The Power Of Passed Links

Links are the currency of the web. That's the reason that Google is the king of the web right now. Google generates more click thrus to more websites than any other web service out there.

But I've been ruminating on a comment that was left on my earned media post a few weeks ago by Ben Straley.

One of the issues not addressed directly in your deck is the impact
earned media has on site visits and conversion rates. With respect to
the former, on average one out of five (20%) site visits come from
links shared via earned media. Microsoft and MTV did a study in 2007
(Circuits of Cool) that found that among the thousands of 14-24 yos
they surveyed, 88% of the links they followed were sent to them by
friends. They're not clicking on ads or even using search results as
much as older demos. With respect to the conversion rates, we find a
pretty consistent 2-4x lift in conversion rates for visitors that
arrive on our customers' sites by way of shared links. Razorfish
published a study a couple of months ago that pegged the conversion
lift from shared links at 4x.

We are working our connections to get a copy of that "Circuits of Cool" study but in the meantime, I've been studying refer logs of every website I can get a look at and digging deep on this search links vs passed links distinction. I want to understand if the 2-4x bump in conversion better. I also want to look at this 14-24 year old demo a bit more closely. My kids are in that demo and they start and end their day at Facebook so it is not entirely surprising to me that for that demo, a passed link is more powerful than one found on search.

To make this simple for me, I've been exclusively focusing on four kinds of links;

1) Google – search (organic and paid), driven by intent
2) Email – a passed link sent via email from one friend to another
3) Facebook – a passed link sent from one friend to many friends
4) Twitter – a passed link sent out for anyone who cares to see it

Here's a "STRAWMAN" of what a typical website (typical of the ones I've been looking at) would see from these four sources:

Strawman traffic

I capitalized the word "strawman" because I am not saying this is the typical traffic patterns all websites are seeing. I just wanted to frame the discussion so we can talk about these various links and why each is important.

From my unscientific survey, I don't see the 2-4x bump in passed link conversion. The "value per visit" column in my strawman is the conversion metric I've been focused on. I do see that email passed links convert 2-4x search. I do see that Facebook links do convert better than search but only slightly. And from what I can tell, Twitter links don't convert quite as well as search.

I don't yet have a framework yet to think about Twitter links vs search links, but the fact that email and Facebook links convert better than search makes sense to me. An email link is a direct suggestion from one friend to another. A Facebook link is a suggestion passed from one friend to a group of friends. I get that those links would be more potent than a search link. And I understand why a Facebook is a more potent link than a Twitter link since Facebook is friends following friends, and Twitter is more like blogging where people follow other people who aren't necessarily friends.

But the other important metric to look at is growth rates. That's where the power of earned media is really showing up. I am seeing links from Facebook and Twitter growing rapidly relative to search links in every refer log I look at. Right now, I am seeing Twitter and Facebook combined at roughly 20% of Google in the average refer log. Six months, that number was less than 10%. Presumably that ratio will continue to grow.

Email is also growing as more people are passing links and clicking on links in email. But it doesn't have the scale of the social media platforms because its one to one and does not take advantage of the one to many nature of most social media platforms.

This is really preliminary thinking I am presenting here. It is not based on a fully representative sample. It is anecdotal at best. But this is an important discussion. I'd love to hear from all of you, many of whom are working in search and social media marketing. If you can, I'd love to see more data like this posted on the web. If you can share data, please post it on your blog and leave a link in the comments, or a simple comment with whatever data you can share would be great too.

I suspect there will be at least one and probably several follow-up posts on this topic in the coming weeks.

Reblog this post [with Zemanta]
#VC & Technology#Web/Tech

Comments (Archived):

  1. aweissman

    Keep in mind re twitter – unless the user/referrer comes from Twitter.com you are not seeing all the Twitter client and api apps that drive it. I think that traffic could be as much as 2-5x the twitter.com referrers – we are just starting to unpack this ourselves

    1. fredwilson

      When you unpack your data, can you please share it and let us all know where to find it (via a passed link of course)

      1. toddml

        We’re working with various twitter (and other) clients to outfit their links with referral context. Over time, you should see more applications filter into the referral data on bit.ly info pages.

        1. Leonid S. Knyshov

          The integrated API key is the only reason why I use Twhirl and switched to bit.ly. That’s also what is keeping me from using Seesmic Desktop at this point.I like to see the effect of my tweets and bit.ly makes it easily accessible. The challenge I see is software automatically expanding bit.ly links, such as @troynt’s greasemonkey script, which can skew metrics. I looked at the API but can’t tell if that could possibly skew the metrics. I am pretty sure this script doesn’t use my bit.ly API key as it never asked me for it, so it probably registers a hit when expanding a URL.I’d like to see a full URL in a twitter client when I hover the mouse over the short link. :)I tend to trust links from people I follow, unless Mikeyy possesses them, but it would probably increase the CTR further if I could see where I was about to get sent. I am aware of preview cookies and other crutches, but they are not integrated. That’s the next step.

      2. Adrian Palacios

        Fred, maybe you can include direct traffic in your strawman stats? Like the aweissman notes, you are not seeing all the traffic from Twitter apps–this shows as direct trafficALSO, a lot of email-based passed links will show up as direct traffic… lots of people at work get links passed to them through Microsoft Outlook (i.e., non-browser based programs) and do not pass referral data as well.The downside is that you will also be lumping in people who know the domain name by heart or have bookmarked a site (for instance, etsy.com is really easy to remember).However, maybe you could look at the trend in direct traffic… if Facebook and Twitter have such ridiculous growth rates, maybe the direct traffic will show a high growth rate (the last column in your table above).

        1. fredwilson

          I’ll add that in a follow-up postOn my blog (not the best or most representative website to study), I get 36% of my traffic direct which has been growingI know that some of that is twitter clients and to a lesser extent email

          1. Mark Essel

            Do RSS Feeds count as direct?

          2. fredwilson

            Nope, I get about 10% of my blog’s traffic from google as a referring link and that is almost all google reader and igoogle

          3. Mark Essel

            Danka shern

          4. fredwilson

            bitte

  2. gregorylent

    realized lately i rarely search, and when i do it is several word questions … almost all links i click on in my travels are from twitter, friendfeed, sometimes blogs …

  3. Diego Sana

    This is an interesting topic, i can’t add much here, but your analysis makes a lot of sense IMO.By the way, in the sentence “I want to understand if the 2-4x bump in conversion better” i believe that what you meant to say was “I want to understand the 2-4x bump in conversion better” (remove the “if”).

  4. Sean McColgan

    Link to the “Circuits of Cool” study [PDF] http://bit.ly/RkJhx

    1. fredwilson

      Awesome! Thanks. We’ve been navigating through senior execs at viacom to get this and it turns out its on the web. The power of the passed link!

      1. matt schulte

        Boy, if this doesn’t say it all – “navigating through senior execs at viacom to get this and it turns out its on the web”. very funny and telling.

    2. Adrian Palacios

      Thanks Sean! Question: how do links in chat programs show in Analytics? If it’s Zoho, etc., I’m thinking that would show as referral, but how about other clients (GTalk, Adium, etc)?

      1. aweissman

        some clients show up – others (gtalk, adium etc) do not.

    3. Mark Essel

      Thanks for the eBook link Sean. Love collecting great sources of info/data. Of course Fred kudos for getting this thought line going. It’s vital to know understand conversions, ballpark numbers and where the growth is. It’s the type of things we’d do in Physics back in the day (order of magnitude approximations to all types of far out problems). Once we have that ballpark narrowed down we can determine what to focus our efforts on.

  5. Derek Ball

    Great article and near and dear to my heart. We’ve been focusing on content publishers and helping them to get more links out in the wild with Tynt. We’ve been finding that the email element is an extremely strong but difficult to track way of sharing links as most people cut and paste from the article into an email. When this happens, your analytics software won’t usually show ’email’ as the referrer, but rather it is ‘unknown’. We’re working to refine this problem (shameless plug here: if you are a content publisher/blogger we are looking for Beta testers for Tynt Tracer at http://www.tynt.com) and gain visibility into exactly what you are talking about.Please do a follow up post when you’ve digested more of your learnings!

  6. maxkalehoff

    Fred, you remind me of an academic paper presented two years ago at the International Conference on Weblogs and Social Media, a scientific conference my colleagues at Nielsen BuzzMetrics founded. My then-colleague and research scientist, Navot Akiva pointed out an interesting finding from a paper called “Traffic Characteristics and Communication Patterns in Blogosphere,” authored by professors from the computer science departments at the Federal University of Minas Gerais in Brazil and Boston University in the U.S.:”[O]ur conclusion that the intensity of traffic directed to a blog through search engines (which use traditional page-rank algorithms) does not seem to correlate with the “real” popularity of the blog, suggests that social-network-based navigation may be playing an increasingly important role in web navigation in general, and blogosphere navigation in particular. On that count we note that in blogspace, the popularity of a blog is more a reflection of its owner’s social attributes (e.g., celebrity status, reputation, and public image) than a reflection of the number and rank of other blogs or web pages that point to it. This highlights the need for the development of page-rank algorithms that take into consideration the social attributes of blogosphere actors (as opposed to solely on the topology of the underlying blogspace), possibly using inference techniques.”The analysis was focused on blogs, but infers broader implications of social media in Web navigation.You can download that paper here: http://www.icwsm.org/papers

    1. fredwilson

      Hi maxAny idea how I can get my hands on that paper?

      1. maxkalehoff

        I included the link in my comment.

    2. Mark Essel

      Great link Max, this post is a gold mine for info.

  7. Prokofy

    Do you have an automatic way to count these referers on a blog? Just counting them manually for the last few hours on my blog, I saw this:Google Search: 127Google Reader: 10 Other blogs: 34Blogsearch: 4Twitter: 2Friendfeed: 2MSM sites: 7News aggregators: 9Yahoo: 3MSN: 3ask.com: 3search.live.com: 3mobile: 1backtype: 1I, too, had this impression that there are a lot more Twitter and FF referrals, and more mobile phone referrals, but in the end, it really is Google and other search engines that produce the most links, and then other blogs, and then the other stuff.Links are the currency of the web, but unlike paid ads, they do not pay out, except for those who sell ads against the attention near them, like Google, and to a lesser extent, for those who place ads if they get sales.There are two kinds of economies on the Internet, the open marketplace, which runs by search without having to form a relationship, and the closed bazaar, which runs by having to form a relationship, i.e. word-of-mouth, friendship link passing, etc. You need both. The latter has higher social costs to form and keep.

    1. fredwilson

      I get my data from google analytics

      1. Prokofy

        Oh, ok. Thanks for giving me an excuse to try to finally wrangle this and set it up on my site, it was elusive to me for a time. I finally got it working (I am not a tekkie) and now I’m baffled by the jargonistic terms like “bounce rate” and such but eventually after there is a 30 day sample or something I will email the report to you.It’s puzzling to me that this analytics seems to be saying something different than the manual look-see, i.e. 215 visits in 4 hours is like this: * 44.65% Direct Traffic * 30.23% Referring Sites * 25.12% Search EnginesIn fact, the actual list shows so many google searches, but perhaps direct traffic contains that too. I’ll have to study up on this topic in the “university” .

        1. Mark Essel

          Oh boy, I hit report instead of reply… (too much wandering out in the sun today). Hope that doesn’t cause any problems.Wanted to thank you for getting me to setup Google analytics (you have a similar typepad blog Prokofy).

  8. Costa Rica

    Regarding “And from what I can tell, Twitter links don’t convert quite as well as search.” Is there a difference if the conversion page recognizes that the source is Twitter (or Facebook) and does not ask you to ‘register’ but instead asks you to identify yourself through your Twitter account?I would assume that Twitter users are particularly attracted by shortness and simplicity, so I would presume that registering by reusing Twitter’s login/password (like TwitPic) or a very short form (like Tumblr) would show a difference in conversion rate.R

  9. Mike Simonsen

    Here’s some quick anecdotal observations, Fred. I’ll dig out our data if I have the chance.I’d guess we easily have 2-4x better conversion on shared links than search. Importantly, I’d categorize searches for “altosresearch.com” or “Altos Research” as a “shared link”. That is, someone said “Go find Altos Research, they do what you need”. All are forms of classic word-of-mouth marketing. If I don’t include the url-searches, then the shared link traffic is much harder to characterize (sample size is much smaller).Finally, I’m sure the advantage is related to the type of product and type of conversion. In my case I’m talking purchase conversions as opposed to free registrations. Our product is a vertical-targeted premium subscription, which is a very different sales beast from an ad-supported somethingorother, or you know, shoes.looking forward to a fuller set of data. Will gladly contribute to get a solid corpus.

  10. Ian McAllister

    I’d also be interested to know how they are doing their calculation. Are they measuring conversion per session, or are they adding up all converting sessions from a single shared link? If the latter, then the increased conversion rate might simply be a function of the wider distribution (resulting in more inbound sessions) you get when links are shared on Facebook or Twitter, which would make sense.If sharing is a numerator and denominator feature, then that would make for a strong case for plastering share links everywhere.

  11. Eray Basar

    I was talking to friend about the exact same topic just a few days ago. I told him that my link clicking habits changed, and at the same time the way I crawl the web. Since using twitter and facebook, more and more passed links are comming to me, which is good – but also bad in a way that I consider it as ‘information overflow’. I surely use google still a lot, but probably I’m searching less for generic keywords. I am not sure about the weighting in your survey though. I consider email link sending pretty much as dead. People don’t send me links via mail anymore. Facebook totaly replaced this option. For professionals, twitter passed links might be even more valuable than Facebook links. The unidirectional character of twitter allows me to follow people that are “relevant” to me, this relevance is therefore passed over to the links these people post. Then again, when old school friends on facebook pass links, these will probably have less relevancy for me.

  12. AndyBeard

    What is your definition of a conversion? On your blog it might be a comment, on Digg it might be a vote… or a new user signup, or just an advertising click. For e-commerce it might be a newsletter signup or a sale.You need to differentiate first time visitors from repeat visitors, and not just daily unique cookies. E.g. an email from a friend suggesting you buy something from Amazon, where you have already been shopping for 10 years isn’t necessarily the same as someone referring a totally new custom. Also note Amazon recently changed the rules for affiliates, and they can no longer directly promote Amazon products in Google paid search.I was recently promoting an affiliate product where in previous promotions I have achieved a $12 visitor value – that is an exception rather than a rule.I know landing pages that convert a visitor into a free subscriber for a teleseminar at 80%+, but it is very hard to calculate the real average lifetime value of such a lead other than based upon historical figures per traffic source.. which should be broken down not just by referrer, but also by keyword.You need to filter by user intent. `

    1. fredwilson

      I agree with all of thisMost of this data comes from e-commerce so it’s conversion to a sale but I’ve also included some content sites where conversion is much less valuableThis was not meant to be a sophisticated analysis in itself, but I did want to facilitate a discussion around this topic

  13. jonathanmendez

    I stopped reading the “circuits of cool” study when it said the data was derived from focus groups and surveys.I think your data would be more interesting if you focused on first time visitors and/or removed brand queries (navigational/recovery goals) from the data. In comparing links I would stay focused on discovery.Great stuff!

  14. Mark Zohar

    Fred,I think it’s quite clear that the combined “intensity” (ie., quality + emotional quotient) and “virality” of shared links is an order of magnitude greater than direct links. I think it’s also fair to say, as you suggest, that the current rate of growth of shared links is an order of magnitude greater than direct links. That being said, I think there are some finer details that need to be considered:1) Not all shared links are equal in value. Shared links that originate from social apps such as Twitter and Facebook probably have a slightly higher intensity and a much higher virality than shared links that originate from email. This may lead web merchants to offer customized or tailored experiences depending upon the source of the shared link; ie., social app traffic = more community focused experience; email traffic = more content driven experience.2) It’s not all about conversions. Sure, every web property wants to convert leads to purchase. But shared links can build brand value in addition to conversions. This can work both ways of course as the recent Domino’s example illustrates.3) Search is still the big dawg. We have to be careful not to live in the echo chamber. With all the Oprah hype, Twitter is still an extremely small source of traffic and leads for web retailers and other sites. The same generally holds true for Facebook. General web search remains the main source of discovery, intent and lead generation on the Web. As more Web interactions move to mobile devices, search is likely to increase in utility and value especially with advances in voice enabled search such as with Google Mobile and Vlingo.Finally, as you posted recently, the rise of passed or shared links may also have a significant impact on the the dynamics and economics of the affiliate network market. Embedding affiliate IDs within shared links is a big opportunity.

    1. fredwilson

      Mark, this is all very important and I agree that we should not overhype twitter and facebookBut some of the data I’ve been looking at is e-commerce data and I’ve seen twitter and facebook pretty high in the refer logs

    2. Prokofy

      Someone should analyze the type of content of the shared links. I think an enormous number of them are shared links about news and ideas and discussions — about intellectual and social life — not directly to some merchants’ site that translates to a direct sale. Even if you factor in the witless clicks on soc/media gurus with books on twitter tips. So then conversion is about selling the ads against the social conversation about events and personalities.

  15. Scott

    What’s scary is that google has all this data in their google analytics backend on thousands of sites. Any idea how much penetration GA has? I imagine it’s very prevalent since I see the google analytics URL on my browser’s status bar all the time when loading a page.Either way, they must be fully aware of referrals from facebook/twitter (passed links) vs. their very own search links. I bet that they’re aggregating and anonymously mining this data to do the same type of strawman. And I wouldn’t be surprised if they anonymously keep track of outgoing clicks in gmail. I probably click on 100 passed links in a gmail e-mail for every 1 sponsored ad.I wonder if Quantcast has this ability on quantified sites as well…

  16. Bahiyyih Maroon

    I primarily utilize numbers to augment ethnographic research so I found both of the passed links to be very interesting, but in distinctly different ways. The blogosphere research (and people’s subsequent comments) suggest rightfully the very urgent need for algorithms that better account for the radical shifts in topology happening in the digiverse. The notion that the blogosphere represents a vastly different territory that than “classic web” in much the same way that two geographically discrete nations are distinct social entities is tremendously important to keep following up on in terms of defining new measurement tools and methodologies. The Circuits of Cool, spoke to me as an interactive anthropologist about the ongoing need for clearer examples and concrete explanations of the very real and important differences in use patterns between Geos. While I’m sure the full length report delivers on the promise to track device usage difference across geos, that doesn’t come out in the synopsis and it’s probably one of the central issues we need to recognize in any effort to engage global markets. In Tunisia for example, “sharing” does happen online, but it’s more popular to share in person. Youths with bluetooth and web enabled mobiles (their term for cells) gather at cafes and hookahs and swap links and static content as part of their social face to face experience. This is important for entertainment clips, because often items fetched from the web are transferred to desktops through various means and then reloaded onto mobiles such that the swap has no way of being tracked. In other words while every indicator suggests user data collection can be safely gathered entirely via virtual tracking, in many geos from the Middle East to India and South America actual user practices are indicating the need for mashups of old fashioned user survey work/research w/ virtual tracking apps. Thankx much for the both links! Very thought provoking info.

    1. karen_e

      Kiefek Bahiyyih, ahlan … wow, incredible comment about the mashup of survey work needed. I wonder when danah boyd @zephoria will enter the discussion.

  17. BmoreWire

    your basically describing P&G tremors.

    1. fredwilson

      Oh well

  18. lavraamides

    email link referrals can be one to many if you want to send a link to a group of people who are not on a single or any social networking platform. I do that often with my family. I’m on facebook; they aren’t. So, keep that in mind when evaluating email.

    1. fredwilson

      Excellent point

  19. Jason

    With regard to the Twitter passed link – i think this topic should be brought up again a year or two from now. With Twitter finally going mainstream, people starting to really understand the purpose of the technology (most people still don’t get it), and tools to analyze the many metrics of Twitter becoming available that sort out the influence of each Twitter user (note the influence of a source like @mashable or @fredwilson vs an unknown user) we are bound see different values for a Twitter passed link.

    1. fredwilson

      Yes, and we need the tools to measure that

  20. Edwin Khodabakchian

    Fred: are you focusing on content site or ecommerce sites as the landing destination of the link? I am trying to understand what is conversion and value of visit in your model

    1. Adrian Palacios

      Edwin, good question… something I wanted to ask.

    2. fredwilson

      Both e-commerce and content and I know that’s an issue to co-mingle the twoI want to be clear that my survey was very unscientific and meant only to suggest that someone, or all of us, should be drilling down on this and figuring it out

      1. Edwin Khodabakchian

        Thanks for the clarification. It would definitely be very valuable to drill down. Beyond the value of shared links what is your perspective on how the sharing infrastructure could get rewarded?

        1. David Weiner, PR Newswire

          Great conversation-starter, Fred. I started writing a comment a few hours ago and felt it was way too long, so I turned it into a blog post.You hit on some critically important notes for people in the communications industry. How much is a view worth online? Comparing this to print is critical as it will determine the future of newspapers (if they can charge more for their online content, their offline content will survive as a result).Should the value be determined by the content or the content creator? What about time spent? What about referral type? How much less if it’s not a unique visitor? I don’t think the “influence” of the referrer is as important as some are saying. Whomever referred you is still trusted to you. The more influential, the more links that person will direct.Anyhow, as always, great post.DW

        2. fredwilson

          By sharing infrastructure, do you mean bit.ly or do you mean facebook and twitter?

          1. Edwin Khodabakchian

            I meant facebook and twitter. Also in the case where I subscribe to your blog in twitter or I am a fan of TED in facebook and I get a link directly from the source in my twitter client and click on it, is that a “shared link” too?

          2. fredwilson

            Yes, I think so

  21. ravisohal

    It seems to make sense that passed links will be increasingly trusted currency. Search is like looking through the yellow pages and passed links are word of mouth recommendations. However, I wonder if passed links will always be inherently less convertible – “I know what I’m searching for and will click on something that meets my criteria” (search) vs “Hey, you might like this” (passed link). Or maybe the issue is that they are less measurable because content distribution channels are beginning to look more like social gatherings.

  22. johndodds

    I’m probably being obtuse but what do you mean by value per visit and how are you calulating it?

    1. fredwilson

      For e-commerce its the total dollars of all the visits divided by the visitFor content its the ecpm of the page view

  23. Jonathan Karon

    Mark’s comment touches on this and I think you need a lateral differentiation.I’m going to propose another class of traffic: Intent-Driven Passed Links. Simply put, I ask people on Twitter rather than asking Google.Consider googling “Django hosting provider” vs. tweeting “Anyone have a favorite Django hosting provider?” My personal pattern would be to do the former, review the options, then seek critique via Twitter.More and more I’m skipping the Google part. So it’s passed link, but it’s a solicited passed link and thus is intent-driven.The popular discussion of Twitter still frames it as a micro-blogging service but there are a lot of people using it like a chat room.

    1. fredwilson

      I had not even thought of thisIntent driven passed links probably convert at a very high ratio

      1. Jonathan Karon

        And thus the vigorous discussion between Magpie and it’s detractors.

  24. LukeG

    Other than brief flashes of personality and entertainment, the utility I derive from twitter is entirely in the links. If anyone’s so inclined, I’d love to see a twitter app that aggregates the links that are shared by my twitter network (spec the folks I follow), strips them of URL shorteners and shows a relative weighting. Think it would be tremendously userful.The implicit recommendations passed through twitter are made especially useful – or at least more interesting – by the fact that follower:following ratios imply dynamic levels of authority. We’ve seen strong but anecdotal evidence that not only do links shared by more “authoritative” twitter users reach a wider audience (a given), but they’re passed along/retweeted at a higher rate, as well. Anyone have more significant data on this?

    1. Phillip Baker

      I’d love to see that app too. Twitter is great for sharing news real-time or otherwise and I think a page of news links from followed people would be interesting.Dave Winer has built something along these lines that collects his own most-clicked links here (he’s also built pages for a few other people): http://twitter.scripting.co…The relationship between people passing links is really interesting. Followers might be actual friends, often not, especially on Twitter which is becoming more of a broadcast medium. Even ‘friends’ on Facebook might not be real friends, it depends how people use it.The relationships between people, the intention in sharing the link and the type of link/content (e.g. entertainment/time-waster, news etc) are all factors underlying the aggregate click-through rates for each category of passed link and whatever is being measured as a conversion.

  25. Larry M

    One has to consider whether the link was solicited or not and if the link was seen or not.I haven’t been on twitter all day. I’ve missed a lot of links that were posted by those I follow. In fact, of my last 800 tweets received, 333 contained links with 10 dups. I’m not likely to go back thru and see what I’ve missed.Facebook has a slower velocity of links, so likelihood of being seen is higher.Google – I’m searching for something, so it would expected click thru rate is highI’m curious to find out click thru rate from twitter search. That should compare much closer to Google. I don’t recall seeing twitter search as a referral in Google Analytics.

    1. fredwilson

      Wow, that’s over 40% of tweets with linksThat is huge

  26. lawrence coburn

    I don’t know about conversion, but here are some pages per visit stats from our community:Google Organic – x pages per visitTwitter – 1.72x pages per visitFacebook – 1.96x pages per visitIf pages per visit is a proxy for engagement, It would seem that friends have a better idea of what people might find engaging than Google does.

    1. fredwilson

      Do you know how many pages per visit you get from a google visit?

  27. Steven Kane

    I suspect the key difference between twitter links and email or facebook is the depth/nature of the personal relationship involvedtwitter links are received by “followers” — at best real personal relationships but mostly either strangers or acquaintances or fans or whateverfred, been reading AVC for a long while, this is one of your best and most interesting posts ever. and thats saying something

    1. Mark Essel

      I’m pretty new to following A VC but in the past week I’ve become hooked. Powerful look on the inside what makes internet businesses work.

    2. Yule Heibel

      I was just reading a blog post by a friend in Vancouver, Boris Mann, about tagging local places with rich content (see his post here: http://tinyurl.com/dcdxta), when I noticed something in his sidebar under “what Boris is reading.” It was a link to an article about local food production, which I had tweeted (and I’m sure others in the Victoria/ Vancouver area had as well) – but it was provided via “Timeline for bmann – MicroPlaza.”Hm, MicroPlaza? What’s that? Being nosy – but also being trusting of whatever Boris is up to – I probed into the link and found MicroPlaza’s faq page: “We wanted a way to discover relevant and interesting items from the people we follow on Twitter. Most of the time something interesting is a link shared by a friend or colleague. So we built MicroPlaza to deliver us the filtered links from our Twitter timelines. It’s our discovery engine, our personal newswire and just so god damn addictive!” (see http://microplaza.com/faq)It’s the first I’ve seen of it, but the service seemed relevant to the discussion here (and of course I want an invite code from them now! = hooked). People are figuring out how to build businesses on providing trusted links and recommendations. Not sure what it says about *commodifying* the “depth or nature of the personal relationship involved,” but it’s happening.

    3. fredwilson

      Wow, thanks for saying that steve. I am trying to post about interesting things I am seeing and paying attention to

  28. David Henderson

    “They’re (the 14-24 year old demo) not clicking on ads or even using search results as much as older demos”Are”links” the currency of the web for Gen Yers? One could argue that friends are their currency, their channel and their content.

  29. underdown

    I am dumbfounded by this article. First of all emailed links and social media links convert better than other passed links? I think the preponderance of evidence paints that assertion as completely ridiculous. Search converts – in terms of a traditional e-commerce conversion – FAR higher than any social media link – the difference is exponential. Email links do convert well on many of my sites but try building a business around them. I question what you consider a “conversion” to draw these conclusions.Links are valuable because of Google’s (and others) algorithm that values third party links’ anchor text as a metric for keyword rankings. As a search engine marketer I couldn’t give a shit less about social media traffic – other than the backlinks they create.

    1. fredwilson

      Can you share some data with us? We are trying to understand this phenomenon. Throwing around words like ridiculous and dumfounded is not as helpful as data.

      1. underdown

        Sorry fred, big fan of your site btw. Of course I have investors and partners on my sites so sharing data is not within my rights. So that leaves two possibilities (false dichotomy?) I’m either a troll or someone with access to data that found it shocking to see this mentioned on a site I frequent. I’ll leave it to you to judge. GL

  30. tzvika

    Fred, you’re not taking into account the no 1 link-sharing tool for that demo – IM. You see it as “direct traffic” on your analytics screen, and my guess is that it work as well as email.

    1. fredwilson

      Can we track IM links?

  31. NICCAI

    Are you missing value by not considering other trusted sources as passed links? What about your blog? Most links I “follow” come from trusted sources via twitter, email, rss, or the respective site associated with my trusted source. Also, how do you validate the passed links from Twitter? Do you consider all url shortening services and assume it is twitter? Just wondering how you factor in the myriad of clients and sites using their api.

    1. fredwilson

      I am relying on google analytics for this data

  32. Gerald Buckley

    Fred, happy to make four sites available to you via Google Analytics. Three of them track at roughly 1M uniques/year. One of them is ecommerce, one of them is VERY vertical search/content and the other is raw content. The last one is a startup getting solid earned media.Your front desk took my number on another matter this afternoon. One call gets it all 🙂

    1. fredwilson

      Thanks Gerald. I’ll call you on Monday. I appreciate you sharing the data with me.

  33. Morgan

    Greetings,I figure this is mostly a call for datapoints; I don’t know if this is helpful, but here’s mine.My ‘conversion’ is a download of my free application; pretty trivial, but it works for me. I normally get paid in egoboo, but this year I was unemployed and my (wonderful!) users ended up paying my mortgage payment. Twice. Since my scenario is very different, the behavior of my users may also be very different.Organic search nets me a 39.3% ‘conversion’ rate, direct traffic nets 34.33%, and referral (the link posted on someone’s page) is 30.23%. (Google Analytics | Traffic Sources | All Traffic Sources, Goal Conversion tab, show: Medium). Twitter.com shows up under referrals, and has a sub-10% conversion rate, Facebook shows up under referrals and has a sub-10% conversion rate. Traffic from both are so low as to be statistically meaningless though. That may be because my software doesn’t appeal in a way that it gets shared on those sites often.I don’t have a presence on Facebook. I have a presence on Twitter, but I don’t promote via twitter. Other folks do promote it, though; I see my application show up on search relatively regularly. My site has a PR (according to online tools) of 5.I hope that info helps!Purely as an opinion, I’d argue it’s a heck of a lot better to encourage folks to blog about your software (or site?) with a positive link over passing the link to a subset of friends. It’s a matter of permanence and audience. The link tells the search engines you’re a useful resource. Even if nobody but the search engine reads that link, it informs everyone who searches on that engine in the future. The incremental value of that over time is much better than a temporary personal referral that is quickly forgotten. This is worse under Facebook and email, which act as walled gardens, and so passed links are lost to the search engines.Now I don’t make a mainstream (read: very popular) app, or run a mainstream site, so I don’t know how well this scales to folks who are talked about all the time. It’s just opinion. It’s also not an ‘or’ world; both types of links coexist, and folks who blog about something are also likely to recommend it person-to-person.– Morgan Schweers, CyberFOX!

    1. fredwilson

      Thanks Morgan. This is useful. But I don’t understand how blogging and twitter is different. I see them as essentially the same activity.

      1. Phillip Baker

        Wow. I think that statement is another 100+ comment post by itself 🙂

        1. fredwilson

          I’ve always seen twittering and blogging as essentially the same thing.It’s just that twittering is easier because the message lengths are short.I took to twittering because it was the exact same experience for me as blogging

          1. Phillip Baker

            They’re both publishing tools and I guess that ultimately it depends how the tools are used. But I think the character limit, the follow system and the centralization of Twitter do produce something different. It’s more about people than posts which are more disposable, I see it as more communication than publishing. I’ve written about it here: http://bit.ly/16Obmu, and in a few other posts.In terms of passing links, the sole reason for a tweet could be to post a (single) link and I think that in itself creates something different. Whatever the other characters are used for, the message in those tweets is “look at this link”. Blog posts can be just as explicit but links tend to be used more as references.I’d be very curious to know if conversion rates for Twitter links and blog links differ.

          2. Seenator

            Fred,I am a bit late to this post but to me there is a big difference between Twitter and Blogging and that is COMMUNITY i.e. a feeling of interconnectedness.The community comes from three main parts:Follow@ Replies# Hash TagsIf and when Ev had invented Blogger and he had built an identity that could follow each other, friend each other etc, then perhaps it would come closer to Twitter. Its the same issue with WordPress.Although LiveJournal and Facebook’s Notes features come closer to Twitter than a generic Blogger and WordPress platform.I am sure there is one more way to look at this but thats how I am trying to wrap my head around this.

          3. fredwilson

            Blogger plus indentity plus follow plus engagement is called tumblr

          4. Seenator

            Then you may have a conflict (Twitter/Tumblr) in your portfolio 🙂

          5. fredwilson

            We don’t think so and have discussed it at lengthThe investors in tumblr (USV and Spark) are also investors in twitter

    2. Phillip Baker

      Morgan, I think it depends how closely your friends’ interests/needs match what you application does. If there is no alignment, it makes sense that organic Google traffic would show a higher conversion % because it allows people you don’t know to match themselves to your app through search.For example, and I’m trying to use your story as a generic example here because I don’t know your app or your friends, obviously. Maybe your Facebook friends are interested in your app because you made it and they are your friends. They might click on the link on your profile to see what you did but they have no need for your app so they don’t download it.There are a lot of human variables involved in passing links that it make it hard to determine the extent that conversion rates are down to the medium alone.

  34. Nova Spivack

    Very interesting line of thought, Fred. Please keep us all updated on what you conclude from all this as you get more data.

  35. Phanio

    Doesn’t it all come down to trust – trust in the person or source of the link? Since late 2000 – Google has really gained the trust of browsers (users). I have a sense that most of the Google results are relevant to what I am searching for – thus I click their links. Many times I receive links in email or in other apps and really don’t trust those from whom they came. If I solicit information from my friends and colleagues, I will trust their responses (thus their links) much greater.I just think that the channel does not matter as much as the source. Imai reported 800 tweets and 333 with links. But, also stated would not go back thru to see what was missed. Thus, does not trust the sources from which they come (or at least does not hold them in very high esteem).This is very similar to the VC industry. If a new entrepreneur goes directly at a VC – the conversion is almost nil. If the prospect comes from a trusted source – the conversion is much, much higher.Thus, with links – do you trust Google or do you trust the users of twitter or facebook etc. I tend to like Google better as there are more choices to compare instead of one link from a friend or colleague.

    1. fredwilson

      Good point on trust. I think most people trust their friends and “trusted sources” more than google

  36. Steen Seo Öhman

    Interesting view will certainly follow the next posts.I think Twitter is very interesting in this context, as it seems links are often posted and often followed, at least compared to facebook.

  37. Swaroop C H

    Another aspect of Twitter is that people tend to ask questions like “What’s a good restaurant in ___ area?” and people tend to reply, often with links.This could happen over email as well if you know a specific person who has that knowledge.But I’m far likelier to get a response on Twitter.

  38. RacerRick

    It would be nice to do a search.twitter of just my friends, and just my friends posts that have links.

  39. Joe Lazarus

    I manage online marketing for a number of businesses across a bunch of different verticals. While I can’t share specific data from my clients, I’m seeing many of these same trends. Here are a few additional thoughts / comments…Google is the top traffic driver for most sites. As with your data, my clients see the largest % of traffic from search, but it’s often a much larger % than your data suggests relative to earned media.Facebook & Twitter are growing quickly as a traffic source. That’s generally true across the board, but the % of traffic from those sites is far lower for most of the sites I’ve studied.The % of traffic from earned media varies wildly by site. I imagine you’re seeing a larger % from earned media due to the nature of the companies in your portfolio, which tend to be Web 2.0 businesses with early adopters who are more likely to use social media sites than more “traditional” online businesses (ex. Etsy probably sees a much higher % of traffic from earned media than eBay, but I don’t have data on either of those companies). I also notice that earned media is a bigger traffic driver for sites that are more dependent on timely, urgent content (ie. people are more likely to share & click links through earned media for an Etsy item that could be out of stock tomorrow or for a breaking news story vs. a television purchase or vacation guide).Email typically includes high value retention segments. Your definition of email… “a passed link sent via email from one friend to another”… is often only a small portion of email referral traffic. Outbound email marketing (ex. Etsy newsletters sent from the company, not our friends) can account for a large portion of that traffic and it tends to convert really well since those are loyal customers who subscribed to company emails.Facebook is cannibalizing online email, particularly for the youth segment. I’ve seen a number of studies that suggest that young people use Facebook not just to send links to groups of friends, but also as a replacement to email (ie. they send a link to just one friend using Facebook email). Many of the people under 25 that I talk to tell me that they only use email to communicate with companies and old people… otherwise it’s Facebook email, IM, etc. Twitter direct messages are also an email replacement. My hunch is that 1:1 communication through social sites converts much better than group communication blasts.Navigational queries skew Google conversion. Typically, the most popular search terms for any business AND the ones that convert best are navigational searches on the company’s brand name (ex. loyal customers search Google for “etsy” as a way to navigate to that site). Often, if you separate navigational queries from other search terms, the volume and conversion metrics tell a different story.As always, thanks for sharing the data and your analysis. Maybe your friends at Comscore can provide a more complete view across a broader range of sites.

    1. fredwilson

      This is great feedback JoeI think you are right that my data is skewed in favor of earned media but it may just be a matter of time before you start seeing these trends show up in more traditional sectors

      1. Joe Lazarus

        Agreed. Mainstream consumers are becoming more comfortable with the idea of sharing links with friends & followers every day. It’s also becoming increasingly difficult for companies in more traditional sectors to grow traffic unless they offer something worth sharing. The days of paid traffic plus SEO / email spam are dwindling.

  40. Ben Straley

    Hi Fred. This is a really interesting conversation you’ve started. I wish I could have jumped in sooner.Following are a few more data points and comments I want to add to the discussion:1.Recap of a Hitwise study published on DMNews – Visits from Facebook are greater than visits from Google on a few major sites across several verticals: http://www.dmnews.com/Faceb… Are sharing and “social discovery” more important sources of traffic than search in certain verticals? Is this a harbinger of things to come across a larger swath of Web sites?2.The average % of traffic from passed links varies considerably by vertical. We suspect some of this is due to the demographics of the audience on a given site. Also, the content found in some verticals is just much more likely to get passed along than in others. For example, we see % of traffic from shared links on games sites well into the 25-40% range. Sites/campaigns that promote strong offers also receive a significant % of their traffic from shared links – in the 20-30% range. In contrast, B2B sites with more static content receive a far lesser benefit from shared links – somewhere in the 3-7% range.3.This is closely related to point 2 above. Content matters most of all. If you’ve got good content and make it easy for people to link to it, it will be shared and attract a lot of visitors. We’ve seen this work over and over. We tracked the pass-along of links pointing to two campaigns running concurrently for the same product (different micro-sites). One of them had a good offer but so-so content while the other campaign had great (funny) content with no offer. The % of unique visitors generated by the pass-along of links to the good offer was under 10% while the traffic from the pass-along of the links to the good content was over 40%. The campaign with good content also got significantly more traffic overall. What data like this suggests is that the prediction you make in your deck about dollars shifting from media to content is a really good one in my opinion. As marketers compete for the attention and interest of their audience, the best way to do this is through content that’s delivered to them via their social graph. This already happens if the content’s good. There just isn’t enough of it.4.We haven’t looked at this yet but I’m extremely curious to know what ,if any, correlation there is between the number (or %) of visitors from passed links and the number (or %) of visitors from organic search 30-60 days later or however long it takes for the search bots to update their indices. It stands to reason there’s a positive correlation between the two metrics but we haven’t done the number crunching yet. Could traffic from shared links be an early indicator of improved SEO performance given the proliferation of back links?5.The strawman you’ve built above is a good stab at an analytical framework for estimating the absolute value of shared links. I think it makes sense to take it a few steps further. If I’m an online marketer paying an eCPM/eCPC/eCPA for some % of traffic to my site and conversions (if I’m doing direct response), then I’m probably also encouraging people to pass-along links via Twitter, Facebook, Digg, etc. While the reach of the latter set of activities is no doubt lower than what I can reasonably expect to obtain from a paid media campaign, I should also expect that the eCPC/eCPA for shared links is much, much lower due to consistently higher click-through and conversion rates for these links. So, as I think about how to optimize the performance of my campaign, I need to be thinking about how I should be shifting dollars within as well as across these very different but complementary sets of activities. Also, my paid media can stimulate pass-along which should be factored into my calculations of the true ROI of my advertising campaign. We’re still pulling the data on this which I’ll share as soon as we have it.6.In most cases where traffic and conversions from shared links are high, we see a HUGE amount of activity driven initially by vertically-oriented blogs and community forums before it goes “mass social” by making its way up to Facebook for example. In calculating true value of shared links, it’s important to factor in the niche sites and communities in your vertical. The numbers might be relatively small but their significance huge. In one case, we saw a niche site send several hundred visitors to a marketer’s campaign micro-site in a 2-week period. This placed that referring site well down the list of top-referrers (something like #25). Not very interesting. But using our sharing tracking and measurement technology, our customer watched as the visitors from that niche site drove well over fourteen thousand additional visitors to the campaign micro-site via links shared in email, IM, on blogs, forums, and, of course, Facebook etc. Niche sites matter.7.Meteor Solutions hasn’t been in business long enough to have longitudinal data to do trend analysis yet. However, we do see sites/campaigns that cater to a younger audience receive a higher percentage of their traffic from shared links. (See my comment about games sites above) When I take a step back and look at my own behavior, I also have a hard time denying the fact that my media consumption habits and behaviors have changed in the last 18-24 months. I’m getting more and more of my information from the people I’m connected to through email, IM, RSS, Facebook, and Twitter. Also, the nature of the searching I’m doing now is much more targeted and specific. I won’t search as much for content or something that’s happening now because I’ve probably already received the link from someone I know or follow. The links that are relevant to me and timely find their way to me these days with remarkable efficiency.8.The most popular mode of sharing we see is email (25% of visits from passed links come from links shared through email), followed by blogs (18% of visits from passed links come from links shared through blogs), video sharing sites (14% of visits from passed links come from links shared through video sharing sites like YouTube), and forums/message boards (11% of visits from passed links come from links shared through forums and message boards). Social networks account for around 9% of the traffic from shared links. I pulled these stats from our Meteor Tracker data which does not yet contain a representative sample of sites of varying sizes across all verticals. It isn’t yet representative of the Web as a whole.

    1. fredwilson

      Wow, wow, wowThis is a fucking great comment. Its a blog post in its own right. Can I reblog it in its entirety?

    2. fredwilson

      BenI posted this in its entirety just nowIts just too good to stay behind the comment linkI linked out to your linkedin profileHappy to replace that link with something else if you’d likeThanks again

  41. Mike Evans

    I”m seeing more and more evidence that this has a lot to do with segmentation. On our site, we’ve got channelized survey responses that say it isn’t Email or Facebook sourced transactions, but Mass Transit advertising transactions that yield the best results. A deeper dive into our survey data shows that it has most to do with the demographics of the group seeing those ads vs other channels.Linking the behavior of the demo to facebook is the right way to go about creating actionable items from this data.

    1. fredwilson

      Thanks! This is super helpful

  42. Doug Schumacher

    Fred,Really like this entire thread.Creativity will certainly become a premium, imo. I recently spoke at the Facebook Global Sales Conference on the increasing demand for creativity in social media. But social media is much more challenging that paid media. Even bad paid media drives traffic and some sales, whereas bad social media is problematic.Social media-oriented marketing is attempting to get a more committed involvement from the consumer, and so it has to reach them at a deeper level. And that requires a much deeper understanding of their wants, needs, desires, technologies, etc, than getting them to laugh at a :30 spot. So while creativity should be getting a good bump in the coming years, I think it’s going to be closely tied to strategic planning, and it’s ability to get the whole process headed in the right direction.http://www.dougschumacher.c

    1. fredwilson

      I agree. Good products and services will do well with social media. Bad ones can try as hard as they can and won’t get far

  43. Brendan Piper

    great read, always like seeing someone brining metrics into the fuzzy science of “Social Media”

  44. Mayank Sharma

    Fred, I have linked this article to a blog post of mine which talks about the accuracy of the statistics provided by link shortening services such as bit.ly. Here is the link. http://zebugroup.com/blog/2009/06/link-tracking-lies-damn-lies-statistics/. I have used this blog post of yours as a reason for a shift from google sending traffic to your site to social media sending the traffic.

    1. fredwilson

      Thanks. I’ll check it out

  45. Kevin C.

    I am trying to determine the value of a twitter link and the multiplication effect of many twitter followers? I don’t see a high conversion rate right now, but it is free and a typical twitter account grows far faster than facebook. With the new apps, you can post on all three with a simple tweet.

  46. fredwilson

    Yup, a few people made that comment and it is right