Where Web Users Spend Time vs. Where Traffic Referrals Originate

“Most of our site’s traffic comes from Google, so most of our digital marketing efforts and spending should be on Google, too.”

– An absolutely terrible way to invest in marketing

On Monday of this week, I published a comprehensive look at how traffic flows on the American web. Unfortunately, some folks in the social feeds where this research was shared got the wrong idea. Yes, Google sends close to 2/3rds of all referral traffic to US websites (of the top 170 referrers). No, that doesn’t mean Google is the best, only, or most important place for every website owner/marketer to put their growth efforts.

Why not?

Because demand creation and traffic acquisition are different practices. I.e.: What people read, watch, listen-to, consume, and visit is different than what sends traffic. You’ve likely seen studies about how Americans use the Internet with stats like:

  • 85% of Americans use the Internet daily.
  • The avg US Internet user spends just under 7 hours online/day
  • 2 hours, 27 minutes are spent on social media, on average
  • 2 hours, 37 minutes are spent on streaming video, on average

Thanks to Datos’ clickstream panel (full methodology described in the report), we’re able to dig deeper with details about which domains received visits vs. sent referrals from January 2023-January 2024. I took all 170 domains that were part of our study and hand-classified these into the 13 categories below:

You can see how dominant search is in referrals: Google, Bing, DuckDuckGo, and Archive.org are all in that category. But when it comes to the share of visits (for analytics folks, these are measured as the number of monthly unique users in Datos’ panel who visited the site), the Search category only comes in 4th!

Let me show that another way—in a pie chart breaking down the share of visits:

Social media websites like Facebook, Reddit, LinkedIn, TikTok, Threads, etc. make up the largest portion of unique visits. According to other sources ~73% of all Americans are monthly social media users, which tracks very closely to what we see in Datos‘ panel (the biggest difference is that I classified YouTube with other video/audio sites like Spotify, Netflix, Vimeo, Hulu, etc. whereas most social media reporting treats YouTube as a social media platform).

Next are productivity sites: Microsoft, MS Office, Live, Github, Canva, Indeed, Docusign, Dropbox, and the like. I was torn on the classification of OpenAI, but ultimately decided it belonged in productivity more than search. Arbitrary classifications are a challenge, but ultimately, it only makes a 1.15% difference.

News sites like CNN, Fox, MSN, BBC, Washington Post, The Guardian, and dozens more are included here, and they collectively get a lot of traffic and engagement. Similarly, e-commerce domains like Amazon, BestBuy, Walmart, and Etsy get plenty of regular visitation, too.

Only after these five other categories does search come in. It sends the lion’s share of traffic out, but it doesn’t receive the lion’s share of time and engagement. People come to Google and Bing to learn and leave (or to navigate the web). Time-on-site for search engines tends to be quite low (though Google’s zero-click answers and SGE investments are undoubtedly designed to combat that).

If you’d like to see my full classification of all 170 sites, I’ve made that available here, and a preview is below (warning, it’s 3,500+ pixels in length).

Full classification breakdown is here

My goal in presenting this data is to help dissuade anyone misled by our “Where traffic comes from” study from investing as though those are the only (or most logical) places to do marketing. The web is a big, diverse place and people consume content, learn about products/services/brands, and form opinions about which problems to solve, what to prioritize, who to trust, and where to go next outside of search engines.

My hypothesis is that an ever-increasing percent of how people use Google is navigational. In 2008, a study estimated that 10.2% of searches were just people trying to reach another URL on the web. That was before Google Chrome had dominant market share and browser toolbars/the mobile web turned almost everything typed into a browser into a search.

Regardless, what I know from these data sources (Datos’ panel, the surveys on websites usage, stats from third parties, all of it) is this: traffic referrals ≠ sources of influence. The latter is far broader than the former. When people watch videos on YouTube, browse news on the Washington Post, scroll through posts on Threads, dig around Etsy, or trawl through the long tail of the web, they’re exposed to the ideas, information, problems, and brands that turn into search queries.

Is a search-exclusive marketing strategy viable? Rarely.

Can a strategy of building brand awareness, interest, and demand outside of search work? Absolutely.

Just be careful not to ascribe attribution or credit to Google when other investments drove the real value. For most brands, a huge portion of search referrals come from branded search queries, and almost none of those originated on or because Google. Yet so many marketers throw branded-search-assisted conversions into “sales attributed to Google.”

The only reasonable move is to invest in the sources of influence to which your audience pays attention: social accounts they follow, YouTube channels they watch, podcasts they listen to, websites they visit, events they attend, etc. Anything else is missing out on opportunity.