Ramp economist Ara Kharazian on how to go viral on LinkedIn

Transcription

Daniel: I don't know why everyone looks so buttoned up and high and mighty, like I'm the best employer of all time on LinkedIn. I got hate messages on LinkedIn. Really? I knew that one was going to do well, like getting a 21. I like knew this is about as good as a result as we're going to get. There's no public data set to answer one of the most basic questions people had around that time, which is how much should I tip? New York Times column was too limiting for him in two specific ways. One, charts in many mainstream publications don't say hello. They look like they're designed for someone who's not supposed to understand it. If you're writing about data and economics, you better have a fucking chart. I don't post that and I still do pretty well. I do think people are very hesitant to post on LinkedIn because it is perceived as cringe. And I would say, don't be cringe. Matt: Welcome to this episode of the Connection Accepted pod. I'm so excited today because we have a really great guest on. We have Ara from Ramp. Ara, thanks for joining us. Ara: Thank you for having me. Matt: So Ara, for the audience's benefit, can you just walk us through high level your background, where you're from, where you went to school, and what led you to Ramp today? Ara: Yeah. I'll start from present day and I'll kind of start a little bit broad, get specific as I can. I work at Ramp. I'm an economist at Ramp. I run Ramp Economics Lab, which is our equivalent of a think tank. So we have first party data from Ramp. Ramp is a corporate card and expense platform. If you think about the data set we have access to, it's anything a business might spend money on except for more or less payroll. So we have credit card transactions from those businesses, but we also have everything that shows up on the receipts. And then we also have bill pay. So that means we have a really great data set that tracks where businesses are spending money, where they're investing, and where they're pulling back. And that's a pretty important data set because there are a lot of companies, private sector companies, but then also government public data sets that do produce research into where the economy is headed. They mostly focus on consumer spend. Business spend, though, is a pretty significantly leading indicator of where the economy is going. And it's a data set. It's a sort of transaction set that is typically private. It's not covered very often. The private sector companies that have access to it don't often publish that kind of research. And also many people just don't have it. There really is no public data set that captures where and how businesses are investing. Matt: And if you are interested in where tariffs are headed and how they're affecting businesses, or you're interested in whether or not there is going to be a recession or a manufacturing boom, one of the places you might want to look is, are businesses spending money in the places where policy suggests they might spend money? How do we confirm these things that we hear about in the news without a data set? So that's what I do at Ramp. I publish research and post about it using first-party data from Ramp and sometimes public data as well. Daniel: But the path to this kind of job is not that standard. I was always into economics. I usually say something like, I came of age when 538 was a big thing. If anyone remembers 538, it was Nate Silver's statistician who sort of made a big on political election tracking and poll tracking. News website, news blog that covered economics, politics, and sports in a data-driven way. And it was like the first site that did what you now call data journalism in a very effective way. Eventually Disney bought it and now it more or less has been shut down. But all of the reporters who worked on 538 have now gone on to like great careers working at, you know, everywhere from the New York Times, Nate Silver's independent work. Matt: I was obsessed with 538. Like it genuinely did economics and data journalism in a way that I thought was smart and presented data in a very interesting and exciting and effective way without dumbing it down the way a lot of people do it. And everything was focused. Every piece of content was focused around one beautiful chart. And if there's anything that I'm inspired by now that I try to produce content to match or anything that I look up to, it's 538 and trying to recreate what I really liked about 538 when I was reading it. But you can't graduate from college and go for a job like that. Cause unless if you're doing something independently, maybe now, but I think I had more to learn about economics and writing in general. Daniel: Yeah. So after college, I worked in economic consulting for a long time and then switched to a normal data science job. And then I ended up at Square FinTech company. And it was around the pandemic where there was a big appetite from the public to understand how businesses were responding to certain post-pandemic trends. You can imagine restaurants, for example, were having significant labor shortages. Matt: We all saw that. Daniel: Yeah. And there wasn't a lot of data about how restaurants were responding to things and how consumers were responding to the post-pandemic labor shortage. Tipping, right? It was a big story around that time. Daniel: And there's no public data set to answer one of the most basic questions people had around that time, which was how much should I fucking tip without being in a way that maintains the social contract but is not offensive, right? And so a lot of reporters would come to us asking how what Square data showed about tipping or wages in the United States or how restaurants and workers were addressing the labor shortage in their own side of the labor market. And I pitched this job to Square to say, hey, let me focus on producing economic research using Square data and putting it out. I think this is going to be really exciting. And we started putting out that kind of data regularly, answering questions about what the average wage was in this or that town for a restaurant worker, data about tipping and what you could expect in your state or your type of business. And it did very well. It was called Square Payroll Index. And then we also put out regular reports on our website. And then I started posting on LinkedIn around that time and developed a fairly small, but I think dedicated following on LinkedIn. And people saw it, particularly people both in the restaurants industry, but also people who were just interested in data and economics and interesting and compelling charts with interesting data sets. And what was really unique about it was that you could not get it anywhere else. There was genuinely no, there was no public data set that captures month over month changes for that level of specificity. Like you can track average hourly wages for the typical worker in the United States. The government puts that out and it's a great data set. But if you wanted to get really specific, like what should a restaurant worker make? What can a restaurant worker expect? Are they getting a competitive wage at their current job in Wichita to New York to Seattle? There is no way to look that up unless if you have a private data set that some guy on LinkedIn is posting about. So that was the niche I filled. And it was a great data set too. So beyond that, we would also post a bunch of other fun stuff. Like when the eclipse happened a couple of years ago, one of the great posts that I did at Square was just showing this strip of activity across the U S that showed sort of an explosion of restaurant activity in otherwise pretty rural, quiet areas of the U S that happened to be in the path of the solar eclipse. And making a really beautiful chart about that that showed something that was both surprising and not surprising at all, but was pretty. And that did very well on LinkedIn. And I think as a testament to what I, what I think really makes this kind of work work is even if you're posting something obvious, like, Oh yeah, people wanted to see the eclipse and spent money around the path. Daniel: There's something that makes people feel good to see their own activities or behaviors or even fun behaviors reflected in an authoritative data set or be confirmed or validated by someone they perceived to be authoritative. Yeah. I think about Peter Walker, the head of insights at Carta. We interviewed him a week or two ago. He also has a very similar approach in that he has access to all this data about startups and funding and everything that Carta has. And he basically looks at the data, finds a interesting trend or observation, and then makes a post with a really nice infographic as well. And I feel like it's similar to what you're also doing with ramp spend data, which no one else really has that level of specificity. So I, when I was a consultant at BCG, we actually, funny enough, used to look at credit card spend data for some consumer projects. And it was not at the level of specificity that you guys probably have, but it was always still very interesting for me to get to look at it. And it would give you a directional sense of like consumer spending patterns, things like that. And I think hedge funds may use it as well as like an early indicator before earnings. But yeah, like you guys have a data set that's so valuable, which is what are companies actually spending on? What categories are they spending on? And I feel like that enables you to put out such interesting research. Matt: Yeah. It makes the work methodologically very easy too. If you talk to people who work with credit card transaction data sets, they often know that there is a skew and a statistical bias to that data set. Not every business spends on a credit card. Businesses that spend on credit card are usually spending a certain type of spend on a credit card. That data is often messy. It's not even a bias thing. It's just not something that you can easily discern. Ramp data also has its own skew and bias, right? Because you're doing a lot of cherry-picking and just doing data practices that don't make sense. And the second reason is that it takes a long time. It genuinely is not easy to make data fit a narrative. Like, if you are going to be unethical with respect to your data practices, it is one of the most unpleasant experiences of your life trying to find all the weird cherry-picking and filter behaviors you need to do to make something fit. It's so much easier to just do the work in a way that is honest and that you think is appropriate and right and makes all the fair assumptions, and then say what you think is most appropriate based on that data. And you can still make that work fuzzy. Like, let's say your priority is getting a really banger LinkedIn post. Like, you can still do that. Like, there are charts that I think are, like, incredible findings that don't do well on LinkedIn. Daniel: And there are charts that I produce that are so obvious that I think no one could possibly think this is interesting that end up getting hundreds and thousands of impressions and press mentions. It was one chart. 996 is this trend that I think originated in China and associated with Chinese working culture, but is now it's kind of emerged as, like, a trend in Silicon Valley working culture, which is like, work from 9 a.m. to 9 p.m. six days a week. And people have been vaguely talking about it over the last year as something that's caught on in a lot of tech companies in San Francisco. People are working more. That's the idea. People are working more, and they're working more on Saturdays. And so, short answers, I produced a chart that showed people are working more on Saturdays. Essentially, what it showed is just that if you look at ramp spend, which is corporate card spend, and you look at where, when and where people are expensing like a weekend meal or something, it's happening a lot more in San Francisco, and it's happening a lot more in San Francisco on Saturdays and Sundays. And it's not happening really anywhere else. Uniquely to SF, it's not happening in New York. It's not happening in other tech hubs. And I showed it in this chart that I think very compellingly showed, like, oh, wow, people are really doing this on Saturday. It's actually still relatively modest of an effect, but relative to the size of a large metro area, but it's definitely something that is happening. And it blew up on LinkedIn to the point that, you know, it did well on the social media platform, but like, even, it's now been a little over a month maybe since I posted it. And even over the last two weeks, we've gotten media citations from large traditional media outlets. Like, New York Times wrote about it. Bloomberg wrote about it. Everything from, like, small tech blogs to the big newspapers. It was in print. I wasn't, this is one of the rare stories where I knew that was going to do well. A lot of times when you produce something, you, I really am not that confident that it's going to do well online. I knew that one was going to do well. Like, when I found this chart, it was like hitting 20 in black. What's the good blackjack score? What do you want? 21. I don't gamble. 21. It was like getting a 21. I like knew this is about as good as a result as we're going to get because you don't see, it's really rare to get a, and I barely worked on it. Like, if you look at the actual metric, it's not that complicated. It's more or less just looking at the share, shift share in transactions that occurs by hour of the week. I mean, there's a couple other ways I sort of did it, and it shows up in other ways, but the chart I made shows it most compellingly. And it's one of those things that taps into a couple really compelling things. People really like seeing how, again, their daily habits are reflected in data. And that did that very effectively. Daniel: And second, there's something almost like controversial about it. Like the idea that tech workers, first of all, controversial, people work in tech. Who would have thought? And then that, oh, they're working more. And oh my God, it's originated from Chinese working culture. Like there's a foreign policy aspect to this story. Like people love that stuff. And not to be glib about it, right? Because there's a lot about this that I think sparked this larger conversation about whether or not this is healthy. That was one of the main questions I got after this. I got hate messages on LinkedIn because I posted about it. And by the way, you can look at my post about it. I was not suggesting that it was good or bad. I was really posting about it, having heard about this trend. Matt: And you're totally neutral. Daniel: Yeah, I was hardly neutral. If anything, I was actually kind of poking fun at it. Like I was making fun of the fact that like, oh, there's this, have you heard of this trend? Like where people think they have to eat meat all the time and get married early. And also somehow that translates into working late on Saturdays. Like I don't look at 996 and think it is a logical series of thoughts. I don't know why the eating meat thing is part of it. If anyone's listening to this, by the way, that's part of the 996 thing is apparently you have to eat meat all the time. Matt: Yeah, that's part of it. Daniel: I don't know why. So there are parts of this that I think are a little absurd. Anyway, you post about it on LinkedIn. And it was one of the posts that LinkedIn featured in its weekly news digest email to everyone who uses LinkedIn and has not unsubscribed from LinkedIn emails. So it went out to hundreds of thousands of people, most likely. And so I got messages from random people that I have no connection to or not in the, and not in my typical audience, people who are retired, giving me life advice and telling me like, hey, I can quote, I have one literally right now that screenshot was like, Hey, I'm an older person. And it was literally like, hey, like, I just want you to know I've had a long, a long life. And I've realized, you know, I just want to make sure you know that if you really want to be happy, point one, point two, point three, don't work that much. I'm like, I wasn't saying I'm like, I'm not doing 996. But people don't see that on LinkedIn. Matt: Right. They, they don't. People really want to give you advice. Life advice is what I've found. Daniel: I will accept all forms of life advice. I just, you know, I don't need it on 996. Like people, people, those kinds of controversial topics just really invite people to share their opinion. And one thing I've realized from this kind of job is that you really can't do it if you are going to be extremely worried about how this or that person might misinterpret what you say because it's just going to happen. Matt: Like if you, if you really are, if you really can't take like almost like the, the occasional overly critical comment online, this is, this would be a really hard job to do. Daniel: A hundred percent, man. So many things that I resonate with you. And I completely agree. I think anyone who's ever gone viral or blown up on social media will tell you that the law of large numbers, there's just always going to be someone out there who doesn't like what you're doing or saying. And in your case, it's not even like that. Those folks are hating necessarily. They almost wanted to give you advice, unsolicited life advice. Matt: Yeah, I got hate stuff too in the comments. I had several people in the comments suggesting that I was a corporate shill perpetuating the idea that we need to all work 996 schedules. And, you know, it's funny because it's not even a, like I was just posting a chart. Like, so it's funny because imagine if I really talked about my beliefs politically or otherwise, right? Like not that I have crazy political beliefs, but you can imagine how social media and the internet and not just social media, the internet, people's nature to sort of see something and feel compelled to respond to something, even if uncontroversial might be hard on people who aren't used to it. Daniel: I want to shift gears to talking a little bit more about LinkedIn. I know you said earlier you started posting when you were working at Square. Why did you initially start posting? Matt: I thought I had unique and interesting data worth sharing and I had a lot of different avenues at Square to post, but, you know, I mean, my job was public facing. Like at Square, at least I reported to the comms team. Like I reported to the team that worked with reporters that has reporter connections that we oftentimes did share with reporters or we'd publish my findings in a press release. And that was great. Like I think it's great to work with the reporters and have, you know, your work presented in a nice press release format, but it doesn't work for every piece of content and it kind of slows you down. There are a lot of analyses that I've had produced in research and writing that I had that I wanted to be able to post in my personal voice with my own thoughts and analysis about it that would look odd in a press release. First of all, it would look odd in a press release for me to say, I think this because just press releases aren't good at that. And second of all, press releases don't have the distribution that I want from social media. So if social media kind of has that benefit. And third of all, it's really fast. So when you're producing something in the fast and stuff like that, right, syndicated in a variety of ways, it necessarily requires a level of review and seriousness that I don't think is relevant or helpful for ad hoc analyses. So it slows you down. Daniel: I mean, there are other people in similar positions, I mean, different positions, actually, who I've had similar stories, like Paul Krugman, the sort of iconic former New York Times columnist, Nobel laureate in economics, famously moved his column to Substack about a year or two ago. He had the great economics column and the great job in economics, right? He just posted a daily blog. And for the New York Times, imagine the distribution, right? And he's Paul Krugman. He supposedly has a lot of power to do whatever he wants, and he did. But when he talks publicly about why he moved to publishing independently on a newsletter platform like Substack, he said that over time, he started to feel that the New York Times column was too limiting for him in two specific ways. One, because you're posting, publishing in the New York Times, there's this sort of voice that you need to use and this editing that you need to go through that he felt both slowed him down and removed the part of his writing that he thought was very important, which is just almost his casual nature. And the second thing is that he couldn't post charts as effectively in the New York Times. Like his column was a text-only column. And if you're writing about data and economics, like, you better have a fucking chart. He couldn't do that. He's Paul Krugman. So he moved to Substack, which has a multimedia function. And, you know, you look back at that kind of decision, it's like, yeah, it's so obvious. Why would he not do that? You know what else doesn't really have space for an interactive chart? A press release. You know, a press release, even if I were to include a photo of a chart, like, it couldn't be interactive. It couldn't have any of the... It couldn't have a button that says download the data. This is another example of this is of how I approach presenting data on LinkedIn and otherwise. Actually, one annoying thing about LinkedIn is that charts can't be interactive anyway, like they're images. But at least you can have a chart that is compelling. There's like that Steve Jobs movie. I think it's the one with Michael Fassbender, because there were two, right? There was the one with Ashton Kutcher, and there was the one with Matt Michael Fassbender. And it came out at the exact same time. And the Michael Fassbender one is better. But there's this funny dramatization in the Michael Fassbender one of when they're coming out with the Macintosh, and they're like running up on time and they're still having glitches because they can't get the Mac to start up and say hello. And I don't know how much of this is true and how much of it was just effectively dramatized. It sounds like it was true, but there's this really funny scene where Michael Fassbender is just over and over again saying, like, the Mac has to say hello. And people are like, why does the Mac have to say hello? And it's like, because computers are boring and impersonal and people don't get them. Daniel: And once a Mac says hello to you, it's immediately understandable and inviting and it almost lends credibility. And I think the same about the data that I post, where economic data, the way people talk about economics, charts in many mainstream publications don't say hello. They look like they're designed for someone who's not supposed to understand it. They're not inviting, they're not interactive. And there are very subtle ways that I try to add features or call-outs in my charts or just small annotations that speak directly to the reader and tell them exactly what the takeaway is. Anything as simple as a get the data button, which is on all of my interactive charts, that allow you to see that, oh yeah, this isn't just someone put it on a chart, there's transparency involved. There's an exchange that I can have with this. That's the equivalent of the Mac that says hello. Now, I don't want people to listen to this and be like, oh, this guy thinks he's Steve Jobs because he has like a funny title on his chart. Matt: That's not what I mean. Daniel: That's a little bit what I mean, actually. I mean, it makes sense, right? Like economic data, not even just economic data, but I do think mainstream publications don't always do the best job at portraying the data in a way that the average reader would understand or find interesting. And I think the fact that you're thinking about that is part of the reason why you've seen so much success with the stuff that you do post because you get it and you understand what it's like to be someone on the other side. You know, sometimes people ask me how I do this job, like what training I need to have. And there's no training for this kind of job. I really do think about it. The most important experience in my life that allows me to do my job is my experience as a consumer. Like when I am producing a chart, I am thinking about my own interactions day-to-day in this economy and what is unique about them and what is interesting about them and what I can find in my own data set and uniquely tell that story. There's that side, me as a consumer and trying to see what kind of, how many of my own habits are reflected in this data set that might be compelling to others. And the second part is I'm also a consumer of the news. And I am constantly trying to think about what I like about how some people tell stories and write about otherwise complex economic topics and what I don't like about how some places do it. Like I have in my head work that I admire and work that I don't admire and work that I'm trying to emulate and be inspired by work that I'm trying to build on and work that I very much want to avoid. Even work that like, you know, appears in like very well distributed avenues and other people on LinkedIn here or there, right? Like I have a sense of what I think I want my data to look like, my research to look like. Daniel: Let's talk about your process when you say you've got an interesting data set and you have a chart, or maybe you're just at the idea stage. What's your process to actually create that LinkedIn post? Matt: Well, first, I'm not on, it starts without thinking about LinkedIn. Like my first step of the process is, well, there's a couple of things, right? So you have content that you can produce from like a regular series. Like I have data sets that I produce regularly, like Ramp AI index tracks, business adoption of AI models and tools. And that's just one data set that I can break out into a million different ways and publish findings based on model company or based on sector or just based on overall trends. And it updates every month. So I have that steady stream of content I can always post about. Matt: Separately, there's content that I can produce that's more ad hoc, that's in response to something I read about in the news or something people are thinking about, a trend that I think I can either confirm or correct. So there's those two avenues and sort of streams of content that can come into my life. Something I post regularly and something that's more ad hoc and requires its own research process. Both will at some point end up on LinkedIn, but I have other platforms too, right? I'm posting on Twitter, I'm posting on Substack too, and I'm posting on LinkedIn. There is something very similar between all of the platforms. Like when I am looking through my results and trying to see what might be compelling, I am starting to, like the first thing I'm revolving around, like what is the result that it can show in a very compelling way in one chart? Matt: Because I don't think people realize this. If you are explaining a technical topic, and this is how journalists work too, you get one technical thing to explain. Like if you have to explain two to three technical things in, in, before you even get to the chart, you've lost your audience. You get one thing. And so oftentimes it's interesting as an economist because there are methodologies that I would like to apply in my work that don't make sense to apply because if I were going to use them and display them in a chart, I would have to explain the methodology in a way that would already lose my audience. Matt: And that I can use what is frankly, a much simpler methodology that still gets at the same point most of the time and shows a much more compelling chart that I don't have to explain because people know it and are familiar with it. Like, I mean, like a funny example is like, you can run a regression. Like a regression is the simplest form of economic analysis. It's not that complicated. I mean, you can make it very, you could be quite complicated and thoughtful about how you do it, but not speak of it, but very simple. And is it, is an almost obvious step to any analysis I should do to run a regression that controls for this or it's a, such a fiscal process that allows you to control for various confounding factors. Daniel: I am never going to publish regression results on LinkedIn because I don't want to explain what a regression is to my audience. So most, more, most of the time I can just publish what a regression is a technical complication of, which is posting a series of averages. That's essentially what a regression is, just more mathematically involved and controlled for. And so I'd rather opt for that. So, so that's the first thing, like I'm thinking about how to make it as, how to reduce my technical vectors as much as possible. You get one thing. Ideally, you don't have to explain anything, but if you do, you get one thing. Then you're done with that. And then LinkedIn can be really cringey. I don't know how much of it is an algorithmic thing when people post that thing about like, well, someone came late to my job interview and you know what, they were the best candidate of all time. I gave them the job anyway. I don't know why everyone looks so, like, buttoned up and high and mighty, like, I'm the best employer of all time on LinkedIn. If that's just the style that's caught on or if that's algorithmically selected for, I'm doubtful it's algorithmic because I don't post that and I still do pretty well. I do think people are very hesitant to post on LinkedIn because it's perceived as cringe. And I would say, don't be cringe. Like, don't do it. Like, when I am posting content, I am filtering through, like, if I, am I going to be embarrassed posting this? Like, do I sound stupid? Do I sound weird? You know, there's a big part of the writing process and also the research process where if you're trying to make a point, you know, when you're when you're sort of writing any kind of blog post that relies on data. I know when it's working and when it's not. It's like writing a love poem too. You kind of know when you're bullshitting yourself. When you're like, oh, I do. I really want a love poem? Do I really love the person I'm writing about or do I really just want to write a stupid poem? You kind of know when, when it works and when it when it works and it's really something that you need to write about or when it's and rather than when it's just a instinct that you have to produce content. And once I feel myself doing the latter where I'm just posting for the sake of posting, I scrap the idea because it's not working. Like it doesn't feel honest. Matt: Yeah. I love that part because as someone who posts on LinkedIn and YouTube and other platforms, I feel like there's a balance between staying consistent but not going so far as to be like, okay, posting just for the sake of posting. And I tell people this all the time. It's like the best way to get more ideas and more content is to just do stuff, look at data, you know, come up with new questions and hypotheses. And the ideas will come. I will also say, though, there is a value to just posting about the same thing over and over again. Daniel: Like for both of my major projects that I've gotten some recognition for at Square, Square Payroll Index and at Ramp, Ramp AI Index, it's just always indices. Can you also just like tell people what the Ramp AI Index is? Matt: Ramp AI Index tracks business adoption of AI models and tools. So more or less, it's tracking as a share of businesses in our economy, at least on Ramp's platform, how many of them are using AI. It matters because if you are, there's not a lot of data available yet that is showing productivity benefits of AI yet in American businesses, nor are we really sure how many businesses are using AI at all, but if you are interested in tracking whether or not AI is going to lead to this significant productivity boom, which is the big hope about it, you would like to see a high share of businesses using it already, using it in a continuous fashion, using it on larger and larger contracts over time, and also to see that it is growing across sectors and not just concentrated in like tech and software, right? So there is no data set that tracks that except for Ramp. And there is a U.S. government survey that goes out every two weeks that has a methodology that I don't agree with or think is accurate or relevant. We can talk about that at length, but I've also written about it. But the point is there are very few ways to actually track whether or not businesses are actually adopting AI in an effective long term. Daniel: When we first started posting on Ramp AI Index, though, earlier this year, early 2025, it didn't necessarily do that well. I mean, it was the first time people had seen it. And so when we first posted about it, it was fine. Like it was a new data set, but it wasn't, it didn't blow up or anything. It wasn't frequently cited. It wasn't cited in that many media stuff. People didn't really know me as someone who posts that often on LinkedIn or certainly what Ramp's brand was yet with respect to data. Also, frankly, I did not have credibility yet because I just joined Ramp. So what are people to know about whether or not the data I'm posting is actually relevant findings or if it's going to be regularly posted? It's a big mistake a lot of businesses make when they start posting on LinkedIn about data. They just post one thing once and it seems like just a static thing that you never hear about again. Matt: Is there anyone else on the Ramp economics team that also has a presence on LinkedIn and actively posts about this? Daniel: No, it's, I'm the, I'm the voice. I'm the voice of Ramp economics lab. And I mean, with the larger data team is the only reason I'm able to do this kind of work because we have a really great, we have really great data sets at Ramp that allow me to do this. But you know, people, it takes time for this stuff to catch on. Like, you will be posting something regularly and then in some ways it's kind of a numbers game where your data doesn't just, you'll post about it regularly month over month and it'll do okay. Daniel: And then one day it just does really well. Like some person with a large following shares it and people suddenly find you, people who didn't know you before, or some other news story comes out that cites your data set because it happened to be at the right place at the right time where you were already producing this thing. Maybe some reporter had seen it, but they didn't have a story for it yet, but they knew that you were producing it regularly. And then when they had the right story, they cited you and then a lot of people saw it and then other people come to you. Matt: And so that's really the thing that I think people don't understand about this work is that they expect it to really do well early on when really you could really be posting about something for months before it just catches. And that happened at Square too. Like I posted about Square payroll index for months before it just caught on because I think some big account retweeted it. Daniel: Really? Matt: And then it got sort of cited in this large story about tipping and restaurant wages. And then the Zeitgeist kind of hit. All the presidential candidates started talking about how we should get rid of tipping or sort of get rid of tipping from like income taxes. And so what are they going to cite in those news stories about the prevalence of tipping? Square data. No one else publishing it. But imagine if we hadn't been publishing that data already. No one knew about it. And then best case scenario, a reporter reaches out to us and asks like, hey, can you turn around this data for me in 24 hours? No. It takes a long time to produce that kind of research. So the story gets written and we don't have it. So part of this is just being at the right place at the right time, but also producing this regular series of data that people start to know you for so that when they do need to refer to it or when the Zeitgeist just hits, you're there. Daniel: And so there is value in just hitting the same note over and over and over again and being putting your flagpole into the ground and being known for like, Hey, I know a lot about business adoption of AI. I mean, one of my goals for Ramp AI index is that we're not even there yet, but we're doing fairly well is that if a reporter is writing an article about whether or not businesses are using AI, I want it to be almost a dereliction of their duty as reporters. If they don't cite our data set, because what else are they citing? Surveys of people on the ground, like just asking businesses if they use AI or no, because the reporters want to use this data set too. The reason why they're not using it isn't because Ramp there is not credible. They're oftentimes not using it because they might not know about it yet. Just because you post something on LinkedIn, it does well on one day or get tens of thousands of impressions. That's not that many people. Most people have not heard about the data that I put out. Even most people in my world of covering AI and economics. Daniel: I mean, it's done fairly well, but I'm I can tell still. I'm in the early days because we'll talk to reporters often and they are surprised they haven't heard about it. Yeah. I think it's a good reminder of just sometimes how much of a bubble we're in. Like the 996 thing obviously has gone viral. I think within like the tech startup. Yes, that one did feel like something people knew. Like that one. I like, like I people when they found like that going, like when I go to something, like people say like, oh yeah, you did that 996 thing. I'm like, that is what I'm now known for. Or those restaurant charts that I made. I made this list of like the top 10 restaurants in New York based on where business, business spenders go. And that did very well. It's funny. No one cares about economics. They just want to know what restaurant to go to. But I think it is like a good reminder. There are a lot of topics and things that maybe we spend so much time within the tech startup scene thinking about that the general population in America just doesn't even know. We were talking with Jason Alvarez Cohen, who's founded a company called Popple a month ago. And he was saying, you know, a lot of people don't know what things like Clulee, Grok, MCP, like there's of analytics that you'd like to see. For example, me personally, I'd love to know the audience segmentation of like all the impressions I'm getting on my posts. Like, who's actually... Matt: Yeah, TikTok does a really good job showing you once you post a video, TikTok will show you like this line chart of where people drop off. And it's really fun and visual on it, especially on the app, where you just like swipe over on TikTok and see that like, oh, I lost X percent of people in the first second and then halfway through I did this overly technical reading of something and lost everyone else. That is immediately actionable and tells you through revealed preferences whether or not my content was good and which part of it was good. Now, some of that you could argue is like, oh, then aren't you going to over-optimize for the first second or something? But yeah, but at least I know that on TikTok, at least I know where I'm losing people. Whereas on LinkedIn, you vaguely see the number of people who viewed your content and you get this unusual number of like how long people viewed your content, but it's unclear what that really means because again, on LinkedIn, different from TikTok, TikTok, like your whole screen when you watch something is one video. So it's relatively easy for them to track what you're doing because they know, hey, I'm watching the video. LinkedIn, I will say, it is harder for them to actually track that metric that I'm looking for because it's a little messier to understand like, wait, is someone watching this video or is it just open or is it on the screens? Where do people drop off? They pause the video, you know, stuff like that. I will give it to them. Matt: It's going to be harder on a platform that most people consume on desktop interface. Yeah. What are your thoughts on video on LinkedIn? Daniel: I have no idea what to watch. I will say, I post videos on LinkedIn. I don't watch my own videos on LinkedIn. I mean, I will actually, that's not true. I will watch videos on LinkedIn sometimes because I'm on LinkedIn for my job and sometimes there's a compelling video that I watch. But the LinkedIn tab on mobile that sort of looks like the TikTok or Instagram reels tab, I really would love to do like a user interview with who is going on LinkedIn to consume short form video content because I have never had that instinct. And I think it's surprising to people that LinkedIn is going into short form video content. Matt: I've tried and experimented with it and I make videos anyway for my other platforms. So I autopost them on LinkedIn too. But I genuinely, because it's so new, like I have no idea. Like, who is the profile of people who is really, don't think it's like a CFO of a company watching LinkedIn videos. But then I also don't think it's like an early college grad, you know? Don't they all have other things to do then to scroll on LinkedIn? Daniel: I completely agree with you. And I don't know if I could name like five people in my personal life that consistently check out the LinkedIn video feed. That said, it does feel like LinkedIn is trying to push this narrative around video being more and more important. And Daniel, my co-host, who's unfortunately not here today, and I were talking about this interview that Jessica Jensen, the chief marketing officer of LinkedIn, with Stephen Bartlett on Diary of a CEO, You know, she's basically saying like, LinkedIn's up by like 36% year over year. And I think there's other stats around like creators who use video see higher growth. But every person I've talked to and asked, do you watch the video feed on LinkedIn? Is like, no, I don't. I've maybe looked at it once, but I think LinkedIn is primarily a text-based platform and people's behavior kind of like mimics that. Matt: I guess one exception where I have seen videos do really well is if it's a really big product launch video by YC backed company or usually those are horizontal, not vertical. And it's content though that they're producing for all their platforms. And so it's not optimized for LinkedIn necessarily. But yeah, I, I don't know. It's interesting. I, again, I will say though, like the videos that I've posted, and I've posted a variety of things, whether they are a little bit more produced, things that we work on with the creative team at Ramp, Vox style edited videos with animations, things like that, versus ad hoc, me just talking to my iPhone about this economic news of the day. And both do fairly well. The only lesson I have, though, or the only learning I have is that I have no learning philosophy is that I try to be, again, a relatively authentic version of myself on the video side too. Video is hard. Daniel: Like it's really hard to talk to a camera without a lot of practice and, and feel like you can speak effectively to a person, even though you're stuck, you're just in a room alone by yourself. And it comes with time and practice. It's kind of the first, the reason why I started doing it was I knew I needed to post this kind of content to get better at it. And you know, you can see that when you look through like old school YouTube people from like when we were kids or however old we were when like YouTube was pre monetization YouTube and it was just like people on YouTube make funny videos. And there was this, you can look through like people who really made a big around that era and see the progression of their videos and their production quality getting better over time. It's not just a technological thing. Like cameras got better. No, it's like they started learning a little bit more about AB and mics, camera quality and how they wanted to present themselves. And simultaneously they also became a little bit more confident in front of a camera and started to be more thoughtful about how they wanted their content to look distinctive from other people. And I still think I'm on that curve where I don't think I've figured out what my content, at least in terms of video, is supposed to look like relative to someone else. I mean, I do think that like charts that I post and my approach to things, I think you could, I think I'm getting to the point where I have a distinctive style that even if it wasn't, didn't have the Ramp logo in the corner, people who'd seen my content might be able to recognize that it looks like something I would have made. I think I've figured that out for images. I think I've figured that out for how I write. I don't think I've figured that out yet for video where I'm still trying to figure out like, what is my personal style that's going to come through in that medium? Matt: Yeah. I love that you brought up the distinction between super high production work that maybe you're doing with the creative team, Vox style versus something more raw, authentic, personal, just like you talking to the camera. And we were talking about this before the show, but I feel like there's YouTube in the past and video content just generally from like the 2000s was very raw, very authentic people just like the camera doing something interesting. And now, you know, over the years, it kind of shifted towards the Mr. Beastification kind of era where everyone wanted super flashy graphics and lots of edits. And you wanted to make things very interesting to people, but also it got very overstimulating. And so now I feel like, especially with more and more AI generated content flooding written platforms, especially it's like, but also video now because there's lots of video generation models. I feel like people are looking for something more raw and more human and more authentic. Daniel: And the easiest way to tell is if it's just the guy with his camera, you know, talking about something interesting. Yeah. The pendulum swings back, always does. And I agree with you. I mean, well, here's where I'm not sure, right? Like do on the margin, because I don't, do I think my content performs worse because I don't do like crazy jump cuts and I don't do like crazy thumbnail on the margin? Probably. Yeah. I think my content performs worse than it would, but I try to produce work that I think I am particularly well-suited to do. Like the other really good advice about anything in life, certainly how I try to produce content is I, don't try to be something that you're not. And I don't think I would be successful if I tried to be like Mr. Beast because I am not Mr. Beast. I don't have his style or his personality. I have my own stuff. And I don't think I could even, I don't think I could even ironically try to create like a crazy thumbnail where I have like my mouth open stretched and like giant impact lettering in my, my shadow or whatever. And so part of it is just recognizing that my content will do better if I produce the content that I'm particularly well-suited to produce. Speaking the way I speak and trying to be smart and thoughtful, but also approachable, telling stories similarly that my dataset is uniquely well-suited to tell. You know, people will often send me ideas of data to produce and the most frequent reason I turned them down is because other datasets could do that better. Like someone at, like there's some obvious ones, right? Like I get people, there's always news stories about like weddings and stuff. They're like, oh, you should do some story about like how people are spending on weddings. And I'm like, look, Ramp is a business spend platform, like we don't really see consumer spend on weddings. So that's an easy one to say no to. And the other one is like, Oh, well, can't you see something like the labor market stuff? I've rarely produced anything about the labor market. So, you know, like, it's is a performative exercise. The way I present myself online and the way I present myself in public speaking engagements is a performance, and it's kind of what I like about it. There's a part of it that is theatrical. I am thinking about how I want to explain something to people in a way that I think is going to be more helpful and better than how other people have explained it to them. That's why they're coming to my content, right? Because theoretically, I have something that they don't know yet that no one else has been able to explain as effectively. And so I am thinking about how to do that, but I'm trying to think about how to do that in a way that also I can do that no one else can replicate. And so I am thinking about how to do that, but I'm trying to think about how to do that in a way that also I can do that no one else can replicate. Daniel: And maybe that means I have a style for my charts where I speak a certain way, or maybe it means that I use humor in my specific way, or I'm a little snarky sometimes. Not overly snarky, like I'm not obnoxious and not pretentious, but like a little bit, like that, because I, you know, for my own comic effect that I think makes sense to be a little bit, you know, to integrate your own personal voice in some ways, right? Like that comes through, I think, in much of what I do. And then a lot of what I do is just straightforward, just reporting on stuff with my own analysis here and there, that can be performed as still in ways that I am offering my best guess about something. Something that I might not even be right about or something that maybe I don't even have all of the data to sort of be definitive about, but I know that people just want an educated guess and providing that is still a performance. Matt: Awesome. So how would you approach making content as a founder or professional? Daniel: I mean, try to think, if you're working with data and you have a unique data set, try to think about what makes your data set unique that nobody else could produce. You can do that in a variety of ways. Think about the skew and bias of your data set and use that to your advantage. Don't try to hide that. If you have a data set that only talks about this side of the economy or this sector, don't try to say things about other parts of the economy or other sectors. Just be the person who is an expert on this specific thing that you have access to. And that's fine. I think people see that as something that maybe limits them. They think, oh well, I want to be someone who can talk about everything in the economy. You don't have to do that. And you will not be effective in that if you are being stretchy with what your data can talk about. Daniel: And then similarly, when you do end up producing a content, you have the idea, you have the topic, think about how you can present that in a way that is unique to your data set and really highlights what's unique about your data set or what you have to say. Do you have something worth sharing or not? Because if you do, it should ideally come fairly easily. You can do that by remembering that, hey, you get one thing, that one technical thing, figure out what that is and then get over it as fast as you can because if you want to articulate something to your audience, think like the consumer of the news that your audience is. Think like the consumer in the economy. Don't think like an economist. Don't think like a founder. Daniel: I'm, by the way, not a founder. I've never found a company. I'm way too risk averse to ever found a company, but I have talked to a lot of founders and the ones who communicate most effectively are the ones that focus the most on how a customer or a user will think and feel. And they don't presume that they have any extra intelligence as a founder. Matt: Like if anything, their intelligence as a founder is because they are so imbued in what their users need, feel, and want. And I feel that way about the content that I produce as well. I do not have any special expertise because I'm a researcher or an economist or any of that. Everything that I do revolves around my experience as someone who participates in day-to-day economic interactions, just like everybody else. Daniel: That makes a ton of sense. What's your approach to writing LinkedIn posts? It's a curation exercise as well. You are being thoughtful about, it's like product design, right? Like you're not just, there's a lot of little micro decisions you make about which element fits where and how you want to organize a user journey, something like that. And I feel similarly about content where I am thinking about how I want to explain something. How do I want to walk through a technical concept in a little different way than other people might have? Where do I want to get people on the ground floor and explain all the basics? Or where do I want to sort of spend some time and energy talking about like that little esoteric thing that I think is still interesting and compelling people? That's a decision that I make. All to say, there's no recipe about how to do this stuff. Like it's a curation exercise. Like it really, and your individual decisions about how to organize information come through in the work that you produce. Every person, anyone who does my job would do it differently than I do it for that specific reason that they believe in prioritizing other sets of information than I do. Matt: What would you say to employees that want to post on LinkedIn but are maybe a little bit scared? Daniel: Check your other company policy. I will say for me, I, my job is in part public facing. Like my job is to post and to be out there, but that only works because I have buy-in from Ramp as a company and Ramp's executive team, which wants to focus on how to publish Ramp research and Ramp data in a compelling way across distribution channels. I mean, this job is really hard in some ways to set up at a new organization because everyone loves the idea of getting news coverage with your company's unique data set. Like that's theoretically great. Who wouldn't, who, what comms professional or executive wouldn't want to see their, their company's data mentioned in the New York times. But then they realized, what does it take to get there? It means hiring someone who, by the way, has access to all of your user and transactional data that you can trust to curate content in a way that they're going to, by the way, post about it online. And also they need to have relatively little controls and review processes because if you put too many review processes on them, like, and you want them to post about something topical, they're going to be outdated by the time it's out. So you kind of just need to trust whoever's in the position to not get your company in trouble. Matt: Oh, and by the way, you want them to be, sometimes they're going to have to post about politically trending topics in the economy. Like once you realize all of those realities of producing this kind of content, most people back away from the idea and they become very risk averse. Many companies and many comms teams and many founders are like, actually, let's hold off. I do think companies that do it well have realized that you can, you don't have to be that risk averse if you hire a group of people to do this kind of work that know the kind of content that the company can produce and be trending on topics and be responsible stewards of user and transaction data. Like you can develop that team. It's just hard to, I think, grow comfortable with that idea without having done it first. And in my experience at the companies that I've done this, it's only worked well when the company wants to do it. This cannot be a little side project. This has to be something that a company's executive team understands and believes as a worthwhile effort to get people to understand and see the product pitch function. Daniel: Do you pitch ramp products in your content at all? Ramps product revolves entirely around taking the messy transaction data that a finance team typically has to deal with and automating all of that. And then showing you insights about how to operate more effectively based on what you find in, in Ramp. And that's kind of my whole ethos for what I produce. I mean, Ramp will show that to you for your business. If there's any skew to the work that I produce, anything that I try to prioritize, it's work that is helpful and tries to highlight how businesses are running and working. And ideally highlight how the most efficient businesses are working so that your business or your organization can also take advantage of these lessons. If there's anything I study, it's how businesses work and how the best ones work. And if there's anything I write about, it's how other businesses can take advantage of those lessons. And that's the promise of Ramp as a product. So to the extent that I can show that to people who aren't already on the platform. And by the way, I'm, I kind of say this, that I'm the only person who speaks on behalf of Ramp that doesn't have to sell the product. I don't, I kind of, as a function of my job, I'm not supposed to pitch the product both because I'm not really trained up on all the product's features, but, and I shouldn't be a spokesperson for the product, but also because I think my work does better when I am perceived as someone who is unbiased and someone who's not trying to sell the company. And luckily when I post about Ramp data, I'm not selling Ramp data either. I'm just selling, not selling anything. I'm just telling you, Hey, like here are the things that you might want to know if you are a business about where things are going. Matt: Well, this has been an awesome conversation, Ara. Is there anything else you want to share?