isabel wang

Web Hosting + Web 2.0 = $

The folks at SWSoft say that their server automation software is in use by over 10,000 web hosting providers in 100+ countries (and their market penetration is nearly 100% in this industry). Together, these companies are home to tens of millions of websites. Over the past 10+ years, their users have created an enormous variety of collaboration and knowledge management tools. Sadly, surprisingly few of these ideas have been adopted by the web hosting firms who played a silent by instrumental role in their development.

Let me give you an example. While I was VP of Communications at EV1Servers, a customer complained that his support ticket was poorly handled. He asked about configuring software RAID on a Red Hat Enterprise server; the technician closed the ticket with a brief note that the issue was beyond the scope of support under his service plan. If he wanted more information, he could search on EV1Servers' FAQ or customer forum.

After performing a quick search, I found that neither the FAQ nor the customer forum contained any reference to software RAID. I found some relevant documentation on Red Hat's website and forwarded it to the customer. Since he didn't feel confident about performing the work himself, I later referred him to another customer who offers sysadmin consulting.

Over the following few weeks I received 4 or 5 other software RAID inquiries. I was able to cut and paste the info I previously collected, but the knowledge that I took time to obtain had no way of propagating across the company's other staff. And the relevant information was not distributed to other customers who might need it. There could have been dozens of other software RAID-related support requests, but I had no means of looking them up.

I suppose I could have created an FAQ entry. If every employee got in the habit of doing so, EV1 might have ended up with a knowledgebase that's as complete as what GoDaddy has. But even there, the customer is unnecessarily presented with an either-or choice. Would he like to search the Help Center or open a support ticket and wait? I know that there is available technology to implement a more efficient tech support process. It would work like so:

1. A customer logs into his hosting provider's support portal and submits his question (for instance, "how to configure software RAID on RHE4?") The system automatically queries the knowledgebase for possible matches.

2. The customer can end the session by indicating that the search results gave him what he's looking for. In this case, perhaps the hosting provider could offer him some small incentive (hosting credit? chance to win prizes?) to update relevant knowledgebase entries with any additional information he might have.

3. On the other hand, if the customer is not satisfied with the  search results, his question automatically enters a support queue, where it awaits human response. When support reps receive tickets, they are prompted to complete knowledgebase entries in addition to answering the customers' questions.

4. It'd be cool to offer customers the option of bypassing the queue for a fee. The hosting provider could offer paid fast-track responses in-house - or create a Amazon-like marketplace in which third party consultants vie for the customer's business. Each consultant's listing might show pricing, customer satisfaction ratings, as well as whether he or she is currently online. If engaged by a customer, the consultant would pay the hosting provider a commission.

Earlier this year EV1Server's trouble ticket system logged its 1 millionth support request. A live human manually responded to each of these tickets. If each incident took 10 minutes to resolve, that's 166,666 man hours the company has had to pay for. Wouldn't it have made sense to redeploy some of the millions this has cost on a backend solution that could turn tech support into a profit center?

Not a Rice Krispies Treat - Yet

In his Wired article on crowdsourcing, Jeff Howe sets forth 5 rules for the new labor pool: the crowd is dispersed/attention deficient/full of specialists. It produces mostly crap, but as a networked community, it ferrets out the best material and corrects errors.

Snap, GoTo/Overture founder Bill Gross' newly (re)launched search engine, shows that some of Howe's rules might not always apply.  Snap has organized a launch contest to solicit public input on its promotional campaign. Visitors are asked to submit and cast votes on advertising and marketing ideas. The company is offering a $5000 prize to the author of the most popular submission.

I was reading the proposed marketing campaigns; most target a too-narrow audience and can't be implemented within Snap's $25K-$50K budgetary guideline. It's one thing to reach a dispersed and attention deficient crowd; there's no guarantee that the group will contain experts on the desired subject matter.

I was particularly unimpressed with the highest-ranked entry. It suggested that Snap should ask MySpace users to add its search module to their personal pages. Snap would pay profile owners $0.50 to $1.00 each time a visitor downloads the search module. In this case, I don't think the networked community chose the best solution among the 154 submissions.

First, Snap would incur significant costs just to advertise the download promo to MySpace users. Also, most viewers of MySpace profiles are already MySpace members - they wouldn't generate much exposure beyond the MySpace demographic. Last but not least, downloads don't guarantee ongoing usage. All in all, this doesn't sound like a plan that would generate cost-effective traffic.

What would I do instead? Nothing. I don't think Snap is ready for a major marketing push. I searched for "Windows web server", and the top result was http://httpd.apache.org/. "Buy shoes" got me a spam site (http://www.bannershopmall.com). Technical terms like "database clustering" and "hardware firewall" yielded fewer than 50 visible results. (Oops - I should back up. Unlike Google/Yahoo!/MSN/etc, Snap doesn't display search results in text. Instead, it has a navigation pane that shows previews of each website. Would be cool - if combined with relevant results.) "EV1Servers" brought up 5 related pages - 2 of those were press releases from 2003!

Maybe the company should stay under the radar; in which case its visitors' 4th most popular recommendation (sign up with Amazon's Mechanical Turk) might be a fine idea. The Turk is a crowdsourcing pioneer. It lets website owners pay as little as $0.01 for "Human Intelligence Tasks" - such as answering market research questions and transcribing podcasts. (One really nifty example is The Sheep Market, in which UCLA grad student Aaron Koblin paid 10,000 visitors $0.02 each to draw a sheep by hand. He now sells sheep stamps for $1 each.)

The submitter of this suggestion thought it might create a viral effect; I'm not so sure. I don't see Snap getting much exposure beyond the very limited population of Mechanical Turk workers. And that would be a very good thing until it's made more progress on refining its search results.

PigeonRank is not enough

About a year ago, I persuaded a Taiwanese customer to sign up for a server cluster at EV1Servers. We had several phone conversations - in Chinese - before he finalized his order.  Although I lived in Taipei until I was 15, my vocabulary back then did not include "firewall throughput" or "database replication". So thank goodness for Dictionary.com; without its English to Chinese translation tool, we would have had a very hard time understanding each other.

I was annoyed at the customer's refusal to speak English. He was a former research scientist and the administrator of a fast-growing website. Why wasn't he making more effort to become fluent in the "official language" of these fields?

Recently, though, I've come across a number of really interesting Chinese-language blog posts - not to mention a Chinese article on server virtualization. The authors and commenters have so much knowledge and insight on such a wide variety of topics.

The English-language blogosphere has Michael Arrington and Om Malik; sooner rather than later (or has it already happened?), an influential cadre of other-language technology opinion makers will emerge. But whereas a significant proportion of their audience also reads Slashdot, blogs on Wordpress and shares photos via Flickr, the English-as-only-language (EOL??) crowd won't have the same all-access pass.

Stats from Technorati show that some 60% of Q1 2006 blog posts were non-English (15% Chinese, 37% Japanese, etc). Some of them might well introduce ideas or facilitate collaboration that lead to groundbreaking technologies. Unfortunately, those who don't speak the right languages will automatically be stuck behind the velvet ropes.

So, if Microsoft wants to challenge Google's search dominance, what they've got to work on is some kind of tool that translates both queries and results to/from every major language on the fly. It'd take more than a flock of pigeons. Maybe Bill G will have to invest in a few million parrots? Or better yet, an army of Big Birds?

Two Amazons

I was reading the Wikipedia articles on the Amazon Rainforest, and Amazon the online store. The former encompasses 1.2 billion acres and is the largest and most species-rich rainforest in the world. But it suffers from rapid, ongoing deforestation, raising serious concerns about the resulting loss of biodiversity. The latter has a market cap of $14.7 billion. And while 13 out of 15 analysts give its shares a "hold" or "sell" rating, I think its fate will surely be the opposite of its namesake's.

Yesterday Amazon.com emailed me 8 recommendations based on the 286 items from my order history. At first I could hardly believe it; 286?? But apparently Amazon was where I bought everything from my iPod and digital camera to two pairs of sneakers and three catnip toys. And this morning my apartment building's doorman handed out Amazon packages as if it's Christmas. Paul the retired Bureau of Engraving manager got a vacuum; Iraki the super do-it-your-selfer got a router (the woodworking rather than wireless networking kind).

But while its product mix and customer base are impressively diverse, what's even more awesome is Amazon's reach.

When I signed up for a TypePad account some weeks ago, I was surprised to find that whenever users create lists of books/CDs/movies, TypePad auto-generates links to those items' product pages on Amazon. Per TypePad's suggestion, I joined the Amazon Associates program. Links in its documentation led to Amazon Web Services (AWS), an initiative that encourages developers to use the company's product data and e-commerce functionality to build their own applications.

Associate-O-Matic (AOM) was among AWS' case studies. After using AWS to develop two online stores, AOM founder Justin Mecham decided to write and sell a simple application for web-based store creation.  I downladed AOM's demo and it worked as advertised. I was able to create this store in one evening using its single-page web interface. While it's neither attractive nor sophisticated, it does offer the basic functionality of collecting orders and forwarding them seamlessly to Amazon.com. My total investment was approx $10 for a domain name and a two-month hosting package at GoDaddy.

And, in addition to casting an ever-wider net through even the newest, tiniest blog writers and website owners, Amazon keeps meticulous track of the orders that come through. Apparently 224 of the items I bought were books. All of them can now be tagged, searched and annotated in the newly launched YourMediaLibrary. Steve Rubel of Micro Persuasion sees the potential for Amazon to become a social networking venue where customers can connect with others who enjoy similar books/music/films.

That would be way awesome. Unlike most social networking sites, where visitors' desire to spend time with one another makes them less receptive to advertisers' messages, Amazon would play a very different and much more profitable role.  Its customers would bond over items they've each ordered - and give each other recommendations on what additional products could be purchased.

Crowdsourcing Zagat

I'm looking at the very pretty demo on AreYouHere. It's a mobile social network that notifies you when your contacts are nearby, and gives you the ability to text message your friends, their friends, and total strangers who are up to 6 degrees away from being your friends.

My first reaction was - yikes! Isn't everyone within 6 degrees of everyone else? But presumbaly AreYouHere has a "my contacts only" setting. And while I'm not sure I'd use it on a day to day basis, it'd be great at conferences or on group trips. Much easier to round up people from far flung corners of convention centers or museums when you know where they are. (BTW, AreYouHere is only available in London/Paris/Geneva for now.)

I'm much more intrigued with Socialight, a service that lets users place virtual Post-it notes in the real world (the company calls them "StickyShadows"). When a Socialight subscriber approaches a StickyShadow, he receives the associated message on his phone.

It looks like Shadow owners can limit access to their content (public, private or people on contact list), but Socialight's FAQ doesn't say whether it offers subscribers any filtering capability. On its list of "how to use" examples, the company suggests that a band could promote upcoming shows by leaving a StickyShadow outside the venue.  But... as a point of reference, I live near the Black Cat, a club where more than a dozen bands are scheduled to play in the coming week. If all of them used StickyShadows, I don't think passers-by would be pleased with the overwhelming number of messages they'd receive.

I'm wondering, though, whether features from AreYouHere and Socialight can be combined to create a Zagat-like service? It would work like so:

1. After dining at a restaurant, users rate their experiences by posting StickyShadows

2. A subscriber could search for data on restaurants near his current location by requesting StickyShadows posted by specific friends who are known connoisseurs, members of his 6-degrees network, or the general public.

3. Better yet, maybe the service should tabulate average ratings from the user-speficied-demographic and output a consolidated list of options - along with special offers from sponsors?

This sort of reminds me of the "crowd-sourcing" article I read in Wired last night. It talked about increasing opportunities for amateur photographers and garage scientists to compete with experienced experts. There's a great deal of knowledge within the crowd. What's needed are low-cost filters that can separate high quality content/ideas/solutions from the noise.

Zagat's content is already crowd-sourced; it gets its reviews from everyday diners instead of professional food critics. But maybe even its annual task of reviewing/consolidating this content could be delegated to the masses by giving them choose-your-own filters?

Dog Dashboard

OK - I think I have it figured out now. Actually, Dogster feels incomplete because there aren't enough points of intersection between owning a pet and spending time on the site. You might enjoy looking at dog profiles - but what's in it for your dog? And you're proud when other people give your dog virtual treats - but he'd much prefer real ones.

Ted the Top Dog writes that his goal is to blend revenue desires with the ultimate priority of keeping the community happy. He's absolutely right. Most of my previous ideas focused on what might interest advertisers.  From a pet owner's perspective...

(a) I still think Dogster can do more with market research. I'm envisioning an Amazon-like system that allows users to learn about (and, ideally, buy) food/supplies/etc through aggregated data on other consumers' preferences.

(b) I decided while dog-sitting for my friend Sophy that she needs a "dog dashboard" - for storing dog-related data (medical records, vet appointments, etc) and facilitating dog-centric collaboration. Instead of emailing me instructions for Mile's care and feeding, she could post it on his profile, along with info on any recent medical problems and new favorite toys. And I could post updates to keep her informed while she's away.

To make a long story short, I think Dogster would have a tighter hold on its community - as well as find more revenue opportunities - by more closely integrating its features with users' real world dog activities.

Bring in the Geek Squad

Just read this article in the New York Times on BestBuy's Geek Squad. A few weeks ago, when the power supply on my old laptop died, I paid them $59 to move the still-functional hard drive to a new computer. They were all done before I finished my latte from the Starbucks next door. Instant gratification; what could be better?

My neighbor Irakli complained that $59 sounds like a racket for a 10-minute task, but he's an extreme do-it-your-selfer who built his speakers, his track lighting system, not to mention most of his furniture, from scratch. According to Robert Stephens, Geek Squad's founder, in many cases what his customers pay for is simplification.

Even Home Depot, whose core customers are DYI fans like Irakli, is increasingly catering to the do-it-for-me crowd. Its website now features 30+ installation services. And Stephens sees potential among a third group that he calls the "I thought I could do myselfers", or users who rethink the possibility of hiring an expert after running into unexpected stumbling blocks.

So I disagreed when Howard Park from GI Partners said that his new investments (ThePlanet and EV1Servers) belong in the unmanaged web servers market. In fact, I would argue that such a market doesn't really exist.

There are dozens of companies that offer web servers for lease with limited tech support. And tens of thousands of budget-conscious website owners who use their services. But a vast majority of these customers did not sign up for the love of installing firewalls and updating operating systems.

The lack of support is not a feature that users demand. It's simply a fact of life for those who need significant amounts of bandwidth and storage space at affordable prices. While some site owners do develop competency for server administration tasks, and others hire staff or find third party consultants, I think even the most technically advanced user would appreciate a safety net. All things being equal,  customers would surely choose a hosting facility with paid help on demand, rather than a data center in which no such option is available?

There are no fewer "I thought I could do myselfers" among web hosting customers than in BestBuy stores.  Their needs have not been met at many hosting facilities because their service providers don't have the resources to build a Geek Squad like support infrastructure.  But GI Partners comes with institutional capital - and a mission to professionalize the industry. In which case, I hope they'll follow BestBuy's example instead of accepting the status quo.

By the way, the Geek Squad's Robert Stephens calls service "nothing but a profit center". He says more and more customers are asking his staff for advice on what technology products to buy. Of course, each product purchase creates a new service sales opportunity. Doesn't that sound like a good path to positive ROI?

Clear

During the longer than expected taxi ride from DFW to downtown Dallas, I managed to carry on a stilted Urdu conversation with the Pakistani driver.

I probably only have a 100 word vocabulary. Still, we were able to discuss Bollywood movies, the sizable South Asian population in Dallas, the too-high cost of living in New York, and my too-fat cat back in DC.

It reminded me a little of trying to explain server virtualization to a banker I met at ISPCON. He started out telling me that he 'didn't speak tech', and was very surprised to realize that he actually totally got the concept. That something so mysterious-sounding could be described in 'regular' words.

I thought both incidents were pretty neat. When people talk about customer education, it's often made out to be such a Herculean task. But maybe the product documentation just needs to be reworded into something as rudimentary as my Urdu vocabulary.

Signs

I woke up 3 hours earlier than usual to go to ISPCON. The conference has been around since 1993. I produced a short-lived copycat event called Web Hosting Expo in 2001. And HostingCon pays homage in more ways than its name. But nobody does it like Jon Price, who totally ought to start his own web hosting event.

The discussion at my virtual private servers panel was pretty lively despite its too-early timing; more on than in another post. What I wanted to write down before I forget are three conversations I heard/had at the show.

1. Paul Stapleton from DH Capital led a very well attended panel on M&A dealflow. Tom Millitzer, who specializes in representing sellers, commented that if a client is organized enough to compile due diligence data in under 2 weeks, the transaction is almost guarantee to go smoothly. And Joe Bardenheier, who's made 42 acquisitions in the web hosting space, said with a straight face that a 375-step integration process follows every deal.

2. Rich Bader told me about his ITIL implementation over lunch. He documents all equipment configurations at his data center in a database, and a committee meets daily to make sure any and all changes are reflected. How does his staff find the time? Don't they have new orders to deploy, and support tickets to address? According to Rich, it's all about forcing yourself to adopt the right habits. If you take into account all the time people waste on trouble-shooting someone else's poorly documented work, your ITIL investments do pay off in the end.

3. Candice Rodriguez and I commiserated about advertisers with multi-million dollar marketing budgets but no ROI tracking systems. We met when I ran ISPcheck (now HostingCatalog) and she worked at Dialtone Internet (now Interland). Now she's the Director of Business Development at the Web Hosting Industry Review.

Oh, and before all that, I got a stern lecture from my 6AM cab driver about how smart people manage their time carefully, so as not to go half-awake to important events.

Some years ago two separate friends gave me the Alchemist and a book on Feng Shui in the same week. Both were about watching for signs from the universe, because the right path tends to reveal itself unexpectedly. Goodness! Maybe the universe agrees with my mother that I should learn to be a more organized person?

Lost Dog(ster)

Recently I met a woman who owns a dog spa. She said Americans spent $35 billion last year on their 73 million dogs and 90 million cats.

I thought of these stats last night, when my neighbor told me his puppy is on Dogster. I looked him up, and, while at it, posted my cat Sara on Catster. Although I'm a fan of dog and cat pictures, the site feels incomplete, even with its ~240K pet profiles. Maybe it's because I see so much untapped potential.

1. In the "favorite foods" and "favorite toys" sections of pet profiles, some owners go through the trouble of listing specific brands and flavors.  It seems Dogster does nothing with this data, other than displaying the text. What a waste of market research - not to mention sponsorship opportunities! Purina would surely be pleased to have all references to Mighty Dog linked to a product page, where it's shown to be the leading brand among brown labs.

2. Another missed opportunities is medical conditions. Sara has asthma; I would love to commiserate with people whose cats are similarly afflicted. And I think there would be no shortage of vendors willing to sponsor an information center featuring treatment options.

3.  Speaking of medicine, might a pet address book/calendar make the site stickier? I can't remember when Sara last went for a check-up; I think it'd be neat to have a central repository of pet-related info.  In addition to vet appointments and medical records, maybe owners could also use this feature to keep track of play dates, pet-sitters, etc.

4. Dogster sells premium memberships for ad-free pet photo viewing. Maybe it could also offer paid profiles to vendors? Pet owners could become "friends" with local vets and dog walkers, for instance.  This would provide the added benefit of allowing prospective customers to ask for feedback.

5. Another premium membership possibility is to imitate Squidoo. Give pet owners a portion of ad revenues generated through their profiles, and/or allow them to set up Amazon and Cafe Press affiliate program links. The monetary incentive might encourage users to promote their profiles, and keep content up to date.

6. And what about a Google Maps mashup - sponsored by a Petsmart near you - where users can see how many brown tabby cats or white Poodles live in their neighborhood? Some sort of eVite/Meetup functionality might be cool also. Some owners might like to meet the 78 other Gray Hounds in their zip code?

Ok - I'm just rambling. And Dogster is already profitable. Has been for about a year, I think. But theirs is a $35 billion/year market, with sooo many revenue generating opportunities they could take advantage of...