Nima Gardideh | Pearmill
High growth companies turn to Pearmill to optimize their paid ads spend. With hundreds of millions of dollars of ad spend under management, the company has built up a considerable amount of data about what’s working on platforms like Google, Facebook, Instagram and TikTok.
On this week’s episode of The Inbound Success Podcast, Pearmill founder Nima Gardideh explains what goes into the company’s experimentation framework, and how they’ve used that framework to drive better results for their clients. In addition, he shares some of their learnings around the ad strategies and formats that are working right now on many of the major paid ad channels.
Get the details on all of this, and more, in this week’s episode.
Resources from this episode:
Check out the Pearmill website
Connect with Nima on LinkedIn
Follow Nima on Twitter
Listen to Pearmill’s Hypergrowth Podcast
Kathleen (00:24):
Welcome back to the inbound success podcast. I am your host, Kathleen Booth. And this week, my guest is Nima Gardideh who is the CEO and co-founder of Pearmill. Welcome to the podcast, Nima.
Nima (00:39):
Yeah, thanks for having me.
Kathleen (00:41):
I'm excited to talk because we've got an interesting topic around experimentation frameworks, um, and kind of what's working now with, uh, online advertising, which I feel like is a topic on everybody's mind, but, uh, before we dig into it, why don't we start with you telling my audience a little bit about yourself and your story and what Pearmill is?
Nima (01:04):
Yeah. Um, so I'm in short words at Iranian Canadian, uh, who now lives in America. <laugh> so I am an immigrant of this country. Um, and I started my career as an engineer, became a product manager in technology companies, um, and have been sort of a serial founder since built some tech companies. And now we're building sort of what I call like a tech enabled services business, where we help companies grow on online using digital advertising. Um, and that's how we've become sort of known over the past couple years is helping a lot of these early stage startups grow extremely, extremely fast, right? So, uh, you know, we now have, I think two of our companies that we work with over the past three years have now IPO'd. So we've been through this like super fast scale, uh, pace with them. Uh, one of them is like, I think the second fastest growing company in the world right now is there it's called ramp.com.
Um, and that's kind of the space that we work in technology super fast growing. It takes a specific type of approach to be able to scale these companies very fast. And so Pearmill comes in and mixes four different types of discipline to do that. So one is medium media buying. So this is sort of what advertising originally started in is you grab budget and you decide where to go, where it goes and you buy the ads on these different platforms. Um, the second is creative. So we have a full production team. We produce all sorts of creative. We even have a studio now in Brooklyn, which has been super fun, uh, to, to witness. Uh, and the last two teams are one is the conversion rate optimization team. So this is the team that after someone's clicking on something, what does that experience like? How, how do we make sure we, um, fine tune it so that the user's not confused and sort of flowing through the process as we want them to.
And the last piece is basically some form of data or engineering. So, um, it goes all the way from helping attribute the, the ads. So which ad people come from when, when, when, before they convert it, um, all the way to automation of, uh, ad spend and an area that I call spend integrity, which we try to reduce essentially human error, um, as you're, as we're handling these budgets. Um, you know, I, one of the examples I use in this area, there's this, um, issue happening recently with Citibank where someone, um, is a human error, someone paid off a loan that was not meant to be paid off. Uh, it was $500 million. So it wasn't, it was a massive amount of money. Um, and now they're in litigation trying to get the money back. Um, and that type of human error is unacceptable, especially when you're, when we're handling in the, in a triple digit millions as well. Um, and so we wanna make sure that even though there's a lot of sort of research and, and, and thought goes behind making these effectively capital allocation decisions, when you're trying to spend money here versus there on, on advertising, how do we reduce the error as much as possible? So we build some software around that. So it's kind of a combination of, uh, you know, human effort, um, processes and then software. Um, that's kind of what Pearmill does.
Kathleen (04:28):
And, uh, what is like the, what's your ideal customer profile? What types of companies do you work with?
Nima (04:34):
Yeah, so we effectively work with, um, either venture back companies, cuz they by design want to grow fast, um, or companies that are growing at somewhere between 10 to 20% month over month. That's kind of what we would like to see. And then we do work early in the stage of the company. So mostly want to see folks that are post product market fit. And usually we, we test that with spend. So if you've been able to spend over, let's say $200,000 a month on these sort of digital channels, then we, we feel like you're close to product market fit, assuming that's spend is profitable or close to profitable. Um, and so that's kind of the, the stage that we start working with with folks. Um, and we scale them up from there, right? So average clients probably spending something close to a million plus right now.
Kathleen (05:24):
So that's so interesting. So you're saying somebody's already spending more than 200,000 a month and it's profitable or very close to that. So like if, you know, I hear that and my initial thought is like, well wait, if they're already spending that much and they've been able to make their ad funnels profitable, what is it that you guys come in and do that helps them like take it to the next level and get even better?
Nima (05:44):
Yeah, that's a very good question. So if you know what happens is in order to keep competing at that level, it actually gets way harder. So it's much easier to do the ad spend at that rate, but it gets harder because at least the major two platforms, there's three platform platforms, let's say Facebook, Google, and now TikTok is starting to become a, a real player. They all have very similar advertising auctions where they will reward the top better, but they also have like all these series of probabilities that goes into how they decide who to reward the impression or to click to. And in the earlier stages, when you're below 200,000, it does, it's, it's usually 200,000 is a, is a good marker. But really what we care about is volume. Like how many conversions are you getting, um, compared to your spend?
It gets harder because they are designed to reward the cheaper, like the folks that are spending less, cuz that's actually the majority of their, their buyers, right? So if you look at Facebook, they have something close to 7 million advertisers and most of them are not anywhere close to spending 200,000, they're spending very little amounts, right? So they're designed their algorithms to actually reward the smaller businesses, which is, which is wonderful. Um, so if you wanna compete at those higher levels, you're just competing against like multiple things, other advertisers who are, who are sophisticated and then also the algorithm is kind of stacked against you. So you really have to push. And so when you work with us, you kind of buy information. We just see so much ad spend go out there that we understand currently what works in these ad networks, what are the different structures? What are the different creative approaches that are working? Um, and what experiments that we need to run in order to discover, okay, within your brand, what are the six or seven things that work that help us scale you? Um, and you're really coming to in, in a large degree, buy information more than anything else and, and tap into a process that's kind kind of continuously working. Um it's
Kathleen (07:52):
So you just gave me the perfect jumping off point <laugh> so you're buying a process and information. And what we are here to talk about is the process and honestly, the information about what's working. So let's start with the process. Um, you have a, an experimentation framework that you use with these companies. Can you break that down for me? What does that look like?
Nima (08:15):
Yeah. So this is a, a framework that we use across all the four different teams. Um, so this is, you know, media buying or we call it growth in inside of our company, creative, uh, landing page or conversion rate optimization as well as like attribution in engineering. It's designed to end up teaching us something, you know, we want to learn as much as possible. So a lot of it is even like sometimes slower in order for us to walk away having learned something. And so what the process is actually like at a high level, quite simple, right? We ideate. So we come up with a bunch of experiments. We want to be running, let's say we've heard. So, or sometimes we know because we've run this experiment before that, you know, splitting up our a groups too much on Google ads is not a good idea. So we're trying to like consolidate them random example. Um, okay. That's an idea we have, we're gonna stack that against a bunch of other ideas that we have and try to see, okay, we have more conviction on this one versus the other. So we're gonna run this one and then we run the test itself. So before we run it, we have to design how long we're gonna run it for how much budget goes behind it. And so that's the design stage.
Then we start running it and after we run it, there is always data that comes out, right. So we tried to isolate as much as possible. So then we can trust the data that comes out. And so that's the analysis stage. So we're now understanding, okay, we ran this test, it looks like the conversion rates were as high and our costs were lower. Um, so that's, that's, that's good news that may, maybe what we need to do is now systematize this learning in a way that works for not only this one client that we ran this test for, but all the other clients that we have, right. So the process goes idea, test, analyze, and systematize, right. We just go through this loop over and over again. And the key part is that we do this on a monthly basis. So our whole company is structured to run. These sprints is what we call them.
Kathleen (10:22):
So you're, I was gonna say, are you, are you managing using agile? It sounds like it,
Nima (10:26):
Yeah, it's super similar to engineering, um, because that's kind of our out of backgrounds are. Yeah. And so we're running it effectively an agile process, but for
Kathleen (10:34):
A monthly sprint or is it
Nima (10:36):
Yeah, it's much it's monthly because, um, you know, in engineering you can just build it two weeks and get results out. Um, in, in marketing, you just need more time to get the data to come back. So, uh, we have monthly sprints week, one of the month is effectively the sprint planning process. Um, we're ideating, we're debating on priority, things like that. And then there's three weeks of execution. Um, and then just goes through that over and over again.
Kathleen (11:00):
Got it. Um, so do you have a certain system for tracking your results? Like how do you, how do you document, um, is there like, is it a Google sheet? Is it a project management system? What are you using for that? So that your team has visibility into it?
Nima (11:20):
Yeah. Um, you know, I think everyone talks about experimentation, but this part, this question just asks is the main problem. Like it's the hardest part is actually tracking. Yeah. This work. Right. Um, so we use click up right now. I
Kathleen (11:33):
Love click
Nima (11:33):
Up. I I'm a huge fan. Yes. Um, and this structure of the accounts are always we've um, by channel. So within each let's say client, we have a separate space for each channel. Um, and each space has effectively this like sprint process built into like, are we in the backlog stage and ideation, are we running these tests?
Kathleen (11:57):
So like if you had, cause having used click up a lot, obviously you can look at it in like a list view and see all your tasks or you can switch it over to like a CanBan board mm-hmm <affirmative>. So if I were looking at your CanBan board, would it, would that be kind of how it's organized?
Nima (12:10):
Exactly. So like you have these statuses for each of the stage of the process, except with like one caveat, which is after the test is run, it, it branches out a little bit. Now we have two Stephan statuses. One is around things that were proven by data and then things that are disproven by data. And then all of them have tags associated with them that are used across our whole organization. So then we can get a sense of, oh, we are proving this new structure on Facebook, let's say across all of our accounts. So now we have this idea of like, oh, it turns out there is this new thing that is emerging at, at the market level that we should be paying attention to.
Kathleen (12:53):
That was gonna be my question because having, having run, I mean, I, I ran our paid ads at, at, uh, two companies ago where I was, cause it was a smaller team now granted much, much smaller scale than what you're talking about, but we would have it like a lot of experiments running at once and they could be anything from like, should we use the color blue or the color, you know, gray more in our, in our creative, or it could be, do we do better with positive messaging or negative messaging, but it sounds like you're, you have two buckets if I'm hearing you correctly. One is like that level of very client specific, like we're getting granular to see what works and then you have the more macro level we're gonna hypothesize that Facebook's algorithm or Google's algorithm is going to reward a certain way of structuring the campaign or way of doing the bidding mm-hmm <affirmative>. And it sounds like the second category is much more applicable across all of your customers. Is that right?
Nima (13:51):
Yeah, exactly. I think that that part is really the reason you would work with an agency of our style. Right. Because then we're gonna come in with these market level layer, learnings that you just wouldn't be able to achieve by yourself no matter how much you're spending, cuz you wanna see if this structural thing is working across industries across geos. Right. Um, and so we are able to pull that data in
Kathleen (14:13):
And, and remind me how what's the total amount of spend you have under management right now?
Nima (14:16):
Um, it's in like the triple digit, millions per year.
Kathleen (14:19):
Okay. Okay. Yeah.
Nima (14:22):
Um, and so the, the, the other part, so this is actually more for structural tests. I think that we learned that this model, this click up model of like, okay, what is the campaign structures, the ad group structures and things like that works. Uh, we also still use it for creative in that we document the learnings after the fact. Um, but there is another tool we've internally built just for creative analysis, because it's just much harder to get that. Right. Um, and experiments are a little bit less, uh, clear, like the learnings are not very obvious. We're no, this color shade works better. Like it's just not as simple as that. Um, and I'll get into that in a second, but just to like wrap up this click up process part, uh, so effectively for every test that we run, the information is there and every campaign and every asset that data gets filled in the card.
So then when you look at it over time, you can get a sense of what happened to this adset or this campaign over time. And so we've automated most of that now where our channel managers get these reports that come into their slack and then us, and then they copy paste it into click up. We had to actually go into click up immediately from design, but then they didn't like it, which is interesting. Cuz they wanted to look at the data manually to some extent, to just feel what it's like, what the changes are like as opposed to just be able to report on it. So yeah, what we do now is just we give them the access and then they, it makes it faster. So they don't have to go to the ad account to pull the data. Uh, but they still do the copy pasting. So then they're creating a narrative of what's happening to the clients themselves.
Kathleen (15:54):
Okay. Um, and, and so one of my questions, cuz I ran into this challenge is around like, like Facebook is a great example. Um, Facebook has the, and I don't remember what it's called now. The type of ad where it will dynamically, you you'll give it like five headlines and five descriptions and five calls to action, five different images. And it will just like mix and match and find the combination that is the best one. And I always found it to be really challenging to track that in a way that that was easy to, to follow because you had so much going on and it was like, how do we, how do we parse from all the different variables to figure out, like, what was it about this ad that really worked well? So I'm curious how you deal with that.
Nima (16:40):
Yeah. You're, you're uh, speaking my language here and this is like a tough area in general. Um, so fortunately, which is super random on Facebook that turns out that that ad unit does not do as well as doing it separately and on your own. So still we've been able to avoid it on, on Facebook.
Kathleen (16:55):
I was gonna ask that cuz I feel like that's a hack, right? Yeah. Like you could take those same five variations of everything and it's like, it's like algorithmic math. Right. And you just come up with all the different combinations and you make them each their own ad and you're sort of, it's more time consuming, but you could do what Facebook is doing in a way that's easier to track. Yeah. But then it's but then always wonder like, is the juice worth the squeeze for all the time you're putting in?
Nima (17:15):
It's definitely not. So you don't actually want that, that much variability. So what we do instead, um, and, and this is how we run creative tests on Facebook and by the way, on Google, that works better. So using their sort of like mix and matching apparatus for search ads, it's called responsive search ads. It just does work better. It's the more superior ad unit. So we have a totally different approach in order to over time, improve on creative there versus Facebook or Facebook. We actually do a lot of work to isolate the information as much as possible. So we have separate creative testing, camp campaigns, each ad unit that we're testing goes into even a separate asset by itself. Uh, so we're structuring things in such a way that we are not letting them cross pollinate as much as possible. So because if you have two or three different ads, ads, let's say in an asset, you are by nature making it so that they are competing against each other.
Kathleen (18:14):
That's what I was gonna ask. Cuz I ran into this, like I wanted to do so much granular experimentation, but then you sort of get to the point, like I was running ads and I was like, we're gonna separate out iOS from Android because iOS is its own special form of hell at the moment mm-hmm <affirmative>. And I wanted to see my worry was that if I had it all pulled together, um, and, and iOS, you know, tracking and retargeting was more difficult that it would like basically screw up the signals for Android by having them all in the same pool. But then when I separated them, I was like, shoot. Now I think I might be, my ads are competing against each other. So like how do you handle that?
Nima (18:50):
Yeah. So the competition is, is, is a problem. But what you wanna do is get a sense of how does this ad do in isolation first and then how does it do in groups? So we actually talk about this gen the concept of does this ad generalize to the larger market or not, right? So you start with isolating as much as possible. So separate campaign, separate assets, and then testing 'em against benchmarks also. So because there were still some, um, let's say leakage at the campaign level. So what, what we do is we have, let's say we're testing five pieces of creative. We have five different assets in that scenario. Then we add two more where we grab on existing high performing ads from let's say the evergreen campaigns and we put 'em in there. So now what we do is we don't test them.
If they're better than the average in general, we test them. If they're better than the hero ads that we pulled in into this specific campaign. So now you're judging them against like the, in the same environment, are they doing better? Are they better than the benchmarks? And then we grab those winners and then we'd see if they generalize to your evergreen campaigns. Like, are they gonna generalize to higher amounts of spend or not? Um, and so this process over time, you get a higher hit rate than trying to like test an isolation and just put 'em out there. Um, and this is like our latest version right now. We used to even have, I think this is like V4 of our creating testing structure. And right now it works very well and we've tested it quite a lot. Um, but it, it is a problem.
I would say, even with this, we don't have a hundred percent hit rate. They don't always generalize. Um, or they, they generalize in the sub segment to the market, but not the whole spend. So we have to be very careful. Um, and then what we do there, it, and this is more the structural side right now. Like let's talk about the creative side. Like, uh, how do you even test ciNimatography for instance, this is my, one of my favorites, cuz it's the hardest one to test. I think like you can test copy a lot, lot more clearly. Yeah. And
Kathleen (20:48):
Images,
Nima (20:49):
Images, or
Kathleen (20:50):
There's like one variable you're changing usually.
Nima (20:52):
Yeah. So effectively what happens is that, uh, our team hypothesis things like they created hypotheses and say, we think this form of student autography is going to work better. And what we're gonna do is we're gonna produce a handful of them for each creative round. And then see if, if we held that one fixed, but then change the other things. We can see that it's directionally moving towards the right area. And over time, some clear answers end up emerging. It's not very obvious that this is the reason, but, and it's not even statistically significant. But what it does give you is that there's directional information that says this approach is better than this other one. Um, and you know, if you spend enough money and time on these things, you end up getting some really interesting results, right? So we have some learnings about south America versus north north America right now, for instance, in terms of shades of the same color, right?
Warmer versions work better. I think in south America than than America, right? It's like all these like nuanced things end up emerging at, out of the data. Um, and then there at the brand level, there's also like obvious things that happen. I, I like explaining this example. We have this client that named Saunder their distributor hotel. And over time we've learned that the photography or the videography has to be a certain way in that some aspects of the apartment have to be always in the photo or the video eventually. And what they are is we need to see that there are high ceilings, there are massive windows in the apartment, so you can get, get a lot of light and there's a kitchen countertop. So people understand it's not just the hotel I get to actually cook on my own as well. So these three things being in there over time has I think doubled our ad spend, uh, it's not ad spend, sorry, return on ad, spend with them and just learning this took us months, right?
Like, oh, these three things have to be in there for us to know for us to increase the average ROAS. And after these learnings come about, we end up calling them. These are the formulas. Like we have to have this, this and this in any photo or image we take, we always still spend something like 20% of our ad spend on brand new ideas that we have no zero conviction on just to like discover new versions of these truths. But, um, eventually when we start taking over the Aspen, 80% of the Aspen goes towards these like formulaic or what we call iterative creative. It's like, okay, we know this is gonna work. We're gonna produce more of this. Yeah.
Kathleen (23:27):
Interesting. All right. So I love this. I love how sort of scientific you guys are. So let's talk about what's working and what's not because we, you talked about people come to you for two things. One is process, which we've covered and the other is information. So let's get into the information side. Um, because I feel like, you know, I, I've worked in E selling into eCommerce and selling into B2B and it doesn't matter what vertical you're in or, you know, whether you're B2B or B2C. Everybody's talking about challenges these days with ad spend and Roaz and platform changes and, and all of this. I'm super curious to hear what you think is working well.
Nima (24:06):
Yeah, let's talk about, I think the different channels let's start with Facebook and then we can go through the other ones. So Facebook has this algorithm, Pete, they internally, I think still, maybe call it the discounted pacing algorithm. And really what it means is that you give an, give them an audience. You give them, um, a target, like let's say I want, um, more purchases. And then what they do is they grab the audience and they figure out who is the least sought after people in that audience that is most likely to take the action that you want them to take. Right? So effectively, who is the cheapest impression that is most likely to convert, which is wonderful. Again, it's like the focused on SMBs, right? Um, and what happens is there are some interesting nuances that happen when, when that is the core algorithm and these are all, these were theories a year or two years ago when we started the business, um, and getting into experimentation in this way.
But now we've proven these, right? So one of the examples here because of this pace pacing, um, and the way that would work is that you want to, as much as possible increase the size of your audience, which is interesting, right? So even if you're like, you know exactly who your customer is and you have those targeting options on Facebook, I would urge you not to target perfectly. And the reason this works is because if you give Facebook a larger universe of people to try to decide between there are people that don't fall perfectly within their targeting apparatus, but somewhere in the data, there is information that tells them that they're more likely to convert to your, for your product. So then you're tapping into this like data apparatus that they have that you don't have access to. So increasing and broadening your audience is actually a good idea right now.
Kathleen (25:58):
So how broad should you go though? Like, should you start with just general targeting in the beginning and like, let it, let see what happens and let Facebook be the one to figure it out.
Nima (26:07):
Yeah. This is like a really debate a topic right now in our company. So <laugh>, I can tell you what, like has worked, um, going straight to broad and by broad, I mean, at the country level, like some of our brands are all of America, that's like the targeting, right. Even some of them are B2B brands, but the targeting is all of America. Right. So, um, but most of them have gone there overtime as opposed to jumping into it. I think for some consumer brands we're seeing jumping into it works. Um, but you do need to be at a certain scale and the scale is around 50 conversions per asset per week. That's kind of what you to
Kathleen (26:43):
Do broad targeting
Nima (26:44):
To do, yeah. To do broad target that this level of broad targeting.
Kathleen (26:48):
Yeah. Because I have, so we experimented with this a little bit at that company where I was running our ad spend and I will say rookie mistake when I first did it, I didn't specify countries. And so all of my ad spend was going to like the developing world, because that was <laugh> the cheapest, you know, most likely to convert, but we weren't selling into those markets. So that was just a dumb, dumb mistake. But like, if you, if you do narrow it geographically, um, you're saying you need to have that certain volume to, to do that level of broad. If you don't have that volume, how narrow should you like, how, if you're starting smaller, how should you target it? How granular should you get?
Nima (27:30):
Yeah, that's a good question. Basically. You want to be, you know, minimum, I would want you to be in the, in the like single digit millions in size if you're in consumer. Um, and probably in the hundreds of thousands of people, if you're in B2B. Right. Okay. And then on the, in terms of how much consolidation you wanna do you wanna be around 20 conversions per week per asset? That's like the minimum you wanna get to, um, if you're doing any broad targeting, um,
Kathleen (27:54):
And when you say conversion, do you just mean like people who click through and fill out a form on your site? Or are you saying like a purchase or
Nima (28:02):
Ideally it's as close to revenue as possible. Okay. Um, but sometimes you don't have the luxury of that.
Kathleen (28:07):
Yeah. I mean like some B2B, yeah. This is not gonna happen
Nima (28:11):
And yeah. And B2B like a lot quite often it's lead. Yeah. But sometimes now what we do is it's MQL. Yeah. Um, when you're reaching a certain scale. Okay.
Kathleen (28:19):
Um,
Nima (28:21):
And then for, for eCommerce, you know, depends on the basket size, but if you're like anywhere below one 50, I want you to be going straight for purchase. Yeah. Um, if you're above that then considering at the cart or like other, maybe even ask them questions and things like that,
Kathleen (28:39):
Having a quiz on your site and somebody can click the quiz. Yeah. Yeah.
Nima (28:43):
Um, for B2B, I'm big fans of a fan of quizzes as well. Uh, that works very well. Cuz then you can define MQL much more succinctly cuz they're just literally giving you the information you want. Yeah. Without a sales staff, talking to them and, and, and tagging them or, or marketing staff tagging them as an MQL. Um, so that that's an area. So like the broad targeting just works on Facebook. And so you want to somehow over time approach it slowly step by step with that said, sometimes we grab an account that's spending 200 K and say, it's time. We feel like there's enough learnings in the account at the account at the at account level. And there's been enough pixel fires that we're willing to go straight to broad targeting. Um, we just feel like it's gonna, it's gonna work. It, it sucks for about a week <laugh> but then it, it does better. And then we look like heroes. Right. So, um, I I've seen some of our paid social managers just go straight to broad recently. Um, but at the larger scales where they just have so much confidence that they're gonna get enough conversions.
Kathleen (29:42):
Yeah. Okay. So broad targeting on Facebook, number one.
Nima (29:46):
Yeah. What else is
Kathleen (29:47):
On
Nima (29:48):
The last thing on Facebook is, I mean, you know, I think all of this is re uh, reliant on two things. One is that you're focus on the right optimization event. So, uh, going towards an a, a, a conversion event that fires enough is highly correlated to revenue if it's not revenue itself. Uh, so we do a lot of work here on like, we have even some linear regression models where we try to fire events when we feel like, okay, we have like a lot of conviction that this visit is a high quality visit. Uh, you can do things like that. Um, but obviously that is like the basics. Now you just need to be fi sending the right signals to these networks because there's machine learning at play that tries to create probabilities of, um, people taking action. So you should be feeding them the right signal.
Um, other than that, the last thing on Facebook before we move on to Google is now you're at broad targeting. How do you target well beyond this just optimization event? Because there's so many different personas you have, it's not just like the same type person buying your product or, or becoming a lead. The only other thing other than the conversion rate event you have access to is creative. So start thinking about your creative as it means to sub target the larger audience. So what we have quite quite a lot now for these at scale clients, we have multiple assets all going towards broad targeting. The only difference is the vibe of the creative, right? So like now we're SubT targeting the larger audience based on how we talk about the customer, how, how we talk about, um, the product, what features we're exposing, what ciNimatography and videography we're using, what typography or colors we're using, and because you have different personas who want different visual language to, um, be before they're know, convinced that they should click on it. So we now create that level of targeting with creative. Um, I think this is the coolest part. You do have to be at a certain scale for it. Um, but it's like the most interesting thing I think our team is doing is now our creative team thinks about, oh, who are we targeting right now within this, like the larger ecosystem of this, uh, market for this product. Right. Um,
Kathleen (31:54):
So is that, so you're taking, you're taking your ads and you're saying we think we might have four personas and we're gonna target the same audience across all these ads, but we're gonna have like this set of ads that is more geared towards persona one. And this one more geared towards persona two. Um, do you ever also, uh, sort of test features and benefits to see which one is gonna resonate the most? Or is it more persona based messaging?
Nima (32:20):
No, it's definitely like, I, I think it's all the above, right? So we'll, we'll sort of iterate through all of those ideas. Is it like the way we pitch the product? Is it like the visual aesthetic or is it just like which part of the product we're exposing? Um, and, and especially for like one of our clients is, you know, ramp is a credit card company, so you have CEOs buying and CFOs buying and they're like have diff very different needs and the different futures of the, uh, the SaaS platform that they have on top of the credit cards just turns out to be much more interesting to the CFO versus the CEO. Yeah. Um, so then you, you speak about it differently. Um, especially in BWP, there is a lot of like matrices and like here is the job title versus the buying committee problems are exactly like who is the economic buyer versus like the, um, hero and all these other stuff that we iterate through all of that. Um, and then sometimes they actually get stuck in one asset, which is interesting, cuz Facebook is so good that they'll realize what you're trying to do here. And then sometimes we've learned that if we put 'em in different assets in groups of sort of personas, um, the assets, like sort of overfit on that part of the market and then they, they get those types of, um, leads or purchases instead.
Kathleen (33:28):
Um, side note, I love ramp. I've used it at some of my past few companies. It's great. Um, so it's cool to hear that you, they're your customer. I'm curious at the, like if we, if we zoom out, um, you know, having, having sold into e-commerce, I I'm hearing a lot from e-commerce companies, especially in the DDC space, uh, because they're so aggressive in terms of how they tackle their advertising and they're, they're so dependent upon the results from it. Um, there's a lot of talk there on what's working and what's not, and I'm hearing, I'm hearing a lot that at least right now, uh, cost per acquisition on TikTok is much lower and TikTok is super effective. So there seems to be in, in the D TOC space, this flood of companies shifting spend over to TikTok. And I don't know whether this is a temporary bubble or I I'd love to hear what you're seeing and, and even beyond eCommerce, like for B2B mm-hmm <affirmative> I don't know what, what's your take on that?
Nima (34:26):
Yeah, I think, well, we're seeing the same, um, in that a good portion of ad spend is now at least new allocation of ad spend is being shipped to, to TikTok if not being stolen from Facebook. Um, and TikTok creative production is much harder, which is interesting. So like it's a lot more trendier, so you have to be matching the trends that are, uh, bubbling up in the sub communities that, um, or like, you know, the, the talks that, um, <laugh> are close to your brand. So we do now, like a lot of research on, uh, the talks that let's say, like we have, um, we have fashion brands right now that are doing spent over there. And, um, what are the approaches that these like fashion influencers are taking in terms of their approach and creative, can we support that? Can we use some of the, uh, existing sort of trends to, to scale up and add and all this sort of stuff, like, sort of comes into play there, question,
Kathleen (35:22):
I'm sorry to interrupt you, but question for you, cuz this is a sort of, you just touched on something that I've had conversations about in the past. And I wanna hear what you think about this. I've had other people tell me that yes, you need to be much more plugged into what I would call like the zeitgeist, right? Mm-hmm <affirmative> like, what's what is, is happening in popular culture and how can we like piggyback on that? But I've had people tell me that in terms of, because you said creative production is harder. Um, and I wanna make sure everyone listening understands exactly what you mean by that because I've had people tell me that yes, that is true. But then they've said actually from a production quality standpoint, that that what performs best on TikTok is actually the less produced seeming, like the more kind of, Hey, I just picked up my iPhone and did the thing, whereas on Facebook, sometimes the more produced stuff does, well, I don't know if that's true or not. That's what I've heard. I'd love to hear what your experiences
Nima (36:14):
Are. Yeah, yeah. I guess like that's a very good distinction. Thank you for asking me that one. So it's harder in that knowing what to produce is way harder, right? <laugh> um, in terms of how well done it is in, you know, qu quality of production, I think TikTok is like early days Facebook in that sense that, yeah, it's you don't have to be overly produced, um, with the caveat that it does depend on which sub segment you're in. Right.
Kathleen (36:39):
But also I feel like the reason people have said that is that the stuff that tends to do well on TikTok feels like, feels like it's in your fee, like organic and in your feed. Yeah. Like it doesn't feel like, oh, I'm being interrupted by an ad. It's just like, oh, here's another cool video. It happens to be an ad. So very much like native advertising in that sense, but in the, in TikTok, native advertising is video.
Nima (37:00):
Yeah. Yeah, exactly. It's like mostly it's effectively like, you know, UGCs sort of content, user generate content. Yeah.
Kathleen (37:07):
But UGC, that doesn't seem like it's UGC. Right.
Nima (37:09):
<laugh> exactly. But like, you know, what works the best I think right now is just like grabbing the popular talkers and then asking them to talk about your product and the way that they do, they do it best. Like they're funny and interesting talking about the thing that they already love and using. Um, so quite a lot of the TikTok production for us is just working with existing creators.
Kathleen (37:29):
That makes sense.
Nima (37:30):
Yeah. Uh, but yeah, I think a bunch of the spent is moving to TikTok. You know, if you've, I really love this old essay that Andrew Chen wrote called the law of shitty clickthroughs <laugh>, um, which, you know, I remember reading maybe like a decade ago and I think you can still find it if you search for it. But effectively what it says is very simple. Whenever there's a new ad network, um, there's all these cheaper CPMs and effectively cheaper clicks, um, that's available to you and your job as a marketer is almost to constantly be looking for these new networks. So you can take advantage of those cheaper times before everybody else realizes, oh, you know what? This is gonna work. So I'm gonna allocate my budget there. And then it becomes
Kathleen (38:13):
Right. It's a market there's supply and demand. As soon as the demand goes up, <laugh> it's gonna get more expensive.
Nima (38:19):
And it just has happened that there's been like, you know, five to 10 years since there's been like an, any real network that was at scale, trying to compete with Facebook and Instagram. So this is one of those times, um, you know, there was like some point, I remember people talking about Cora being potentially yeah. The solution. It just didn't scale. And it was not as good. Um, same with Reddit when they launched and Pinterest, when they launched, like, it was just not the same, but TikTok is really nailing it. And what's cool about TikTok is just, you just see it over the last year, they started the ad network. We, I think spend something like a couple million dollars of, at some point didn't do really well. We stopped. And then they worked on the algorithm and then they called us up again. They were like, we think it's time for you to try it out again.
We tried it and it just worked. So they are iterating on it and they care a lot about performance and they're doing a good job on the, uh, on just building the, the apparatus for marketers. Um, uh, but just to a, at risk that maybe the larger problem you, you talked about, which is mostly commerce companies are effectively, uh, using paid as the path for growth. Um, naturally paid gets more expensive over time. It's a big problem. Yeah. I think it's a huge problem. I think most e-commerce companies should not be raising venture capital that much. I don't think they should do it because, and this is the, the, the tricky part. I think you can build very, very good e-commerce businesses so long as you're, um, sort of timeframe for becoming public or being like a big company is not five to 10 years, but it's actually 20 to 40 years.
So if you change your time mindset, mindset on, I just want to grow 20% to 50% year over year. You can do that for a long time. Only spending a certain amount of money, not increasing your ad spend every year, just because you have to relying on good product production and continuing to produce new products, uh, to augment the, let's say, AOV on your, on your marketplace. Like you can really build amazing businesses so long as you're not tied to some artificial growth rate that's set by your investors. So that's the, I, I, I, I, we love working with DTC companies that are very close to profitability, and then we make a deal with them to get 'em to profitability first, and then we'll grow them more. And we, we change our, um, pricing model for them just to get to that. And then we get a bonus when we get them to profitability and things like that because of this philosophy that we have, we just think the e-commerce market is just too saturated. It's too easy to enter it. Um, you know, if you're smart and understand how to work with manufacturers, you can effectively start a brand within a couple of months.
Kathleen (41:06):
Um, not even, I mean, people do it with drop shipping.
Nima (41:08):
Yeah. It's like couple days probably. Yeah, exactly. Uh, but if you're making maybe something harder and unique, it'll take a little bit longer. Um, and so the market is super saturated, but if you have this long view of how you're gonna exist as a company, I think it's just still a, a great area to be in. Um, and unfortunately it's still muddied by venture capital to some extent. So you gotta be careful.
Kathleen (41:31):
Yeah, definitely. All right. So we're almost at time, but I, I wanna make sure I ask you before we start to wrap up the last question, which is, we haven't talked about like Twitter or LinkedIn, any just like high level thoughts on what's working there.
Nima (41:49):
Yeah. So Twitter is quite confusing, cuz it basically has been a hit or miss for like years for us. We always spend on it, but almost never are happy with it. Uh, I would say a handful of clients right now are enjoying spender in that like it brings on conversions that are valuable and interesting, but you have to be in a niche. So like ramp, I think is one of the only ones actually out of all our clients, because they just happen to be very popular within like techies. And, uh, that's an, it's a good area. Like if you're like selling to techies, Twitter turns
Kathleen (42:19):
Out tech or, or DTC, I asked because like the DTC founders are all over Twitter. And so I was trying to experiment with it, but like none of them advertise there, but they're all active on there. <laugh>
Nima (42:30):
Yeah, they are active. So if you're selling to the DTC founders, maybe it's a
Kathleen (42:33):
Computer. Well, that's who I was selling to <laugh>
Nima (42:35):
Yeah. It's like the founder class is on, on Twitter. So it's like an interesting area. Um, LinkedIn targeting is obviously excellent, right? Yeah. So you can go after exactly who you want, but they know it. So that's the problem. So it's very expensive. They're sort of baseline CPMs are very high.
Kathleen (42:51):
Yeah. Wasn't it. Last time I looked, it was like six or $7 or not, not CPM. Sorry, per click.
Nima (42:56):
I mean, oh yeah, yeah. It's it's, it's quite high. Yeah. Um, the CPMs are way higher. Um, the, basically what I think is happening right now is that we use LinkedIn, but more as a channel just to get clicks as opposed to conversions and then wet target them on all these different platforms that are cheaper. Um, a good amount of spend still goes there, but there's been a recent surge of these like sort of data companies coming in and saying, you know what, we're gonna help you spend on Facebook, but have LinkedIn level targeting. Right. So now there's clear bit advertising, metadata that IO say primer.com. We actually use all of these products cuz we're always like putting, putting them against each other to see which one works better. But
Kathleen (43:38):
What have you found found?
Nima (43:40):
<laugh> we like clear bit right now. I think the most metadata that I owe is the second best. Um, yeah. It's just access to data. And what you do is effectively that you go on these platforms and say, well, I want CEOs of companies between X and Y number of employees with this much revenue using this form of software, you know, whatever targeting you get on LinkedIn and actually much more because now they're like buying data from S yeah. And like it's stitching it together and then they upload an audience for you on Facebook and they can target them. So it's very good for B2B. Um, it hasn't, it's not so good that you stop spending on LinkedIn, but now you're, you have a competitor to LinkedIn advertising in a way that works and sometimes even cheaper.
Kathleen (44:19):
Well, and it can be kind of part of a more full funnel strategy. Cause like we use metadata at my last company and I think you're right. Like I've always looked at LinkedIn as a top of funnel channel and then how do we, how do we retarget them and get them someplace else where we have better ways of actually turning it into a meaningful conversion. So
Nima (44:35):
Exactly. I think that's the right move. Um, and I think that, you know, it goes to mention maybe YouTube is maybe sometimes for them.
Kathleen (44:41):
I don't even, yes, I should have mentioned that. I can't believe I've forgot that.
Nima (44:44):
Yeah. It's super, it's like doing very well right now. Yeah.
Kathleen (44:46):
That's what I've been hearing.
Nima (44:48):
Yeah. Both for B2B and B2C. It's like wonderfully useful. It's a little bit hard to track obviously, cuz it's mostly view throughs. So you have to be like very rigorous at running incre mentality tests and you know, putting your spend there and then turning it off and seeing what happens or increasing it dramatically just to see what happens in the differential, in your conversions, uh, to get a sense of what the real value is, but it is valuable and we keep proving it over and over again. Um, there's probably a ceiling there, but um, if you haven't done any spend that I highly recommended it's, it's, it's quite useful. Um, and I, you know, I'm sure you've seen monday.com ads at some point. So you should know that it works for B2B, uh, but it, it works across the board for us. And uh, especially if you combine it, um, there's this new campaign type called performance max, um, campaigns, uh, within Google that you can use. And so that area is new and is working quite well. Um, and unlocking new spends at Google for us.
Kathleen (45:41):
Great. All right. Well, I feel like I could literally talk to you all day about this topic. So we do not have all day and I wanna be respectful of your time. So we're gonna shift gears over and I have two questions. I always ask my guests, um, that I wanna pose to you. Mm-hmm <affirmative> the first being, and this is a great subject for this with paid ads. Like all the marketers I've talked to say, there's so much changing in the, in the world of digital marketing and particularly digital advertising, um, that it can feel a little bit like drinking from the fire hose to keep up with it all. Mm-hmm <affirmative> so how do you personally stay on top of all of this?
Nima (46:14):
Um, well first of all, thank you for having me, by the way, it's been super fun to talk about this, obviously, you know, your stuff and, and it's, um, always good to talk to someone who understands what I'm talking about. <laugh>
Kathleen (46:24):
I love nerding out on this stuff with people.
Nima (46:28):
Yeah, I think that's a very good question. So I talk about this in the context of mastery, a lot in with our company. Um, and you know, if you look at like other professions, like being a doctor, for instance, it's very similar, there's new things happening all the time. If you don't keep up, you're not gonna be the best Dr. Best therapist. Um, and you have to be constantly on the look at, and, and maybe in marketing it's a little bit faster than those things, cuz it's not like research, it's literally, you know, a handful of people who run experiments and all of a sudden it shows up. So the most important thing is just having a very strong process, right. Um, and the personal process of learning, um, we have this like series of channels that you can subscribe to on slack, on slack, for instance, where everyone shares, what they're reading, um, and thinking about.
And the moment there is a form of an idea for an experiment, it gets grabbed and put, puts into the ringer of like, okay, we gotta try this out and see if it's gonna work or not. Um, and on the personal level for me, I, I have rituals around it, right. So I will go and meet with the team and just ask them direct questions, just like how you, you ask me all these questions. I will go and be like, okay, what's working. What do you think is working that now? And um, what is your sort of hypothesis underlying it? Um, what is your new intuition on this stuff? Um, and just having these conversations over and over again, not just with my own team, but also other marketers out there, uh, and having sort of like part of your job is to do this is to learn and, and ask people how they think about these markets. Um, and it's hard. I think it's part of the job and I find it quite interesting cuz it's really, I dunno, I find it cool. So I, I I'm I'm game to always have these conversations and now we have a podcast where I bring on marketers and just grill them about this stuff. So, uh, just like yourself, I think it's like a very good form of learning, um, to just be curious and continue being curious. Right?
Kathleen (48:24):
Yeah. My podcast is, is how I learn. Definitely. Mm-hmm <affirmative> it's through people like you. Um, alright. Second question. The podcast is all about inbound marketing, which the definition of that is, is for interpretation. How I define it is anything that naturally attracts your right customer to you. Is there a particular company or person that you think is setting the standard for what it means to be a great inbound marketer today?
Nima (48:48):
Oh, that's a good question. Um, I think let's go through the obvious ones and then maybe there's like another one I like, um, I think people don't talk about Tesla as an inbound marketing machine. They, I think they spend like very little if or if or nothing on advertising. Um, and they're quite good at it. Uh, obviously their model doesn't work for everybody, so the, the learnings are not maybe as generalizable. Um, but the other one
Kathleen (49:17):
That's okay. I mean it's Elon Musk love him or hate him. He's a master at branding.
Nima (49:23):
Yeah. Incredibly good at it. Um, you know, one of the best in the world Notion I, I would say is probably another one. And what I like about Notion is, um, they take this idea of branding extremely seriously. And if you meet Ivan for instance, and I think you probably received emails from Ivan, cuz it says Ivan from Notion still to this day, when you get a notification from Notion, uh, it, it is who he is like everything at Notion, it's just the manifestation of the founders and what they believe in, in the core principles of design and how it rose in the, in the organization. And you can watch their talks and you can see how they thought about their offices, what they look like. Every copy that's written, everything that they express externally is so well done and exactly Notion ask right now I can tell you if I don't see the logo and I see read, copy, or look at visuals, I can tell you if it was Ivan and his team or not. Um, just because they just do this work extremely well. Um, and, and I'm a big fan of them. I think there's a company behind them actually had, um, their CEO on my podcast is animals. Um, they did a lot of their content marketing stuff and um, they did a, a wonderful job, I think, um, taking those ideas that, that their, their team had and, and making it work extremely well.
Kathleen (50:55):
Yeah. That's a, that's definitely a good agency. I've heard them mentioned a lot on here. Yeah. And I actually interviewed somebody who, when I interviewed him was, was at animals. He's not there anymore, but um, but love the work they're doing. Um, alright, well that brings us to the end. So before we wrap up, if somebody's listening and they wanna learn more about Pearmill or they wanna connect with you and ask a question, what is the best way for them to do that?
Nima (51:18):
Yeah. Great question. So you can find us on pearmill.com, P E A R M I L L.com. I'm also on Twitter and Gardideh Um, and you can also listen to our podcast, which is around mostly I, I, I speak to founders and marketers around hyper growth and it's called the hypergrowth experience. Um, it's more mostly focused on maybe organizational design and process on how to get this stuff done. Um, and yeah, that's, that's it. Thank you so much for having me. This was super fun.
Kathleen (51:46):
Yeah, this was great. Thank you for joining me if you're listening. Um, and you like this episode, I would love it. If you would head to apple podcast sand leave the show a review, uh, and if you know somebody else doing amazing inbound marketing work, send me a tweet at K@athleenlBooth and I would love to make them my next guest. Uh, you can find all the links to reach Nima in the show notes, which are available at kathleen-booth.com. Thanks for joining me this week, Nima. This was great.
Nima (52:13):
Thank you for having me.