In this episode Eric Siu and Neil Patel discuss recent findings by Rand Fishkin (Moz, SparkToro) that a large portion of social media referrals are incorrectly recorded as direct traffic, a phenomenon known as "dark social." They highlight how platforms like TikTok, Slack, Discord, Mastodon, and WhatsApp show 100% of their traffic as direct, while Facebook Messenger and Instagram also have high misattribution rates. The conversation then shifts to the impact of privacy consent and cookie opt‑outs, which further inflate direct traffic and make traditional analytics less reliable. Both hosts argue for moving beyond sole reliance on Google Analytics toward multi‑touch attribution, media‑mix models, and brand‑recall studies. Practical workarounds are shared, such as asking leads how they heard about the business, capturing first‑ and last‑touch URLs, and surveying email lists. They also describe their current measurement stack—custom dashboards built in Google Data Studio, Looker, and HubSpot—emphasizing a shift from pure traffic metrics to business‑focused KPIs like pipeline, MRR, and LTV. The episode concludes with a call to adapt measurement practices to the increasingly opaque, privacy‑driven web.
Chapters
All right. So we're going to talk about how dark social falsely attributes significant percentages of web traffic as direct. So this is a post from Rand Fishkin. So he formerly, he founded Moz. He's the founder, one of the co-founders of SparkToro. And so Rand Fishkin over here, you can see on the SparkToro blog. So he basically says that at a high level that— actually, I'm going to pull, I'm going to go all the way down here. So takeaways, I'm going to go to takeaways over here. Takeaways. If you are a marketer or website owner relying on analytics platforms, Google Analytics, or anything else to provide accurate comprehensive information about where your visitors came from and how they found your website, we recommend revisiting those assumptions. A substantial portion of social traffic referral— social referral traffic comes without proper accurate referral data and is misclassified as direct. If you're attributing all direct visits to type-in visitors, email, text message, offline brand campaigns, or dark search, because of dark social, you're almost certainly undercounting the impact of social media marketing, social referrals, and word-of-mouth networks. Now, let me call out one more thing real quick before we move on here. So in this graph up here, you can see the results at a glance. You can see basically 100% of TikTok traffic is attributed to direct. Same with Slack, same with Discord, same with Mastodon, if anybody uses that. WhatsApp is 100% as well. Facebook Messenger, 75%. And then Instagram is about 30% or so. So let me explain dark social really quick and then Neil, you can give your reaction. So dark social basically just means it's like the social media posts that you post and you're basically not able to attribute it, which is why it's called dark.
But you know what's funny? You're going to see even more direct traffic because when people come to your website, typically you have consents. Like, do you allow cookies? Do you not? Do you allow tracking? Do you not? Most people say no. Like we looked at our stats the other week, over 68% say no. In essence, they're saying, don't track me. That's a lot. Now that's over a week. It's not the whole web. And yes, our audience is marketers, but still, you know, we have millions and millions of visitors a month. So the sample size is large enough. It's going to get worse and worse. We're going back to a model in marketing where You have to start doing things like media mixing models and things like that to figure out where you're getting an ROI and where you should be spending your money. You know, people want more privacy, they want access, or they want to control the rights to their data. And I'm not saying these are bad things at all. It's just changing the web and marketing landscape to being more of a quote unquote traditional model where you got to start doing multi-touch attribution and figure out, all right, Where are people coming from? Brand recall studies, where are people actually buying from and how can we go and take that data and figure out where to allocate our budgets?
Yeah, look, 5, 10 years ago, everyone's like, everything's going digital. You have to be able to track everything. The reality now is things are becoming less trackable. And what you could do is you could, for example, when we have a lead that comes in through the Single Grain website, we'll ask them, hey, how did you hear about us? Like, where did you come from? We'll also see the first URL, which where they submitted. Or where they came in from. We'll then also see the URL, the last URL where they actually filled out the form. So we're looking at first and last touch attribution, multi-touch, more ideal. But then we're also asking which search term they put into whatever platform they came from. So we're trying to get that qualitative data. And then let's say you have people on your email list as well. You can also survey this list and say, hey, where did you first hear about us? So it's not like this stuff is completely opaque. There are ways for you to capture your audience. When the leads are coming through, you can just capture that data. Yep.
Look, I think with marketing, you have to start doing, companies should, I think they need to make a much bigger investment in brand recall studies. Look to see where your visitors are coming from and look to see where your buyers are coming from. That'll give you a good understanding of where you should be investing your dollars. Don't just rely on analytics. The data is getting harder and harder. to track becoming accurate because of a lot of the laws and the changes that are coming, especially with the cookieless future that's happening soon enough, right? So you need to start investing in brand recall studies to figure out where you should be investing your money.
Yeah. I mean, Neil and I used to look at Google Analytics every day and now like I can't even look at it because it's so difficult to slice and dice. It just makes me not want to log in anymore.
Dude, when's the last time you logged into your Google Analytics?
I log in like maybe once every 1, 2 weeks or so because I do have reports set up, but it's just like, it's not easy. Like you can't even, I, you know how like you can, you can used to be able to make the graph like it was sorted by daily where it was hard to read. So then you get sorted by weekly and then monthly. You can't even do that anymore.
Yeah. Yeah. I haven't logged in in a long time. I probably haven't logged in in like 2 months. But on that counterpoint, I do have a daily dashboard that pulls in a lot of analytics data that I see every single day. So it's, Kind of like the same thing as logging in. It's just we've created our own custom reports that break down everything we need. So then I just look at that every single day and it's not just visitors for me. It looks at visitors, sources, lead conversion, pipe, pipe to close, average sales cycle, MRR, estimated LTV, et cetera.
Yeah, it's interesting. The dashboards I look at now are either GDS or Looker dashboards. And then also we look at our HubSpot reports as well, just to look at pipeline as well. So we're looking at more business side of things, I think. So I mean, maybe when we used to look at this stuff pretty hardcore 10 years ago, we were more focused on traffic.
Yeah.
Now we're more focused on kind of general business. So, you know, time changes. So that is it for today. Please don't forget to rate, view, subscribe, and we'll see you tomorrow.
