for AI from their own wallet has roughly doubled in a year. And within that group, spending is very concentrated.
[00:02:42] The top ten percent of people [00:02:45] paying for AI out of their own funds bring in roughly half the revenue, and the top one percent bring in about twenty percent. So there is a huge power users effect. The point of me sharing this is just to reiterate [00:03:00] everyone's using it, no one wants to pay for it.
[00:03:02] When people are actually spending money on these tools for themselves, they're mostly going to productivity, development, creative tools, things that are [00:03:15] used to build, make, or sell. Not these, like, everyday common consumer queries asking to identify a bird that someone saw in the backyard or finding [00:03:30] things or comparing stuff, getting relationship advice, all of these use cases that, you know, the general consumer is, is using them for.
[00:03:37] When people are spending money, they are doing it to be productive
[00:03:42] And on the podcast, one of the, um, [00:03:45] the authors, Olivia, passed on this line from one of their portfolio companies, Eugenia Kuyda, who is the f- co-founder of the early AI friendship app Replica, and is now, uh, the founder of the consumer app, agent app [00:04:00] called Waabi, and she said something really smart.
[00:04:02] I love this: "Most people aren't looking to save time. They're looking for ways to spend their time." And that's why social media, [00:04:15] Netflix, th- these are all the most used consumer digital products. Most AI so far helps people do things faster, easier, be more productive, save time, save money, and that's not [00:04:30] as compelling as a value proposition for most people who what's really pulling them towards subscribing for something, it's sport, it's entertainment, it's social
[00:04:42] It's community. Now, [00:04:45] shopping, shopping is one of the few categories that sits on both sides of that line. I wrote earlier this year that not all shopping is fun. Rebuying milk every week is a chore that I would happily hand to [00:05:00] an agent, while indulging in some retail therapy at the mall is time that I want to spend.
[00:05:06] There is, there's all different types of shopping. What I find fun to shop for, someone else doesn't. So while [00:05:15] most AI shopping tools so far have been built for that first job, the h- handling tedious chores job, the second one, the fun part of shopping where people spend time and [00:05:30] attention and it is a form of entertainment and self-expression
[00:05:36] There's been a bit of a mixed bag. Certainly, there are companies out there that are trying to move forward, like the concept of AI [00:05:45] try-ons and things like that. But this is where most people, and I'm still skeptical of this use case too, is like, why would I want to outsource someone buying my, you know, summer vacation outfits?[00:06:00]
[00:06:00] If you are someone that enjoys that kind of thing. So, um, that there are m-many use cases here for different types of people and different occasions. Some of it is fun, and some of it is tedious. The third [00:06:15] insight that I highlighted here is that growing paid subscriptions could stunt growth for these AI companies.
[00:06:23] Now, venture capitalists in consumer technology had this old rule [00:06:30] of growth. Well, the old rule was to build up lots of users as fast as possible, often running at a loss in order to secure a land grab and acquire as many users as possible. And then [00:06:45] the sustainable revenue model comes later with advertising.
[00:06:49] But AI doesn't work that way because the running costs are way too high to just give the main product away for free. And that is why [00:07:00] Olivia at Andreessen Horowitz says that she doesn't want the share of paying users to keep growing from that four and a half percent that I mentioned earlier. So they did this breakdown looking at what percentage [00:07:15] of top web products offer each revenue model between individual subscriptions, usage charges and advertising, and platform and transaction fees, and this is amongst cons-consumer AI [00:07:30] companies.
[00:07:30] They found that about eighty-five percent of apps make money from subscriptions and sixty-two percent from credits or token top-ups. Only thirteen percent run ads. And Olivia calls this an [00:07:45] unnatural inversion of how the consumer internet usually works. Most big tech companies make their money from ads or transaction fees, not from subscriptions, so this is odd
[00:07:57] Now, if the vast majority of your [00:08:00] users won't pay you and each of them costs real money to serve, you have to find another way to earn from them. And that is why I believe that despite some discussion of alternative ways [00:08:15] to fund these consumer AI agent apps, I firmly believe it's going to be ads. I'll share more on, on that in a couple of weeks
[00:08:29] poor [00:08:30] audience targeting is frustrating, but for retail media teams it can be costly too. With Growth Loop's composable commerce media solution, you can turn your first-party data into hundreds of [00:08:45] high-value audience segments and launch campaigns faster. After partnering with Growth Loop, instant commerce pioneer Gopuff scaled from a hundred syndicated audience segments to more than six [00:09:00] hundred, and now it takes less than forty-eight hours to turn around a custom segment for one of their brand partners.
[00:09:08] Learn more about how Growth Loop is powering Gopuff's best-in-class [00:09:15] retail media operations at go.growthloop.com/breakfast. That is go.growthloop.com/breakfast next highlight, I [00:09:30] think a wallet is coming. So OpenAI, as we all know, has run up to a billion dollars in annualized ad revenue really quickly. It could be even higher now 'cause that figure was from [00:09:45] August. And, um, this is the type of milestone that it takes years and years to get to for a publisher or for a consumer tech app.
[00:09:57] And Olivia at [00:10:00] Andreessen Horowitz gave two reasons that it happened so quickly this time. First, there were-- there are just so many users to sell against. Just the usage of this tool was so rapid, uh, so quickly that they were able to sell [00:10:15] against that one point two billion weekly active users. Secondly, ChatGPT just knows so much about each user, so in theory it has rich context, and in theory it could charge more and [00:10:30] see higher conversion rates.
[00:10:32] Now, is that what advertisers are saying right now? I'm not hearing it personally. I'm not hearing so many runaway successes of [00:10:45] brands advertising on ChatGPT. It's super limited right now in terms of the targeting, the reporting, the functionality. Um, but in theory, yes, there is the potential for even better targeting capabilities [00:11:00] than the current kingpins like Meta and Google because of that depth and breadth of conversations that people are having with them.
[00:11:07] And then this is the part in the podcast where Olivia drops this bomb and just kinda leaves it hanging out there. She mentions something called [00:11:15] Login with ChatGPT, and she says, "I think a wallet is coming." Um, so that was just left hanging out there from one of the world's most successful venture capitalist firms
[00:11:26] And I've been thinking a lot about what that would mean, [00:11:30] and where I've got to so far is a ChatGPT wallet would matter the most for retail media because it gives OpenAI proof of purchase. Retailers' biggest advertising advantage, or [00:11:45] one of them, is closed loop measurement. They can show an ad, and that it led to a sale because they see the ad impression and the conversion at the end.
[00:11:54] A wallet would give OpenAI that too, and across every retailer where people can [00:12:00] pay with it And so in my agentic shopping series last year, I argued that once an AI platform sees the transaction, retailer's first-party data stops being this exclusive [00:12:15] thing. And exclusivity is what has been the selling point, especially for off-site retail media to be able to close that loop Now, there is a whole lot more to unpack here in terms of implications, but until there's [00:12:30] further evidence of this wallet coming out, I will save my breath for that one.
[00:12:36] And my final highlight from this research is that AI shopping is still an open field. [00:12:45] The bigger risk here is ChatGPT replacing the retailer's site as the place to shop, because that is a true existential crisis for a retailer, let alone their media business. But that all depends on building a great [00:13:00] shopping experience, which ChatGPT hasn't done so far.
[00:13:04] And to be fair, none of them have. Shopping is still one of several categories where there is no AI-native winner that has [00:13:15] emerged yet. On the podcast, Olivia asks whether they will ultimately live inside of personal agents or as standalone apps. And Josh, one of the other, um, partners from the [00:13:30] company, says that shopping really needs visual browsing, not a chat window.
[00:13:34] You wanna see options, you wanna explore, you wanna have things customized for you
[00:13:38] And as we know now, these new personal agents are running into retailers' walls. [00:13:45] Amazon famously shut off access for Meta's new Muse agent almost immediately after it was launched. But at the same time, other retailers like Shopify, Instacart, other, uh, like OTAs, like Expedia, [00:14:00] OpenTable, they have started building official integrations.
[00:14:05] And, uh, what I've talked about previously is that Amazon can play this game of agentic commerce chicken and block everyone [00:14:15] because it can afford to. It has the traffic, it can hold out against agents. It could also, in the future, decide to reverse that decision. Um, but many other retailers don't have that enormous traffic [00:14:30] and, advantage and, the frequency of purchase- purchasing and the logged in data that Amazon gets through its Prime membership base So if you don't have all of that, then you need to think very [00:14:45] carefully about blocking all of these agents.
[00:14:49] I appreciate that many of these available options, none of them sound particularly attractive, but lots of things to unpack here for [00:15:00] retailers thinking about what this means for their business, for their media business, but also for their retail business at its core as well. Well, that said, I was hoping for this to be short to capture some of my [00:15:15] highlights from this research, but, uh, there really is a lot to dig in there.
[00:15:19] I, I do recommend checking out that blog post. We'll link up to it in the show notes. Also, if you prefer to listen, find that episode where they recap this research. Thank you [00:15:30] for listening. Have a great weekend, and I'll catch you on Monday.
[00:15:34]