Search Vs Social
Back in 2008 and 2009, probably because of our investment in Twitter, I was obsessed with looking at acquisition traffic from search and social and comparing the two. Here’s an example of a post I wrote at that time looking at the two channels and comparing them.
At some point, I became convinced that websites would eventually see more acquisition traffic coming from social than they were seeing from search, which was the king dog of Internet traffic at the time. It was a hotly debated issue but, again maybe because of how long we were on Twitter, I was convinced social would be king some day.
I stopped obsessing about that issue sometime around 2010 and moved on to mobile as the thing I thought about and wrote about the most.
But today, when I was looking to see if the traffic to AVC had declined a lot while I have been away and blogging about all sorts of non work stuff (it hasn’t), I saw this chart of year to date acquisition traffic to AVC.
Social, and for AVC that means Twitter and Disqus, brings 23% of the visits to AVC. Search brings less than 20%. And it’s been that way for a long time now. Well over three years.
It’s kind of funny to think that we wondered and debated about such things back in 2008 and 2009. It’s not a debate any more and its not something to wonder about. It’s reality now.
I expect certain versions of our bitcoin and mobile ecosystem conversations of today will seem equally obvious in 5 years’ time.
Ana – quick, which ones 😉 ?????
Or like the Springsteen Rosalita lyrics some day it will all seem funny:Now, I know your mama, she don’t like me ’cause I play in a rock and roll bandAnd I know your daddy, he don’t dig me, but he never did understandYour papa lowered the boom, he locked you in your roomI’m coming to lend a handI’m coming to liberate you, confiscate you, I want to be your manSomeday we’ll look back on this and it will all seem funnyBut now you’re sad, your mama’s madAnd your papa says he knows that I don’t (have any money)Whoa, your papa says he knows (that I don’t have any money)Whoa, so your daddy says he knows I don’t have (Papa says he knows that I don’t have any money)
I think search will always have a place for the long tail. More specifically, I think social is great for content-oriented sites while search is better for commerce-oriented sites. So, for sites like Amazon, Yelp, or even Zillow (they are closely associated with a large transaction), I think search will continue to provide very substantial traffic.
yup. i totally agree.
You could argue e-commerce simple hasn’t embraced social. Amazon for example does not make it a fluid experience to share your purchases or even monetize your influence through your network. Or they could certainly lower the barriers and integrate more deeply into Twitter and Facebook.
.eCommerce sites that are driven by creating a community will have something to talk about rising to the potential cult levelAppleJLM.
SHARE PICTURE OF CAT? HILARIOUS!SHARE “BOUGHT FOOD FOR CAT”? BORING!
Bang on.Task completion is the next big functionality wave that’s going to hit local, and it starts with someone finding something, consuming the content and then taking action.
Statistically, you have one data point here. In an extreme, almost all traffic of twitter comes from social, but that doesn’t mean social is the major traffic channel.
Have you done any posts about this?
No, I don’t have a opinion on the comparison. I just think this comparison is not fair.
I am surprised for more temporally relevant information (such as a blog) where the conversation is such an important part of the content that search is as high as it is.Did you plot traffic source vs post age?
a lot of “organic search” is really direct. see this chart of the top keywords
This isn’t (necessarily) direct – this is just Google swallowing the keyword used to search.
Yeah, (not provided) means the user came from an HTTPS google page (which should be basically everybody now). (not set) is the “keyword” shown for direct traffic in GA.
where the conversation is such an important part of the contentNot to mention the fact that that conversation (last I checked, might have changed) is or was not available by search.
YupNo one would fund or build a business where cust acquisition was SEO or PPC based today.I have serious scars from the crash of search within that same time period.
Also within expectations…your bounce rate is a little higher and new users lower with the social vector…I read that as a fan base that responds to direct article shares. It’s not just about the numbers…I’d take the engaged social stream over the organic search tire kickers any day. Of course I would take any of your numbers, any day 🙂
Google Analytics’ attribution methodology under-counts direct visitors (sometimes dramatically). If a user visits your site via a search engine then comes back directly every day for a month, all those visits will be considered search referral visits. Same with site referrals. Direct visits only get credit if there is no other referral source for the prior six months. This probably makes sense for an e-commerce site that needs to understand how they initially attracted a customer, but doesn’t work so well for a content site that has regular visitors who may be clicking to your site from a variety of sources. The referral settings can be overridden, but that’s how it works stock.
Very interesting comment. Wasn’t aware of the repeat visits attribution, surely the same would be true for social? If someone finds AVC through a tweet then everything for the next six months gets allocated to that.
Analytics is generally messy.
Depends how the initial google analytics attribution model is set up.
You touched on mobile, but didn’t share. What percentage is mobile and how has this changed?
you have data from Disqus on this generally that you can share across all the sites? By category?
Disqus has seen major traffic growth since 2010 and the share of mobile is way up. Mobile traffic accounted for about 1% of our footprint then. Today, mobile is about 30% and continues to climb.
Thanks for sharing Ernest!
You’re very welcome! I won’t be sharing any more numbers today, but what is also interesting is the percentage of comments that come from mobile devices (excluding tablets)… It’s higher than what I’d expect. =)
so what kind of user experience is avc/ disqus on mobiles Jim?
That would be for Fred to disclose…not me.
Bars were always more popular than libraries.Funny we questioned it.
That’s an awesome line
Yes, but google is still a pretty popular library.
INTERNET STARTED BY PEOPLE IN LIBRARIES.ONES THAT LIKE BARS CAME LATER.
I missed you!
NEXT TIME AIM BETTER.
How many libraries do you know that are worth more than every bar, COMBINED?
separate subject … personalized search from google is really starting to bug me .. i am so much smarter than their algorithm thinks i am in terms of knowing what i am really looking for
Google only needs to be smart enough for the average user “normal” not for you.
I find the breakout of new users most interesting. Organic search (22%) is slightly higher than social (19%), which is impressive given that search has been, as you put it, the king dog of acquistion. I’m not surprised that traffic from social is higher overall because people can more or less subscribe to AVC via FB or Twitter. Can you project from how the data are trending of when social may overtake organic search for new users?
I would think that your newsletter would bring most traffic to a site like avc.com
You’ve explained often enough your problems in finding content, e.g., blogs, video clips, music, on the Internet.Why? In part because ‘search’, that is, Google/Bing with just their keywords/phrases, are at best poor at letting people solve the problems you describe.In particular, such ‘search’ will have to be poor at feeding people to AVC, and various cases of social will be more effective.But social is fundamentally manual, that is, we should call labor intensive, and, thus, severely throttled at scaling.So, how to get a solution that will scale, e.g,, software to eat the problem as in “software is eating the world”? Well, much of the challenge, and a big advantage of solutions via social, is somehow effectively enough getting at or approximating the user’s intended meaning. Sure, as Page/Brin along with Norvig might say, real artificial intelligence (AI) might be sufficient. It is also, for now and the foreseeable future, infeasible, unrealistic, pie in the sky.But, is AI the only possible automation, scalable, software solution? Nope. So, how about intuitive heuristics, the social graph, natural language processing, …, that is, the stuff easy to find and read about? IMHO, nope, won’t work.What will work? I can think of only five parties interested now in how, my gray tabby kitty cat, my plump Siamese kitty cat, me, myself, and I. But, last night I got some more code written, really just routine code in one of the Web pages, e.g., use some ASP.NET classes to insert some data in some text boxes!
direct is so big too! over time, i imagine lot of search and social traffic converted to direct (i did!) because of the content and community here on avc.
Fred – I am curious to hear your views on “attribution”. With the proliferation of tech & touch points, are we fooling ourselves by looking at metrics in a “silo” fashion? I imagine the reality is perhaps more complex & involves interaction of multiple touch points.
It reminds me of the ‘Apps vs. Mobile Web’ debate, and via which medium mobile users will engage and spend more time.I was walked through comScore’s Mobile App Report this morning, and was reminded (empirically) how it is NOT EVEN CLOSE insofar as where users spend their time.But we used to be less sure. And like most things, it depends on the business.I love posts that I call ‘Internet Memoir,’ that give us this perspective.
as an aside, i’d be interested to know what percentage of blog comments are written using mobile phones.
I’d be curious about the direct traffic and how they first discovered AVC. Between social & search, which traffic is sticker?
Hmm but I wonder how much of that behaviour is distorted from regular visitors rather than acquired. Depending on the time of day you tweet out I might decide to check in via twitter or directly or type avc into the search bar. Its random for me.
.Traffic can only come from where the people are located. More bandwidth and time is now dedicated to social so it comes as no surprise that people are coming from where they are located.Big eCommerce sites will have more traffic coming from search and big social sites will have more folks coming from social.JLM.
What is the time frame for these stats – last 30 days?
year to date. i explained that in the post
Thanks, I somehow missed that part. These are really impressive numbers.
All this is telling is there’s a stable diversification of traffic sources.
Funny how fickle social is though. If an entrepreneur walked through your door and said we are going to get “customer acquisition via social”, not sure you can rely on it. They have to have an awful lot of traction to prove it’s real.
i view social less fickle than the google search algo’s. i suspect that most digital content companies (buzzfeed, gawker, BI, Huffpo, etc) generate a significant portion of traffic from social. So back to Fred’s point, social has scaled to the point that it should not be viewed as a niche or less than ideal approach vs search.
Your blog is inherently social. It would be interesting to see the numbers for ecommerce and non-personal websites. Also, which converts/transacts better? Eyeballs aren’t the only metric for performance. What is the level of intent for each audience? We often see the level of purchase intent higher for organic search users (those looking for a particular product). Given that, my guess is that search is higher on the intent chain compared to pure social traffic. There is no doubt that influencers like yourself can help seed intent and even generate a number of transactions, but conversion success is likely a function of influence level.In short, Search vs Social isn’t purely about traffic or even user acquisition. It must also look at conversion success, and long-term retention (social is likely your retention mechanism).
That’s a strong conclusion for a sample size of n=1. Just because the reality played out on your site because of how you’ve promoted the site doesn’t mean social drives more traffic than search engines. What have you done to promote SEO on this site? Not much. What have you done to promote social? Quite a bit.
Maybe the problem is the year over year decline in interest in entrepreneurship and venture capital. From Google search trends:Venture Capital is the first graph and then the second one is entrepreneurship.
We need Units for that Y AXIS!!!!!!!!!!!!
That’s one of the big annoyances of Google trends. All it is is percentage search volume relative to the top point (100%). So you can know half of the way to the bottom has 50% search volume of the top point but we don’t know how many queries it is.—Sent from Mailbox for iPad
How about cuting the data by Mobile Search and Mobile Social? That might be interesting.
I’m not surprised. Social is a much more natural outlet for content than search is. While you look for things to read, you tend to discuss what you read with people
” It’s not a debate any more and its not something to wonder about. It’s reality now.”For this site. I’m guessing that there are plenty of car insurance sites that would have a very different picture.
Be really careful, because “Direct” is more of a “we don’t know where it’s coming from” than people typing the URL in their browser :)In the day and age of HTTPS, the Referer, which is used to measure this is now obfuscated, resulting in an increased of “direct” visits…
I’ll bet the majority of those “Direct” are “Search” indeed, namely https.
Yes, good reminder on how HTTPS changes analytics.
I think it really depends on the contents. This blog has good, shareable information for social or linked people. Some other things like local products, commodities and transaction-driven contents are still driven by ‘Google´s empire of generics’.Social still misses the most valuable traffic of all, that is, the traffic with purchasing intentionality and this makes things quite uncertain, still in 2014.
I was thinking about exactly this with regard to the Scottish referendum this week. All of the news that I was reading about it was solidly “no”. I thought – “this is odd because there must be loads of passionate ‘yes’ news out there too”.It made me think – democracy is built on reasonable debate and compromise. If people (as I do) consume more and more news via social which reinforces their opinions more and more, I wonder if social media is not (only) the silver bullet for “democratising information” but also the bullet which kills democracy as people become too entrenched in their views and cannot compromise.Hmmmmmm….
Hey Fred! 80% bounce rate is not a very useful information.Consider tweaking your bounce rate calculation to a more suitable setting (e.g. count as bounce if below X seconds on the page).The default Analytics settings are more suitable to ecommerces, not as much to content websites.Take care!
I’m not seeing a big disparity between search and social ratios on a lot of the sites I work on. Publishers got slammed pretty hard by Facebook algorithm changes designed to disadvantage them (if they’re not paying for placement) last year and that may have something to do with it, but even on news sites I advise search and social are pretty even, and for sites that do a lot of reference and evergreen content, search is slightly higher. Not by much, granted, but long tail is pretty powerful for reasonably optimized sites. (Commerce sites are a different story, though. Search all day long.)Also: your direct traffic is crazy. (By which I mean crazy awesome, not rigged.) I agree with one of your commenters that some of that may be https confusing analytics, but for an indie publisher, I’d expect direct to be 15 – 20%. You have an incredibly loyal audience.
It’s a bit sketchy to declare victory for social over one blog’s set of numbers. The two money quotes I’ve seen in these comments so far are these:1. How many libraries do you know that are worth more than every bar, COMBINED?2. You could argue e-commerce simple hasn’t embraced social.Let’s expand on those comments.First, where Social works is really a function of what you might be willing to be distracted by while being social. That’s why Google is so valuable. You’re willing to be distracted during search because the distraction is precisely the object of the search. And in fact, it is so lucrative because in addition, the distraction you often are seeking is to purchase something NOW. That’s a very valuable distraction to control indeed.What follows is an obvious conclusion. Don’t assume Social could succeed for e-commerce if they would simply embraced it. Believe me, many have tried and it fails because people are not as interested in being distracted from their socializing to go buy something. It just doesn’t work nearly as well and you’ve got some of the more forward-looking market strategists starting to de-emphasize it as a result. Even straight up advertising doesn’t work nearly as well as search for the same reason.Never mind buying behavior (even though it is absolutely critical to monetizing these silly platforms), aren’t there other reasons for social to succeed over search? It doesn’t seem like it, but maybe. If nothing else, social has been here long enough to demonstrate whether it has the potential to be a major paradigm shift for all things or just for socializing and it sure looks like the latter more than the former. Can somebody invent an entirely new platform that is revolutionary? Maybe, but the network effects of the old ones sure make that hard.The other fallacy is in the quality of traffic. Fred says, “more acquisition traffic coming from social.” A literal reading of that suggests “acquisition traffic” = “new users.” Sorry Fred, but your Google analytics above show more new users coming in via Google still. And I’ll bet if you could go back in time and tag every one of those social visitors to know how they got here originally, the overwhelming majority would also be via search. So Social is the new RSS–great for getting visitors to repeat, but not so hot for attracting new visitors.Let’s forgive the rumors of search’s demise as they are surely overstated.
For eCommerce Retail, B2B, Travel, and pretty much anything besides niche content, Social has really failed as a customer acquisition source. I don’t think we have ever seen an “acquisition channel” with a bigger negative ROI than Social. Our company (Bounce Exchange) is privy to data across nearly 1000 enterprise eCommerce sites. Email drives 50-200x the sales of social. Search is 25x-200x. We saw 3 exceptions across nearly 1000 data points. 2/3 3 exceptions have lost nearly all the social traffic they had and 1 shut down operations. No one has really talked about how much social has failed from a commerce standpoint in a lot of ways.
Great vision. It wouldn’t have been obvious in 2008.Still, I think this analysis is too superficial.In this case, social might drive more traffic than search. Great.How engaged is this traffic though?From your table, I can see than social generates significantly more bounce, social visitors spend less time on your site, and view fewer pages per visit, than search.Mathematically, if you multiply the number of unique visitors by 1-bounce rate, search generates more actual visitors (91k for search vs. 77k for social). In you take into account the number of ‘actual page views’ rather than the number of unique visitors, the difference is even greater (135k for search vs. 97k for social).When you think about it, you can expect someone who specifically searched for you to be more engaged than someone ‘stumbled upon’ your content on social media. I assume this is a different approach of content consumption. Social visitors seem to have a shorter attention span.I’d be interested to see your conversion metrics (newsletter signups, maybe?) depending on the source of traffic.Interesting stuff nonetheless 🙂
Same Thought- Include some sort of metric assigning quality of user. That being said, I dont see social ceding its run any time soon
Great point! A quality metric based on user… so tooling on this right now.
Are you still a heavily invested in Twitter? And would you comment on the fine line between pitching investors such as yourself when you were so instrumental at Twitter, when there is potential conflict either by money or connections, and pitching VC’s that have no hand in social (which is a task near impossible).
yes i am still heavily invested in Twitteri generally will advise entrepreneurs not to pitch me if i believe their pitch would put me in a conflicted position
Thank you. I really respect you. Especially answering what could be construed as too personal question. Always wanted to work with you, but you are invested in some competitors of ours. Maybe someday. BTW, because you are such a sensible voice from the VC community, I am wondering if you would ever revisit you Employee Equity posts. At least an update on the RSU which at the time you quoted general council as ‘an adventure in tax land.’ That was years ago.
I came from the disqus newsfeed…. In some ways, social (like twitter) is a form of search… with interaction. I always go there first to find what’s trending or recent up-to-the-second news events, and can read related opinions. Google is about 4 hours behind at best – and that’s just for what THEY think is important and relevant – not what IS.
that’s true, thanks for ypur sharing
los medios de comunicación son muy importantes en el mundo globalizado por que permite manifestar inquietudes sociales y pensamientos sobre la tecnología y su utilidad social de conocimiento entre otros es una cultura futurista. o quien puede decir que este pueda contactar seres de otras galaxias con códigos de lenguaje evolucionado de interpretación? o ver imágenes del mas allá desconocidas que ya están en programas como discovery channel y otros. entonces el manejo y utilidad de las tecnologias son trascendentales e inherentes al ser humano por lo desconocido y el saber mas
I doubt that using data from other people in that way will work very well.
but that endorsed results from qualified connections would be accessible somehow, so we can skip the sifting.Depends on what you are searching for is actually the issue. I might think a search about “Lancaster PA” or “Guitars” “random thoughts on scum music industry execs” that you endorse would be good but not necessarily about “LED Replacement Ceiling lights”.
My standards are higher.Such a source can work well with a lot of additional context, but for a software approach that context will be missing.For a software approach, part of the problem, really debilitating, is the issue of interest. So, that a friend of mine that I respect likes movie X doesn’t mean that I will like movie X when I buy a DVD for a birthday gift to my niece. Context.Broadly can’t get a reliable recommendation unless know essentially what for. Sure, with a lot of social context, can know want for, and, then, social can work.But concentrating on the person making the recommendation is not enough and also need to include the relevant interest of the person receiving the recommendation. Indeed, Fred has indicated his frustration finding blogs he will like. Okay, give him the most popular of the 100+ million blogs tracked by Technorati. Right: That won’t work. Think I’ll be happy with a music recommendation from the Top 40? Nope. Lesson: Recommendations from what others like, by itself, won’t work at all well. Instead, need more, especially about the interest of the person to receive the recommendation. And there don’t want to know the person but more directly their interest, and that not in general but particular to the query, time, and whatever else the person has in mind.Ah, such are some of the challenges!
The question is when will google be smart enough to know what your thoughts on are about the music industry vs. just that you are a music fan or that you can play guitar? In other words I knew enough about you to offer the “scum” comment but they wouldn’t know that most likely.I think facebook would be better able to figure this out or twitter. Definitely disqus could if they wanted to analyze comments that people make.Facebook could figure this out by seeing who you associate with and the things that they say and how they say them.
> The questionNo, merely A question, and the one Page, Brin, and Norvig seem to want to address. But it’s a poorly selected question, like saying that building a Mach 10 plane will be sufficient to fly across the Atlantic, that is, is sufficient but not necessary, asks too much, takes a difficult but solvable problem and, for a solution, poses a much more difficult, for now impossible, problem. Not the first time someone did that. Indeed, one of the keys to progress is good problem selection, and there don’t try to solve a difficult problem by first solving a much more difficult problem that, really, don’t need to solve. Sounds like a simple lesson, right? Why make this mistake? Because just imagining that some such AI or some such solution would work is really easy; seeing with much confidence that a much easier solution will work is much more difficult. Or, it’s easy to buy a nice house for $10 million. Harder for $200,000. But getting the $10 million is harder than finding a good house for $200,000. But we can understand how Page/Brin think first about the $10 million solution. Actually they are thinking about the $100+ billion solution, or the priceless solution, until someone has a sufficiently good idea for AI. And Page/Brin don’t know how to work in the world of such ideas. I mean, they wrote some C++ code to computerize a library card catalog subject index with results sorted on a measure of gross popularity. But they did this work at just the right time in just the right place. E.g., in a few minutes, Bechtolsheim writes them a check for $100,000 and soon introduces them to KPCB.In one step more detail, there is the broad idea that looking over the shoulders, at the e-mail and posts, etc. of a person will permit saying a lot about, say, what movie, car, shirt, snack food, blog, recorded music, etc. they will like will be doable in an effective way. I say, mostly no. E.g., for the movie, shirt, or CD, they might be buying that as a gift for their wife, child, niece, nephew, etc. They might be buying opera tickets, their first, with next no hints before.There are ways to proceed, but there are chuckholes in the road, and not all the good paths are obvious.