About
I am an Associate Professor of Marketing at the University of Maryland's Robert H. Smith…
Articles by Daniel
Activity
21K followers
-
Daniel McCarthy shared thisVery nice piece on the growth of the online sports betting market by Matthew Suh, particularly with Gen Z. Everyone is probably aware of how the market has been booming but I think this video does a great job of showing how it has boomed even more so, specifically among Gen Z'ers. Some of the stats he mentions: -58% of 18-to-22 year olds placed a bet over the past year -67% of students living on campus are active bettors -52% of Gen Z investors took money originally earmarked for investing and used it for sports betting -26% of Gen Z'ers conisder sports betting to be a part of their long-term financial strategy, way higher than any other generation Nuts. For more on how sports betting is deepening among consumers (overall, not specifically among Gen Z'ers), see Kenneth Wilbur and my (newly updated!) article on this: https://lnkd.in/gVWegQgK Entertainment is one thing, but it's quite another when it is viewed as your long-term financial strategy, when you're taking money that would have gone into the S&P and diverting it instead to betting on sports, and when it grows to represent a substantial percentage of your income. https://lnkd.in/gXKvXPYPAfter the Launch: Trial, Repeat, and the Income Incidence of Online Sports Betting LegalizationAfter the Launch: Trial, Repeat, and the Income Incidence of Online Sports Betting Legalization
-
Daniel McCarthy shared thisIn San Francisco for OpenAI DevDay tomorrow, and today I got to visit the HQ with the ChatGPT Pro community. What’s impressed me most about OAI is how closely the team listens. Several times now, I’ve shared feedback and seen it reflected in the product shortly thereafter. I have not interacted with a more customer-attentive company, which, for all the people who have heard me go on about customer centricity, is striking given how product-driven the frontier companies have to be. Angie Luo, Kevin D., and Pei J. have been wonderful advocates for us, and meeting them (and Tibo! and Joseph Bartlett and Michael Wall!) after all those Slack conversations was kind of surreal. Of course the best part is that Sarah Rose Siskind from the community even made a comic book featuring us all as superheroes, which they were handing out. Apparently I’m “Doctor Data,” although I’d like to go on record that the data is not doctored!! Great way to kick things off in advance of the official programming at Fort Mason tomorrow
-
Daniel McCarthy shared this🚨 WORKING PAPER ALERT 🚨 Really excited to share “Paying for Itself? Value (Mis)Alignment in Subscriptions” with Shin Oblander, Meng Yang and Young-Hoon Park. We study ~half a million subscribers across eight food delivery programs, including DashPass and Instacart+. The big question we tackle is when do consumers come out ahead, when do platforms, can both win at once, and what contextual factors inform these outcomes? Every dollar a subscriber saves in waived fees is fee revenue the platform gives up. Extra spending has to generate enough profit for both sides to win. Subscriptions increased spending across all eight programs, but thin margins left little extra profit to share. At our baseline contribution margin of 3% of consumer spending, just 6% of matched subscription relationships delivered BOTH subscriber savings and direct platform profit. Even if platforms could wave a magic wand and choose who enrolls, they would raise platform profits by selecting people who would lose on their subscriptions. Many consumers also keep paying despite substantial losses. Monthly subscribers with the largest accumulated losses cancel at roughly the same rate of those near break-even! Some practical takeaways: 1/ Before signing up, look back over your order history. Would this subscription have saved you money, net of subscription fees? Using the six months before program launch, 79% of eventual subscriber relationships that passed this test came out ahead (and were > 2x more likely to sign up in the first place), versus 39% of the others. A “no” doesn't guarantee a loss, but the odds aren’t good. 2/ Monthly buys flexibility. For annual plans, savings decline within the prepaid year. If usage fades, you're still on the hook. Annual discounts can be good for consistently heavy users but monthly plans allow early exit. 3/ Simple stopping rules can act as insurance. On a monthly plan, consider canceling for good when total savings since signup fall below total subscription fees paid the first time. Applied to our monthly plan data, with purchases unchanged, this would have cut total lifetime losses by about 80%, while average lifetime net savings fell from $104 to $99, only like 5%. A case for savings alerts and quicker cancellation. 4/ Scrutinize “pays for itself” advertising. A few big winners drive up the average. In the full sample, monthly net savings averaged $6.24 per subscription relationship, while the median relationship LOST $1.35 a month. Claims should show the median and the share who recoup their fees over a stated period, alongside averages. 5/ For platforms, more spending isn't enough. Creating more room for both sides to win may require better margins or benefits consumers value that cost little to provide, including benefits funded by merchants or partners. This one was a lot of fun to write. So many threads to tug on. Some weekend reading for anyone interested in the economics of subscriptions. Paper link in the comments. Enjoy!
-
Daniel McCarthy shared thisAlways love to start the week off with a nice C3 (customer cohort chart)! Well, for the most part. They provided a tabular cumulative version of it. But I'll take it. Moneyview’s IPO filing tells an annuity-like story in which they acquire borrowers over time, and the resulting borrower acquisition cohorts keeps generating repeat lending over the next four years (albeit with what appears to be some weakness in year 4). Newer cohorts are also generating more cumulative lending at comparable ages. A lot of other good disclosures in here - total users and monetized users, marketing efficiency measures, and a cohort-level "registered users monetized over time" figure (basically a free-to-paid conversion-type figure). Easy to see why companies share this as it helps prospective investors understand the value of their customer base. Great to see this kind of customer disclosure in company filings! For an Indian lending platform to boot - CBCV is a framework for any firm, not just US-listed ones. Prospectus: https://lnkd.in/e9GNBJrb CBCV article with Peter Fader in HBR, IPO disclosure article with Aswath Damodaran and Maxime Cohen, Theta C3 article, and original CBCV article in the Journal of Marketing are below.
-
Daniel McCarthy posted thisOne thing I often say is that revenue from customer acquisition tends to be more volatile than revenue from repeat purchasing. ODDITY, the beauty and wellness company behind IL MAKIAGE and SpoiledChild, is an interesting example of this. In Q2 2026, net revenue from first orders fell approximately 40% from a year earlier. Net revenue from repeat orders fell approximately 20%. So the mix of orders skewed towards repeat vs new, but only because repeat went down and new went down even more. This is part of the reason why I'd discount the value of a firm that is very reliant on customer acquisition, all else equal, vs a business that is getting most of its revenue from repeat orders. The latter is steadier more predictable revenue (it also says something about product-market fit). Management says this was due to an advertising algorithm change that pushed people away from "try before you buy", and that apparently they ended up in worse ad auctions for pricier prospects. It says that ads can prompt existing customers to buy again (so the algorithm problem hurt repeat orders), and fewer customers acquired earlier in the year meant fewer subsequent repeat purchases. So for one, this ties into something that Peter Fader and I had many heated discussions with equity analysts about regarding Wayfair back in the day. Some of the sales and marketing is for acquisition, the rest is for repeat, and they should feed into CAC and the contribution margin, respectively. So yes, we wholeheartedly agree that you don't want to dump it all in CAC. But no, it's not a free lunch, it just means you spend less to bring 'em in but make less profit as part of the ordinary course of business. There is also a trap in focusing on the repeat-sales mix. So many businesses hype up the % of sales or orders coming from repeat buyers. Yes, good to know, but as we see here, repeat revenue can become a larger share of the business even while repeat dollars are falling, if new revenue is falling harder. Decompose and look at these quantities separately. Also speaks to the value of test and learn. It sounds like they rolled out this algorithm change prematurely. Sometimes you can't run small tests, but here I don't see why they couldn't have. Experiment! Lots to learn from examples like this. CBCV is such a valuable lens to think about these issues through. Links below.
-
Daniel McCarthy posted thisA weekly ChatGPT Finance email helped me cut over $1,150 a year in subscriptions I no longer needed. It also introduced me to an absolutely ridiculous cancellation process at LogMeIn Pro. The first sweep took about six minutes… Canva, Business Insider, Paramount+ and Cinemax through Prime Video. Grammarly was next. ChatGPT put the recurring charges in one place, helped identify mystery subscriptions, and found the cancellation links and email addresses. That was enough to get me over the hump. The other cancellations were remarkably easy. But then, sigh, there was LogMeIn. I couldn't access the account management page on my phone. On a computer, the subscription page offered one- and two-year renewals but no cancellation button. Eventually, I found a form that generated a support case. Their email then told me to call within 72 hours or the case would automatically close. I'd submitted a cancellation request, and now I had a deadline to call them about it. On the call, they asked for the case number, billing address and last four digits of my credit card. Pardon my French, but give me a f’ing break. I was trying to stop a $349.99/year subscription from renewing. I was surprised this was still happening, given all the talk about making subscriptions easier to cancel. The FTC did adopt a “click-to-cancel” rule, but a federal appeals court struck it down in July 2025 over a procedural failure in the rulemaking process. Maryland, where I live, also enacted its own law, effective June 1, 2026. For covered subscriptions, cancellation must be as easy as signup and available through the same channel. If my subscription falls under that law, I wonder how a mandatory phone call is legal. A lot of great discussion around this sort of roach motel tactic by Robbie Kellman Baxter. Terrible for word of mouth and reacquisition, even though it may squeeze an extra payment or two from unsuspecting people like me.
-
Daniel McCarthy shared thisNice customer cohort disclosure from Funding Circle! Its FlexiPay and credit card business grew revenue 83% in H1 2026, with 90% coming from customers onboarded before 2026. The chart shows borrowing balances by customer vintage, so we can see how those relationships develop over time. What a nice chart. You can see the stability of business over time. They bring in a new cohort of customers and the business from that cohort is like an annuity. Not expanding like SaaS, but hey, this isn't a SaaS business. They also do a good job of laying out the customer value logic. Acquisition costs and expected credit losses upfront, but then they get repeat revenue over time, and an illustrative 12–18 month payback. This is why looking at customer economics alongside current profits matters. Same as in SaaS. I'd love actual cohort profits, acquisition costs and retention too. Persistent balances alone don't establish attractive economics. Still, it's great to see customer cohorts and LTV-adjacent calculations. More of this, please! Sources in comments including more on the virtues of the C3 from Theta. It's my favorite single customer-related disclosure.
-
Daniel McCarthy posted thisŌURA filed for its IPO and WHOOP says it is next. I was a longtime WHOOP user (only stopped because I kept losing them for some reason... I now use a POLAR Vantage V2 for workout protocols). Needless to say, I watch this space closely! This was a very solid S-1 filing. There wasn't a C3, but was a nice cohort disclosure: weighted-average month-12 Paid Member retention was 85% (by vintage, 81% in FY23, 85% in FY24 and 87% in 9M FY25). But 85% is still month 12, and we don't get to see the tail. Retention often goes up with tenure, and FY23 cohorts were mature enough to show more. If the tail were great, one imagines Oura would show it. Mild red flag. Now the fun part, unit economics! Ring 5 is $399-$499; Ring 4 starts at $349. Filing data imply about $145 of hardware gross profit per ring-equivalent after allocated cost of revenue. So the Q is then, what's CAC? I'd treat all $258M of S&M as acquisition spend. Some may support retention, but once you own the ring, the product, not another ad, should drive renewal. They don't directly disclose adds, but we can estimate it as 3.1M rings sold x 89% first-ring x 94% paid conversion = 2.59M (all disclosed). This implies CAC of $99, less than that $145 of hardware profit. The ring covers it 1.45x up front. Noice! With 40% organic acquisitions (disclosed), assign S&M only to the 60% nonorganic and CAC is $166. Hardware falls $21 short, recovered in four paid months. This denominator issue was central to my Rent The Runway analysis and ultimately mattered. Not as big a deal here thankfully. Link in comments. Value after acquisition: I use 87% retention in year one, 85% thereafter. At $69.99/year, 89% gross margin and a 15% discount rate: 6.8 expected billed years, $476 membership revenue, $424 undiscounted GP, and $211 PV gross profit from the post-acquisition membership stream. Add $145 ring GP: PAV is $356. Subtract $99 CAC and estimated gross-profit CLV is $256, a 3.6x LTV/CAC. Solid! "Nonorganic-only": $190 and 2.1x. Still good. Means marketing can drive profitable growth too. I wish we had gross adds, true CAC, later-year retention and cohort contribution costs. Still, the **pre-IPO** Peloton-like dream is plausible: hardware funds acquisition; recurring membership is upside. Spreadsheet in comments to mess around with these #s yourselves - those who know me will recognize this is the PTON spreadsheet adapted to Oura 😄 This sort of CBCV analysis is what we live and breathe at Theta - check us out! The disclosures themselves are a testament to their importance, otherwise, why disclose them at all? Great to see. Relevant links in the comments.
-
Daniel McCarthy shared thisAstra is great, but man oh man has OpenAI's Astra Ultra been a game changer for me. It feels like what Sol Ultra was always aiming for. It does big things, but does so efficiently. It just works. For our paper “Expanding Markets or Capturing Share? The Competitive Effects of Subscriptions on Spending in Restaurant Delivery” (conditionally accepted at the Journal of Marketing Research - woo hoo!), I needed a synthetic-data replication package that jointly reproduces all of the main findings and is not too far off in general. Basically, you could think of this as a monster moment-matching problem which is a bear to do with a generative model of the synthetic data. I’d spent weeks on this with Sol Ultra, using Pro to rein in the over-engineering. Over 8 billion tokens I'd gone through. Within a DAY of switching to Astra Ultra, we’d largely cracked the core analytical problem. There was still a substantial amount of iteration and auditing afterward to ensure that everything was formatted and presented properly, but that breakthrough was remarkable. This dovetails with my last post about that Zitron article. So many uses are now practical that were not. 0 to 1. Man, what a model. Congrats to the team. https://lnkd.in/eeTj-eDN
Daniel McCarthy commented on a post
9h
I just hope I don’t become a villain in part 2 of all this 😂
Daniel McCarthy replied to a comment
13h
Taimoor C. I have no idea on what basis they can actually feel that way. Somehow I suspect the messaging they are exposed to, although maybe I am giving them too much credit
-
Daniel McCarthy liked thisDaniel McCarthy liked thisOne of the best parts of OpenAI DevDay this week had little to do with what was announced on stage. I was invited to be there as part of the OpenAI Pro Community, a small group of users who get early access to products so we can test them, break them, and give feedback. But I mostly wanted to go because I had never met many of the people in this community in person. The group includes an immunologist, a comedian, a journalist, a physicist, a mom-preneur, a farmer, a musician, a few professors, and people from all sorts of fields and backgrounds. Everyone seems to be finding completely different and surprising ways to put AI to work. I spend a lot of time around AI, but being around this group still gave me a long list of ideas I had never considered. Thank you Kevin D., Mamie Rheingold, Angie Luo, and Pei J.!
-
Daniel McCarthy liked thisSuch a fun project making a graphic novel for the openai "Super Users". Welcome, Doctor Data! - The Token Joker 🦹♂️
-
Daniel McCarthy liked thisDaniel McCarthy liked thisBreaking news: our field experiment with The Washington Post shows that generative AI search can diversify collective attention while giving readers more information in common, helping pop the filter bubble 🤩 Read more here https://lnkd.in/gJKG9tir Brought to you by Lee, Lee, Lee, and Lee! Heeseung Andrew Lee, Dokyun Lee, Gwanhoo Lee, Dongwon Lee Thanks to The Washington Post team!: Eui-Hong (Sam) Han, Patrick Haney, Anuradha Uduwage
-
Daniel McCarthy liked thisDaniel McCarthy liked thisthanking all y'all for the wonderful birthday wishes, I'm highlighting Ashwin Malshe अश्विन मालशे and the core team celebrating the day with me at Sam Adams for #MarketingDynamics few years ago, and Amy Jo's reminder that age is just a number...and a better opportunity to #BQ !
-
Daniel McCarthy liked thisDaniel McCarthy liked thisI’m at OpenAI DevDay today and there’s been a flurry of announcements. The most important ones for media and creators: dots, plugins and MCP. If you work in this space and haven’t been paying attention to this shift, well… you should. They all point in the same direction: agents becoming a new interface between people and information. They can monitor, retrieve, recommend and connect directly to services and content. OpenAI is far from alone. Meta, Microsoft, xAI, but also Instinct and Poke are all pushing toward the same idea: persistent PAs that people can delegate to without thinking about the underlying agent infrastructure. Agents for normies. Which means the next battle for distribution may well happen inside a user’s personal AI assistant. Wrote more about what that could mean for media, creators and publishers on my Substack https://lnkd.in/gf7EAMV6
-
Daniel McCarthy liked thisDaniel McCarthy liked thisHi folks, here are some updates about the Frontiers in Service Conference. First, my heartiest congratulations to the folks at Hanken in Helsinki, Finland, who did a stellar job of hosting the 2026 Frontiers Conference. I am proud to report that the folks at Hanken gave me the Christian Grönroos Service Research Award--a lifetime achievement award in the service field. That was especially meaningful to me, given the importance of Christian and others in the Nordic countries in launching the service research field. The photograph shows me receiving the award from Christian. Also in the photo, from left to right, are Professor Johanna Gummerus and the department chair, Professor Maria Holmlund. The 2027 Frontiers in Service Conference will be held July 8-11 at the University of Maryland, preceded by the AMA SERVSIG Doctoral Consortium. The conference website should be up by November 1.
Experience & Education
Recommendations received
2 people have recommended Daniel
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content