Ep 535: Ashvin Melwani On How Obvi Fixed Meta Pixel Poisoning and Recovered Ad Performance

0:00Ash Melwani:Listen, I've tested thousands of ads. We've done iterations of our landing pages. Like, I know what works and I know what doesn't work. And I know when something is off of my ad account. Because, like, you don't just see this, like, sharp decline overnight for no reason. So, if you are a health and wellness brand and you don't want to get flagged, 50 % of the people are making their first search on TikTok. They're not Googling it, quote-unquote. They're going to TikTok and searching for it there. If you're not there and you're not showing up there, you're missing a massive opportunity. do I keep this ad on or do I turn it off is not easy.

0:34Ash Melwani:That's why having media bias to understand when to do and when to pull certain levers is very, very important. The name of the game for 2025 is just survive and hope for the best.

0:53DTC Podcast Host:Ash, welcome to the DTC podcast. It's been a long time. You graced our stage in Las Vegas. I guess that was in 2023. I've been watching you from afar. I know you've spent over$50 million on Meta. You have helped drive Obvi to over$100 million in sales. Big congrats. How are you doing?

1:13Ash Melwani:Good. I'm good. I know it's definitely been a minute. I've been trying to cut back a little bit on the traveling just so I can focus on the brand a little bit. But, you know, here and there, there's a few things that we love going out to. And hopefully we'll see you guys soon.

1:28DTC Podcast Host:Yeah, we're going to be doing some meetups, I think, in the next little while. But until then, I've just been I've been enjoying I think you're one of consistently one of the best tweeters on or Xers on the topic of the you know, the deep topics on meta. And I wanted to kind of walk through a little bit. First of all, maybe just catch us up, catch us up a little bit on the growth of Obvi. What I want to dive into today is kind of your story of having to rebuild your pixel, because I think that's something our listeners could benefit massively from. But maybe before we get there, catch us up a little bit on Obvi's growth trajectory kind of since the last couple years.

2:03Ash Melwani:Yeah, absolutely. So last year, we had our best year ever, we grew 100 % year over year, changed the strategy, cracked the code on Meta. Things were just humming along. Retail was humming along and everything was great. I think it was like December slash January of this year, end of last year, when obviously this update for Andromeda started to come out. The health and wellness policy started to take effect. And so we definitely felt it. We got hit with some of the flags for health and wellness, which is really interesting because everybody's getting hit in different ways. There's different combinations.

2:43Ash Melwani:So it's like if you're a health and wellness company that is just like, you know, just a supplement brand, you know, you're getting flagged, but the consequences are different than if you were like, say, a pharmacy company, right? You know, for those who aren't familiar with why this is even happening is Meta is going through lawsuits for violating, you know, laws. And so they're essentially just putting everybody under this one umbrella and saying, if you're somewhat related to health and wellness, we're just going to flag you and we'll deal with the review later. But then once you're flagged, you're getting deprioritized in the auction.

3:19Ash Melwani:And that's what ended up happening with us, right? People are like, oh, like, are you sure? It's like, it may be your creative. It might be your landing page. Like, listen, I've tested thousands and thousands of ads. We've done iterations of our landing pages. Like, I know what works and I know what doesn't work. And I know when something is off with my ad account, because like, you don't just see this like sharp decline overnight for no reason. Right. And so what we ended up doing is, and to be fair, January 1, literally New Year's Day, our ad account got restricted. And obviously, we have a really good rep, we weren't doing anything wrong, we got it reviewed.

3:53Ash Melwani:But then ever since then, we've had this like flag for health and wellness. And so, you know, a few things that we did were, you know, we've tested, you know, different ad accounts, we've tested different pixels, domains, and, you know, there's so many things that we tried to get, you know, performance back on track. But some of the things that that actually made a difference was, you know, one being the actual pixel data and the health of our data, right. And so working with the right partners to get that set up correctly, you know, I've been telling people like, listen, yeah, there is that free app on Shopify to connect your pixel from Shopify to meta.

4:29Ash Melwani:That is the bare minimum. Now, like I would 100 % take a look at some of these companies that have tools that do this in the right way. And, and, you know, like right now we're using this company called up stack where the Michael, who is the founder has essentially created a way to not send the data that Facebook doesn't want back to meta and only send the stuff that's super relevant and can actually help performance, right? So if you are a health and wellness brand and you don't want to get flagged, setting up your pixel in a way so that you're not suspicious or under that radar from meta. So fixing the pixel health and data was very, very important.

5:16Ash Melwani:Obviously, number two being like CAC is always going to increase year over year. How do you just get better? Coming up with new landing pages and offers. We had a product that we were selling a one-month supply, a three-month supply, and a six-month supply. What we ended up doing was taking our top three selling products and creating a bundle and just this kit, which allowed us to take our AOV from $65 to$100. Not only did we increase AOV, but it allows us to spend more and afford a higher a CAC. And then I think the other thing here is honestly, like just buying behavior is so different now. Like people are seeing and hearing about brands and products for the first time on TikTok.

5:59Ash Melwani:And, you know, if you actually look at the data of like where people are searching for things, 50 % of the people are making their first search on TikTok. They're not Googling it, quote unquote, they're going to TikTok and searching for it there. So as a brand, if you're not there and you're not showing up there, you're missing a massive, massive opportunity. And so that was something that I also realized, you know, maybe later than I would have hoped, but, you know, having a presence on TikTok shop, making sure that that is up and running, even if you're just running that at breakeven, the awareness play that you can get from there is massive, right?

6:38Ash Melwani:Like I tell people like a million views on TikTok can drive so much growth versus like a million views on like Instagram reels or, or Facebook, right? Like people are using TikTok with intention and they're on Instagram and Facebook reels, just scrolling to just doom scroll. But people are heavily influenced by what's happening on TikTok. So if your brand is not on there, I don't care what your, your demographic is like, everybody is on there now. It's not just, you know, 16 year olds. It is a full range of people like our demo, our, you know our target demo is 35 plus like they're on there so if you're not there you're missing out and so i think this is the year of one diversification for us to trying to build build away from this over reliance on meta and then potentially figuring out okay well we're now in 20 000 doors nationwide across multiple retailers is this the moment that we pivot away from dtc and really go all in on retail and tone back the more direct response stuff and just go full branded tv spots influencers tiktok whatever it is and just like now solidify our space in this like market and um support retail so trying to figure that out this year is kind of like it's definitely not an up year it's like really flat for us and and it's fine we're kind of just getting through the year, obviously, there was a lot of uncertainty with tariffs and just consumer sentiment.

8:06Ash Melwani:And like, honestly, the name of the game for 2025 is just survive and hope for the best. So that's kind of where we're at right now.

8:14DTC Podcast Host:It's been a while since I've actually been running an ad campaign. But I remember that feeling of like, when you know, something's off. And so what what what actually happened with it's funny, I didn't even think of collagen as being as part of that category of what would be potentially, you know, fall under the health restrictions. Talk to me about what happened that made you tick over into that category.

8:35Ash Melwani:Yeah, honestly, I don't know. And I actually tweeted this out maybe two months ago where like I showed everybody like, hey, here's what we got flagged for. It's not a direct flag. It's an indirect flag. But from a lot of people that I spoke to, it's like even if it doesn't necessarily say you're flagged for your domain, if there's a random domain associated that is flagged, could you be penalized in the backend? And the answer to that was yes. Right. There is no official documentation about this. This isn't like, oh, like, you know, there's a, there's a blatant bug that's going on with meta. And because there's this one team, this small team of health and wellness people that are like just focusing on this, you have no access to them.

9:17Ash Melwani:My rep has no idea. A bunch of people's reps have no idea. Like I've been talking to other, you know, agency owners who are well connected within meta that have their contacts that are like super high up at meta essentially saying like, yeah, like this is an issue. This is a known issue. We don't know when a fix is coming. So now it's like, you're, you're kind of playing against this behemoth that there's just a, that's just a black box. And so, you know, what we ended up seeing was the one thing that my rep pointed out, which was really suspicious was the day that we got flagged, I get a little bit tactical here, but what you can do is there's a ratio of landing page views divided by total outbound clicks, right?

10:03Ash Melwani:So say you have a hundred outbound clicks from your ads and you get 90 landing page views, right? That's a healthy ratio of 90%, meaning 90 % of their traffic let the page load and they actually got to where they needed to go. If that ratio is sub 80%, you know, 50, 60%, that means half of your traffic isn't even letting the page load and they're going right. And so on the same day we got flagged, our ratio went from 85, 90 down to 55. And so what does that even mean? It means one, the traffic that we're getting is half of it is bouncing before they even get to the landing page. And if half of the traffic is bouncing, that means your CAC is literally doubling.

10:47Ash Melwani:And so this happened, I can literally pinpoint day by day, this ratio being anywhere from 80 to 90, the day of the flag, it hit 50%. And I was like, okay, this is definitely an issue. And then we go in, we get the domain reviewed, the appeal was rejected, which made no sense because the domain actually had nothing to do with obvi at all it was a random it was a it was a domain for a um a funeral home in florida it made no sense right but it's now associated to our brand and so i talked to a few people and this is when michael from upstack actually introduced uh or he reached out and said i've been seeing this across multiple brands this is all falling under this health and wellness bug that's going around let's see if we can fix you know some of the pixel stuff right so literally switching over the different integration with his system of like removing certain, you know, information, sending it back to Facebook.

11:46Ash Melwani:I'm not even kidding you, like seven days after our ratio started to go back up. And so that in itself is a signal. It's a positive signal back to meta saying, Hey, people are engaging with this page again. Let's give them better traffic, this and that. And then you can literally see our CPMs come down, our cost of traffic comes down, our CPAs start to come down. And I'm just like, this had nothing to do with us. Like we were doing everything right and then this random domain gets flagged on our account and we're getting penalized it made no sense and my rep had no idea his boss had no idea and so that's why i'm like if somebody comes to me like whether it's on mentor pass or you know somebody messages me and it's like um can you audit my ad account the first place i'll go now is events manager and see what's going on within there because we had a solution set up before all the scores look perfect everything there was no errors whatsoever just a different solution that was sending back information and perhaps a little bit better than than others fix the issue and so it's like you gotta just make sure your pixel is like in place because with andromeda's update it is the system is way more faster to pick up on irregularities and so it's like if there's something that's broken you will start to get deprioritized right away.

13:03Ash Melwani:Whereas before it would take maybe a week, two weeks for the system to catch up. But because this AI is moving so fast, it'll know when to deprioritize you. So like if something is broken on your website or pixels not flat or firing, or, you know, there's a, there's an ad that's getting too many negative comments, you will get deprioritized so fast that you'll be like, Oh, my website's broken. Like, no, it's the algorithm punishing you because something is broken.

13:30DTC Podcast Host:Because it's a self-fulfilling prophecy probably. We've seen this on some Google accounts where you get a bad signal in the mix and then it just trends down to the point where, right? It's sort of, it's a loop, that doom loop.

13:43Ash Melwani:And it's just a faster response time for meta. So like, which is great because if you do do well, that's a great signal and you start to perform better, right? Instead of like, oh, I got to wait a week, two weeks for things to optimize. Like things are optimizing way more quickly now, which is great.

13:58DTC Podcast Host:So when you launched this new pixel and rebuilt your campaigns, great opportunity to put your sort of new thinking into practice when you do the redux. What was your sort of thought process with how you built the campaign back, maybe creative and audience wise? Were you basically duplicating what you had already done before or did you start with kind of a new thesis?

14:17Ash Melwani:Yeah, I mean, it's a great question because, again, like even throughout this entire year, because like since January performance was like kind of shit. And so a lot of it was trying to figure out the best way to run the ad account post Andromeda, right? Also, you would have like really good like runs and then you would have like performance would fall off a cliff, right? And that was happening. And if you go and look back at the conversations, like again, we have this little bubble on Twitter, but like they're pushing algorithm updates consistently. So like when there's updates like that, you're going to see volatility and we felt it, right?

14:55Ash Melwani:Now, for me, it's like, okay, well, what I was running in my setup I was running last year, is it still the best practice post this update? And so I wish it was a little bit quicker, but after talking to a lot of people, it's like, okay, well, one, over-excluding can hurt you. Two, having or not having enough diversity and volume in your ad account can also hurt you. The way that you test and testing by concept and angle and having like three or five variations of that angle and testing that by itself can also hurt you, right? And all of your best practices that you were running for the last, you know, 12 to 18 months completely changes overnight because of their new system, right?

15:45Ash Melwani:So now Andromeda has this AI that says I can process way more ad creatives in a single second than I could previously, right? So what does that mean? If I give meta one ad concept and I do three to five variations, and my variations are very different in the sense of maybe I have the same ad, but I'm testing a different headline, Facebook is going to be like, well, this isn't diverse enough. I'm just going to prioritize one of these ads. And if one of these ads doesn't work, then the whole thing is not going to work. Now, instead of giving Facebook five variations of one ad, can I give it five unique ads with maybe two variations, right?

16:26Ash Melwani:So at least it has a way more, way more options to pick from. And instead of testing those five new angles separately, can you test them together, right? And so what Meta is doing now is like, okay, if you give it 10 creatives, completely different, all visually looking different, angles are different, messaging is different, whatever, all the differences you possibly can do, it's going to look at these 10 and say, okay, well, this angle doesn't necessarily work. This version doesn't work. This one doesn't work. Here are the ones that actually work. It gives the winners more context to see what actually works versus giving it one variation.

17:08Ash Melwani:And there's no additional context to say, is this the best of what I can provide to the audience? Right. So the way that I started to test was different. I loosened up on my exclusions. I wanted to give meta more signal. So like I used to exclude website visitors and customers, right? I'll still exclude customers, but I removed that website visitors because that warmer audience is fine. I want them to see the new ads because what that'll mean is the warmer audience is going to start clicking on these ads and engaging on them and then potentially convert. And even if it's maybe slightly retargeting, that's still a positive signal back to Meta and Meta is going to prioritize me again in the auction, right?

17:47Ash Melwani:So I removed the exclusions. The other thing was instead of testing 50 ads a week and that being like 10 net new angles with three to five variations and then having 10 ad sets, I'm cutting that down to two or three ad sets with like 10 to 15 ads within them, right? So now not only am I consolidating and competing less with myself, so there's less auction overlap, I'm able to test a lot more batches quickly. And so that was probably the biggest thing and what I changed in terms of testing. I was big, big, big on cost caps last year. For my testing, I've kind of moved back to automatic bidding just to kind of force spend, right?

18:32Ash Melwani:Because I have these 15 ads, I need enough spend to go through every single one. So testing on the lowest automatic bidding, but because I'm now consolidated, whereas I would have 10 ad sets, I have to give each ad set$400. That's like four grand to test 50 creatives. Now I'm giving three ad sets$500. So instead of testing for$4 ,000, I'm testing for$1 ,500, right? So it kind of balances itself out. And then just scaling, right? You just have your winning campaign with all your winners in there and letting that go. And yeah, still leveraging manual bidding as like a side campaign.

19:10DTC Podcast Host:So whenever there's an opportunity for something to spend, I'll spend.

19:13Ash Melwani:But those are probably the biggest changes that I made, which was consolidating a little bit more, making sure that the creative output was way more diverse than it was before. So my team knows to only give me two variations of a single idea now instead of three to five. And then, you know, giving them more resources to create more diverse and volume.

19:41DTC Podcast Host:Higher contrast tests as well, right?

19:42Ash Melwani:Exactly. Way more higher contrast, you know, making sure that there's an equal split between static ads and video ads. And then within that, all the different formats that you can do, right? So like static ads can be, you know, your product with the main headline on it. or it could be an us versus them, or it could be a review. It could be an offer. Like there's so many different formats. So making sure that we have all those formats is important. And then on the video side, I mean, there's so much you can do with video. UGC is just a single format. Like you have podcast style ads, you have street interviews, you have, you know, founder, you know, ads, whatever it is, there's so many different formats.

20:19Ash Melwani:You just have to make sure you're, you're giving the ad account as many formats as possible and letting it rock. And then the last Last thing I would say is for CRO testing, I'm not even kidding you, man. I just ran a test last week. I changed the above the fold copy. I took the top three performing ads right now. Whatever was my copy on that ad, changed the copy on the landing page to kind of just match that a little bit. 30 % lived right there. Right. And so when you start thinking about like, OK, well, this landing page can always get better. Well, on a weekly basis, what can I test on this landing page?

20:58Ash Melwani:Like even if it's just copy, even if it's just like changing the imagery, those like 10 to 20 percent wins stack and will allow you to continue scaling. So I'm making sure that there is no week where I'm not running a test just to learn something. Even if it's just like testing your CTA button copy instead of buy now, it's claim offer or the first image that shows what is that image? Is it a lifestyle image? Is it a render? And then thinking through offers, how can you give as much value for the most margin for the business? So those are the things that I've been focusing on a lot. And then obviously this TikTok shop stuff is a whole beast on its own too.

21:38DTC Podcast Host:Before we dive into that, I just have a note here. I don't know if it's still relevant, but I saw a tweet about Proxima and using AI built audiences during the reset. Was that is that something that's still working? Maybe describe that a little bit. Yeah.

21:49Ash Melwani:So Proxima, one, I've been Alex is amazing. The team is amazing. We had an issue probably three years ago where our ad account was just like so focused on this one audience that was brought in from this like one influencer that we really leverage in our advertising. Long story short, that influencer, when she stopped working with us, the ad account was trained for her audience, right? And her audience isn't really our target demographic, but now Meta thinks it is and it's only going to go after that. Now, what Proxima did was it kind of forced the algorithm to relearn who our audience was based off of the high quality data points that they gave us.

22:30Ash Melwani:So using lookalikes to kind of retrain the algorithm. Proxima has always been like a tool in our ad account where I would, if my scale campaign, you know, is a single ad set going broad, I would have a separate Proxima ad set with all their audiences, all the same winning ads and a CBO. So one day Proxima does really well, one day Broad does really well, limiting my risk by giving two audiences Facebook can go into and scale for the day. And so when ASC started to roll out, I didn't necessarily have a use case for Proxima anymore because it was just Facebook would decide the audience for you, right?