About Tom Evslin

Video Profile of Tom Evslin

Follow Tom Evslin on Twitter


Add to Technorati Favorites!
Powered by TypePad
Member since 01/2005


Mass Customization Example - FeedBlitz

FeedBlitz (a company I’ve invested in and am a board member of) enables mass customization (see yesterday’s introductory post) for both publishers and subscribers of online content.  It may well be that no two people want to read the same list of online publications; even those who read the same publications may not want to receive them the same way. Moreover, even though you and I both subscribe to the same publication, we may both be interested in different articles from it.

FeedBlitz started by solving a single problem for publishers: delivering blog posts by email to readers who preferred receiving posts in their inbox to visiting blogsites each day or using a feedreader to browse RSS feeds (which many readers still don’t understand). For the publisher, it handled soliciting and storing the reader’s email address, accepting changes and cancellations, keeping track of various stats, physically converting an RSS feed into as many mail messages as needed, and helping assure that this mail – which the reader had requested to receive – actually got through the reader’s spam filter.

Subscribers keep control which is actually to the advantage of reputable publishers who want to be read. Verification assures that people are actually voluntarily signing up to receive this email and not being imposed on by someone signing up on their behalf. Subscribers can cancel with FeedBlitz at any time and know that the publication WILL stop arriving in their inbox; subscribers can use FeedBlitz’ interface to change their email address once for all the publications they receive through the service.

That was the service pretty much as it launched way back in 2005. But this is the age of mass customization; you could tell by the requests for enhancement that FeedBlitz got both from publishers and subscribers.

Although pundits talk about “convergence”, what’s really happening in communication is divergence. When FeedBlitz began, its supported only subscription by email. Now a reader can choose to subscribe to the twitter, the text message, or the instant message version of a publication. Note that the publisher doesn’t create each of these versions by hand – FeedBlitz does this customization from the RSS feed produced by the publisher at the instruction of the reader. My guess is that FeedBlitz will end up supporting an increasing number of delivery options on behalf of publishers who must deliver choice and customization to keep their readers but can’t each afford to master the complexities of each of these delivery mechanisms and all the new ones that may be invented.

Responding to requests from both publishers and subscribers, FeedBlitz added filtering articles by tags to allow readers to choose only those articles relevant to themselves from a single publication or to allow publishers to create focused sub-pubs from a more general publication. More mass customization.

Publishers have increasingly gotten the ability to tailor the look and format of their publications as delivered by FeedBlitz – just as publishers control the look of their websites. The group of publishers using FeedBlitz grew beyond bloggers to people who think of themselves as newsletter providers, local organization coordinators, or even direct marketers – but is still restricted to those publishers who readers make a verifiable request to receive the publication.

Some of these new publishers want to go even further in mass customization – they want each message delivered to each subscriber to be unique and relevant to that subscriber. This is natural in a world where Amazon greets each of us differently and Google searches for what it believes we really want based on our history. FeedBlitz Newsletter Edition had to be invented to allow that level of individual customization.

This all sounds like an ad for FeedBlitz and, to some extent, it is. I think the company and founder Phil Hollows have done a good job of recognizing and accommodating the growing demand by consumers that what they receive be relevant to them and physically delivered the way they want to take delivery. FeedBlitz’ customers are the publishers but FeedBlitz service is to help the publishers meet the needs of their subscribers; it’s really Reader Relationship Management (RRM).

Web tools including FeedBlitz are very cheap – free in some cases. Although Google and Amazon spend large fortunes to mass customize their web real estate, small publishers and community organizations can also afford to personalize. In fact, they can’t afford not to. None of us will stand for one-size-fits-all content anymore.

Web 2.0 is Stuck at a Local Optimum

Sketch of a fitness landscape. The arrows indicate the preferred flow of a population on the landscape, and the points A, B, and C are local optima. The red ball indicates a population that moves from a very low fitness value to the top of a peak. Illustration by C.O. Wilke, 2001.

Sketch of a fitness landscape. The arrows indicate the preferred flow of a population on the landscape, and the points A, B, and C are local optima. The red ball indicates a population that moves from a very low fitness value to the top of a peak. Illustration by C.O. Wilke, 2001.

Tim O’Reilly blogs: “What really needs to be done is not just to connect the various social networks that do exist in internet network-of-networks style, but also to social-network enable our real social network apps: our IM, our email, our phone. Where, I keep asking vendors, is the Web 2.0 address book?... When one of the big communications vendors (email, IM OR phone) gets this right, simply by instrumenting our communications so that the social network becomes visible (and under the control of the user), it seems to me that they could blow away a lot of the existing social network froth.”

When his happens, we’ll have gotten past the local optimum which Web 2.0 is stuck on (which ironically, is that Web 2.0 is optimized for GLOBAL rather than local groups). Implicitly Tim is saying this because he is looking for the vendors of INTERNET apps – email, IM, Phone – rather than vendors of Web apps to be the source of the needed innovation.

[If you already know all about fitness landscapes and local optima as these terms are used in evolution, skip the next paragraph.]

Visualize a landscape pimpled by peaks of various sizes. Visualize a population in which random mutations occur and are heritable or copyable. In this case the population is a bunch of web entrepreneurs so we’re not getting into any creationism debates (today). Height represents the number of unique daily visitors to a web site. Mutations which lead up are rewarded by more capital; mutations which lead down are punished by loss of access to capital just when it’s needed.  Standing on a peak can get you acquired.  The problem comes once a population is on a peak. It is very difficult to get to a higher peak across a fitness landscape because you have to go down to go up. And mutations which take you down are punished.

For very good reasons Web 2.0 services optimized themselves for the global communities which the Internet enabled; this was a good peak to climb and totally unoccupied because, before the Internet, it was impossible to knit global communities together without regard to distance.  Web 2.0 services also optimized for low initial costs of entry and viral marketing; as I’ve posted before, that’s a great strategy for the first social networking service and maybe even the tenth; it doesn’t work for the hundredth because everybody else is doing it.

The great environmental change affecting web businesses today (evolution is a mechanism for coping with change) is that we’ve passed a tipping point for use of the web by local organizations – organizations which already exist. Since it is now more likely than not that the majority of members of most local groups in the US have some sort of broadband access, these groups are ready to go beyond IM and email and even cheap phone calls in their use of the Web.

Web 2.0 entrepreneurs know this. In an excellent taxonomy of social networks, Liz Gannes blogs on GigaOM: “Everyone and their mother wants to build white-label social network to serve an existing interest or community these days, but most of the stuff I’ve tried using is pretty crappy.”

Why is what they build so crappy? Because they’re trying to use the stuff they so successfully developed to serve global groups to meet the needs of local groups. Trouble is local groups have different needs.

Local groups (unlike the global groups formed by the social networks) already exist. This is not about forming groups; it’s about serving them. Some may not even want new members.  Few of them will see the Web as their primary means for getting new members.

Access control may be much more important to local groups than global groups.

Most local groups are not populated by us nerds. They want easy more than cool.

Local groups may not care much about free. Those that have web sites are paying too much for design and hosting and getting too little usability in return.  They could pay something substantial (compared to what Web 2.0 services charge) and still come out way ahead.

From a marketing POV, the decision to use a web provider to meet the needs of the existing group in NOT an individual decision like using del.icio.us or digg; it’s a decision made by whomever or whatever committee is already in charge of communication for the group.

Please feel free to use comments to add to the list of differences.

The great Web application(s) for existing groups will come. My guess is that few if any these applications of them will come from those who have been successful with Web 2.0; they’re trapped on local optima by their prior success when the market was a different place.

The Local Web won’t be Web 2.0+ or Web 3.0; it won’t be the semantic web; it’ll be its own unique self, a branch from lower on the evolutionary tree. And it’ll be huge.

Related posts: (note that it took me awhile to realize that the Local Web is NOT Web x.0)

For Web 2.0 Success - Think Local, Act Local

Web 2.0 – The Global Opportunities in Local

Local – The First Life Opportunity

The Newbies are Coming

Stumbled Upon


Ordinarily the end of December is a low point for blog reading.  Somehow people seem to have other things to do which are more important to them.  So I was pleasantly surprised to see Fractals of Change readership as measured by SiteMeter spiking upwards and had to drill down and find out why.


What clearly happened, according to MyBlogLog is that a post I wrote last May has been Stumbled Upon.  Some member of www.stumbleupon.com liked what I wrote about Why a Great Programmer is Worth 50 Good Ones and recommended it.  Other members of the social network agreed with this opinion and the site began recommending the post to both members who stumbled similar sites and members who say they like things which are stumbled by the members who stumbled this post.

One of the differences between stumbleupon and other social networking sites is that stumbleupon does NOT keep an overall list of the 100 most popular posts or even give average ratings for sites.  In their FAQs, they explain why:  “Because these statistics foster a competitive atmosphere that would promote spamming among owners of the sites in question. StumbleUpon is based on showing people sites personalized to their particular interests, not on absolutes like average rating and top 100 lists.”

The key part of this answer, I think, is the second sentence.  How relevant or worthwhile a site is depends on your interest.  A rating is only relevant in terms of what stumbleupon can infer about your likes and dislikes from the stumbling you’ve done and the people in stumbleupon you’ve chosen to associate with.

Judging from very limited experience, the differences in technique between reddit which uses overall ranking and stumbleupon which looks for individual relevance results in a difference in reader behavior.  Below is a graph of a spike in FOC readership which came when it was heavily recommended on reddit.


Interestingly, it was the same post that was recommended both times.

Here are some differences:

  1. The reddit spike came almost immediately after publication; the stumbleupon one came seven months after.  I’ve noticed that something which doesn’t make it to the top of digg or reddit almost immediately, never does.
  2. The reddit spike was much briefer; the one from stumbleupon spread out and relatively long lasting.
  3. The stumbleupon members typically read multiple posts; the reddit members (like most casual visitors to the site) just the one they were pointed to. (look at page views vs. visits in the stats above).
  4. You can’t see it in the charts but someone stumbled another post yesterday - Managing Programming for CEOs Part 1 – Decompiling Programmer-Speak - and significant traffic is now going there as well.

Too soon to know whether the new stumbleupon readers become regular readers (not many from reddit did) and/or email or rss subscribers.  But I’m impressed by this use of dynamic peer clusters for finding apparently interested readers so you will notice that, at the bottom of  each post, it is now very easy to stumble that post.

Will also keep writing from my programmer (nerd) POV.

You Can’t Get There From Here talks about the technique of dynamic peer clusters as does Dynamic Clusters – Maybe You CAN Get There From Here.

Social Tagging and Fractals of Change is a previous post about social networks.

FeedBlitz Version 2.0 – Reading Blogs in Email

Disclosure:  I am an investor in and a board member of FeedBlitz.

Last week FeedBlitz released Version 2.0 of the world’s largest blog to email service (technically RSS to email).  You can find all the details on the FeedBlitz blog.  This post is about my favorite new feature:  the smart use of the tags most of us bloggers put on our posts (for example, this post is tagged “FeedBlitz” and “Weblogs”). 

Tags are useful both to publishers and subscribers in FeedBlitz 2.0; in this post I’m talking about the use of tags by subscribers because I’m already taking advantage of that feature.

When I first invested in FeedBlitz I thought it was mainly useful for those readers who only look at a couple of blogs per day.  Blog addicts, I thought, will continue to use web-based feed readers like Bloglines and Newsgator.  Wrong, at least for some of us!  I found that I was spending too much time on blog-reading and switched my favorite blogs to email subscriptions.  I live in Outlook, anyway.  Why not get my favorite bogs there?  Getting blogs through email has the additional benefit that I can read them when I’m offline (still does happen, unfortunately).

Now, with smart tags, I get an automatic way to arrange the incoming posts AND a way to filter for subjects I’m particularly interested in or NOT interested in.

I’m about to describe a bunch of things you DON’T have to do in order to receive blogs as email using FeedBlitz. In fact, I recommend that you DON’T do any of them if you’re just starting to use FeedBlitz. It works fine without my tweaks and takes just a few seconds to set up. But, if you’re a tweaker….

Let’s start with the filtering.  When you log on to FeedBlitz, the dashboard lets you click through to a list of your subscriptions.  If you click on the underlined title of a particular subscription, you get a dropdown menu with “Subscription Details” as one of the options. Choose that and you get a page which includes the fields below:


Suppose you’re a nerd and you only want to read the posts in Fractals of Change which I tag with “nerding”.  OK, then, put “nerding” in the include box and that’s all you’ll get.  If you’re a normal person who hates my nerd posts, put “nerding” in the exclude box.  Simple.  Now you get the emails you want and you don’t get the one you don’t want.

How do you know what tags a blogger uses?  With my blog it happens to be easy because they’re at the bottom of each post.  For blogs that don’t have this, hold that question and see below.

Now for some Outlook customization.  Different mail programs are different but I’m just covering Outlook here.  First thing I did was make a rule to automatically move incoming blog emails from FeedBlitz to their own blog folder.  You don’t need to be a nerd to do that.  Here’s how:

  1. Create a blog folder by selecting “New” and then “Folder” in the File menu.
  2. Click on the “Create Rule” icon in the menu bar.Ruleicon_1

  1. Select “From FeedBlitz” and “Move e-mail to folder:” in the dialog below.  You’ll also have to click “Select Folder…” to select the blog folder you created in Step 1.


  1. Click “OK” and you’re done.  How’s that for easy.

Now a little customization of the blog folder doesn’t hurt. Mine looks like this.


You do this customization or whatever suits you by choosing “Arrange By” and then “Custom…” in the View Menu.  Notice that now I can see the tags (Outlook calls them “Categories”) for each blog email.

Not really a lot of work and it lets you keep up with what is relevant to you from your favorite bloggers. But, remember, you don’t have to do any of this customization to use FeedBlitz. It’s just capabilities that are there when you need them.

How Readers Find Blogs

There are five basic ways that readers find new blogs to read: recommendations by strangers aka social tagging services (see immediately below), recommendations from friends, recommendations by robots, links from other blogs, and search engine queries.


This chart shows “walkin” readers to Fractals of Change; these are readers who click on a link or type in a URL to get to the blog.  The vast majority of readers are subscribers but that’s not what I’m blogging about today.  Readers become subscribers after they walk in.

There was a huge spike Friday (actually late Thursday nite) in walkin readership.  It was all because of a social tagging service called reddit. A user named lupin_sansei submitted my post from last Monday on why a great programmer is worth 50 good ones.  Zoom.

When a post is tagged on reddit, it gets a few seconds or minutes of fame up high on the new list page (depending on time of day and day of week).  Reddit users watch this page and a number of them apparently scrutinize each article.  Registered users (registering just requires setting up a username and a password) get to vote articles up or down. 

Some function, which I haven’t been able to reverse engineer, of time and net votes (plus one for up, minus one for down) determines how high on the reddit hot list the reference to a post is.  Only 25 entries per page so there is a high value to being on the first page.  If your post doesn’t quickly get on the first page, it quickly disappears from view. The entry for this post stayed on the top page for almost a day before sinking down to page two, page three, etc.  It is now ranked two hundred and something and fading to oblivion.

Members have an incentive to vote because the service uses their up and down votes to deduce a personal profile and make recommendations which take this into account.  I haven’t used reddit long enough to accumulate a profile so don’t know how well this works.

Two other social networking services are digg and del.icio.us.  I posted about them here and tried to get readers to use them to demonstrate support for another post of mine to the FCC.  Didn’t work.  People tag what they feel like tagging and I’ve never been able to guess what posts of mine will have temporary tag fame and which won’t.  Nor have I been able to influence this.

Fred Wilson posted recently that “user tagging is fundamental.” He was talking about services like those above.  I suspect he’s right.

The second way blogs get new readers or one time readers is through recommendations from friends.  On Friday between about 10AM and midnight only about 17% of the visits to the post on programmers were directly from reddit; 74% of the links were from an “unknown” source which almost always means that someone emailed someone else the URL of a post.  Since traffic was at about eight times its normal rate. These email references to this post (which was not my most current post by this time) were almost certainly as a result of the reddit entry at one or more removes.  But, even on a normal day, many readers come to the blog because a friend told them to.

The third source of blog traffic is by robot referrals.  Memeorandum.com and its brother service tech.memeorandum.com are two such.  Both of these calculate the ranking of a post depending on the number and authority of the links to that post.  “Authority” is a mysterious function of both readership and whether the author of the linking post is a recognized authority on the subject of the post he or she is linking to.  I sort of count as an authority on telephony stuff but not on politics (more the pity).  The last time walkin traffic on my blog soared to ten times normal levels was when the robots which are tech.memeorandum decided that there were lots of important links to my post on AT&T overcharging soldiers for prepaid card calls from Iraq and Afghanistan and featured my post on its front page for more than a few microseconds.

The only other time there was an order of magnitude increase in walkin traffic was when BoingBoing, the granddaddy of all blogs in readership, linked to a post of mine about John Battelle.  Zoom went the walkins.  Links from other blogs are the fourth way that blogs get new readers.

My blog and many others got initial readership from links.  Bloggers Fred Wilson, Brad Feld, Jeff Jarvis, Jeff Pulver, Michal Parekh and many others have been very generous with “link love.”  Clusters of bloggers with similar interests (but not always similar points of view) link to each other frequently so, after a while, we are already sharing many readers so are not as good sources of new readers to each other as we once were.

Plenty has been written about search engine optimization (really optimizing your site so search engines will find it).  I don’t do that; it’s hard enough just to write the posts.  But, on a normal day, about half my walkin readers come from search engines.  May be because I write on a lot of different subjects.  I still get a lot of hits from the word “pornography” because I posted that Pornography Drives Technology back in March of 2005.  Wonder how many of these stick around to become regular readers.

Somewhere in all this may be the secret to blogging fame and fortune.  But notice in the graph above how fleeting fame is.

Now on Kindle!

hackoff.com: An historic murder mystery set in the Internet bubble and rubble

CEO Tom Evslin's insider account of the Internet bubble and its aftermath. "This novel is a surveillance video of the seeds of the current economic collapse."

The Interpreter's Tale

Hacker Dom Montain is in Barcelona in Evslin's Kindle-edition long short story. Why? and why are the pickpockets stealing mobile phones?

Need A Kindle?

Kindle: Amazon's Wireless Reading Device

Not quite as good as a real book IMHO but a lot lighter than a trip worth of books. Also better than a cell phone for mobile web access - and that's free!

Recent Reads - Click title to order from Amazon


  • adlinks
  • adsense