SmithShop

Why I downgraded Twitter from "Strong Buy" to "Hold/Sell"

Sat, 12 Mar 2011 03:12:00 -0800

Ok, so I don't actually make recommendations and, since Twitter is privately owned, it's fairly difficult to do either of the above.

I've have always maintained that Twitter--not Facebook--is the horse to bet on. Facebook, despite its hundreds of millions of users, massive amounts of traffic, stratospheric valuation, and media attention is, at the end of the day, just a social network--a website. Facebook's competitors:

Bebo
MySpace
Friendster
Orkut
etc.

The problem with social networks is that they fade. Facebook is so massive that it can probably see the ceiling in how many users a web site can sign up. A web site that peaks only has one way to go--down. Ask MySpace, Bebo, and Friendster. Facebook will spend the rest of its life trying to stay relevant, keep users engaged, and squeeze every last cent out of them. Yawn.

Twitter?

Twitter isn't a website, it's a brand new communication channel. Twitter's competitors:

Email
Websites
SMS
Snail mail
Radio
Television

Communication channels have a way of sticking around.

Remarkably, Twitter Co. OWNS this communication channel and ultimately can do whatever it wants with it.

If ever a company was right to grow first and monitize later, it's Twitter. It has reached a critical mass and it's not going anywhere. Twitter will be here forever.

Why the downgrade?

By telling developers to stop building clients, they are telling developers "We are a social network" rather than "We are a communication channel". But they are wrong. They are a communications channel and they are telling developers to stop writing tools to make better use of it. It's the same thing as telling Alexander Graham Bell not to make a phone, Jack Smith and Sabeer Bhatia not to write Hotmail, Google not to build Gmail, Mozilla not to make Thunderbird, or RIM not to bother with the Blackberry.

Email clients have driven email to greater heights and created opportunities for companies like SendGrid, Campaign Monitor, Constant Contact not to mention the million other companies that drive sales and tell stories via email.

Twitter has made a huge mistake.

This decision isn't going to kill Twitter as a communication channel but it's going to slow its growth. They need to take the shackles off of developers. No more rules. Send and consume Tweets however you want. Send as many advertisements as you want. Make the best clients that filter out Twitter spam and provide the best user experience. Make an app that makes it easy for groups to Tweet together. Make Twitter for the enterprise. Encrypt Tweets. Charge for these apps or give them away. Do whatever you want. Tweet, tweet, tweet.

My Suggestion

Create a two-tier Twitter ecosystem. Create two queues. Free Tweets in one queue and the Tweets that people pay Twitter $.01 to send go in another. Delay the free ones by an hour. There you go--Twitter is now making a billion dollars a year.

Permalink | Leave a comment »

14 Day HN App » Day 15: Where is Rankoro.us?

Wed, 01 Dec 2010 02:23:00 -0800

Well November 30th came and went and, as of yet, Rankoro.us is a no-show. My laptop power supply/chord died and it's going to take a few days to get a replacement. I'm going to shoot for Sunday, December 5th, assuming I can get going again by Friday.

Permalink | Leave a comment »

Price Convergence and Mailing List Services

Thu, 25 Nov 2010 02:10:00 -0800

The company I work for sends a lot of newsletters. It's an important part of our business model, a revenue driver, and a significant cost. We've switched providers twice since I came on board. Our first provider (I can't even remember the name) was one of the "old school" services--not very polished, not terribly expensive--but, well, we got what we paid for.

We switched over to Campaign monitor because we felt the delivery rates at the initial service just weren't cutting it. Campaign Monitor proved to be an incredible service. It had a fantastic API, allowed us to segment and target our list in multiple ways, and had very high delivery rates. Unfortunately, it was rediculously expensive:

Our mailing list has about 80K subscribers and we send about 3 newsletters a week. That makes for (3 per week * $5) + (3 per week * 80K subscribers * $.01) = $2415 per week or around $10,000 per month. Now, in fairness, they offered prepayment/bulk discounts that would allow you to lower the $.01 per email figure down to $.006-$.007 per email but that required you lay down $10-20K all at once. Very few small companies can do that.

We switched over to Constant Contact because they offered all-you-can-send (within reason) packages priced according to the mailing list size:

For our list of ~80K subscribers we pay about $450 per month. Compared to the best-case, super bulk pricing of Campaign Monitor, it was about 1/3 the price. The switch easily saved us $15,000-$20,000 a year

Unfortunately, well, you get what you pay for. Constant Contact isn't a bad service, it just isn't as polished as Campaign Monitor. They have a good API but some of the Campaign Monitor tools and features just aren't there. We had to make some sacrifices and work arounds.

I'm not changing mailing list providers again. Constant Contact is fine, they have good customer service, and their price is competitive. So competitive in fact, that the entire industry seems to be heading that way. Mailchimp's pricing structure is eerily similar:

So how does Campaign Monitor survive when their prices are so much higher than the others? They don't, they adapt. Here is their new pricing structure:

It's still a bit more expensive at certain price points but hey, you get what you pay for! I'd gladly pay their slightly higher prices for their (atleast slightly) better product but the switching costs are too high so I'm staying put.

Permalink | Leave a comment »

14 Day HN App » Day 7: What is Rankoro.us?

Wed, 24 Nov 2010 02:14:00 -0800

Good progress last night. Long way to go. Thought I'd take a break and explain a bit what Rankoro.us is...

My November Startup Sprint entry, Rankoro.us, hopes to create new conversations around always controversial "Top 5" lists. I know there are dozens of apps where you can create and share lists and LivingSocial flooded Facebook with them a year ago but I think they failed in a few ways (or didn't even try):

They had no class. Not in the way Quora and Hacker News do. LivingSocial Top 5 lists had a real spammy Farmville thing going on.
They didn't create thoughful conversations. Most of these lists are a "I'll show you yours if you show me mine." Conversations were limited to Facebook comment threads which tend to degrade to mainly LOL's and OMG's pretty quickly.
Their UI's are secondary. As Hipmunk (flight search), Workflowy (first TODO app I've ever really liked), and Tumblr have shown, you can take an old "idea" and make it new again with a thoughtful user interface.

Rankoro.us is going to move the conversation off of Facebook and onto a site built completely around lists and the conversations they create. Here's why it's going to work:

Nothing is more contentious than a Top Five list. Opinions are like.... err... bellybuttons... everyone has one. I think Top Five lists in particular spur better conversations than a simple question, etc.
Initial product research has been positive. I'm convinced. Mom thought it was a brilliant idea. A friend of mine said she'd "totally use it".
There is going to be a reputation/karma system. I hope this makes people want to get involved and, more importantly, encourage high-quality comments and thoughtful lists.
It's going to be self-promoting. I'll spend sometime on SEO but the majority of the promotion will be by the creator of the list.
The UI will be great. Ok, since I've 2 full days and 2 nights left to write it (day job, soccer tonight, and then Thanksgiving) , the MVP might not be the bee's knees but if it gets any traction at all it'll be slick in a few weeks after launch.
You control the conversation. All lists will be public but to get involved in a conversation you must a) have a list yourself (put up or shut up) and b) be "connected" to the user via an invitation/follow model. (Note: The workings of the commenting is still being figured out).

I've included a few Microsoft paint drawings I made to try and convey the idea to my friend Pepe who is doing the design work.

Figure 1. The list creation flow. Left to right, top to bottom.

Figure 2. The Top 5 page and conversation.

Figure 3. Slightly refined Top 5 page.

Please sign up for the announcement email-you can't have late adopters without early adopters or follow me on Twitter (@rankorous).

Permalink | Leave a comment »

14 Day HN App » Day 6: Clay Rested and saw a gig

Tue, 23 Nov 2010 03:59:00 -0800

No development on Day 6. I had chicken wings and went to see Blitzen Trapper play at the Workman's Club. Good show--entertaining--but not really my type of music.

Permalink | Leave a comment »

14 Day HN App » Day 3-5: Coming soon page live and development underway

Mon, 22 Nov 2010 02:28:00 -0800

Well, I had a pretty productive weekend--worked Friday night, Saturday day, and most of Sunday (had to take a break to watch the Falcons move to 8-2!). I finished setting up the hosting for Rankoro.us--two Rackspace Cloud server instances (an admin server and a public facing server) with php, MySQL, memecached, all sorts of monitoring (nagios, munin, AWStats, webalizer), etc. This always takes longer than I would like--one of these days I'm going to checkout chef/opscode.

I put up a basic "Coming Soon" page on Sunday morning to hopefully gather a few email addresses or twitter followers before the launch. The coming soon pages are just html, a few images, and a small script that adds the email addresses to a Constant Contact mailing list. With a little luck I won't just be announcing Rankoro.us to my mom next month.

Actual development got underway on Sunday--44 subversion commits as of this morning. The database is fairly well mapped out, I've got login/register functionality built, and I'm onto the main two pages that compose the app. In order to make sure I have *something* to launch next Tuesday, I'm going to iterate the UI from "simple" and "server-side" to "rich" and "ajaxy". Worse case scenario I'll have a slightly-less-polished-than-I-would-like version.

The November deadline is definitely what I needed. It's got me really focused and I am pretty confident I'll make the deadline. Hopefully the end product will be fun/useful for people.

Permalink | Leave a comment »

14 Day HN App » Day 2: Project, Hosting, and Build Set Up

Fri, 19 Nov 2010 02:07:00 -0800

Day 2 and I'm already behind. I had hoped to have a landing page up and working but instead I opted to set up a more robust hosting platform. An extra days work here will make life a bit easier next week. Here's what I did accomplish:

Set up source control. I use Beanstalk's subversion hosting. The $25/mo I pay (for up to 25 projects) is probably a bit high but having a solid source control/repository is definitely worth the money. One of these days I'll look into Github.

Set up initial project. Rankorous is going to be written in php with a mysql database. It's what I know. I copied the last project I was working on and removed any project-specific code and voila I have the latest evolution of my "framework" that's grown over the last few years. I set up a few databases (each project I work on has 3 identical versions of the database: one for development, one for running unit tests, and one I use w/ my home-grown ORM/code generator). I currently use Eclipse for development but might make a switch soon.

CDN Setup. I typically use a CDN (Rackspace Cloud files) to host images and other assets (css/js). It improves page load time and reduces server load. For Rankorous, I set up a few buckets (or whatever they are called) that I'll use for hosting.

Server setup. I always set up my hosting and build process at the beginning of a project. Working with people around the globe, I find the ability to push changes up to a staging server during development greatly speeds up the development process. For this project, I'm going to have two servers:

Public Web Server. The public web server will handle most of the web requests and runs Fedora, MySql, Squid, Apache, and memcached.
Admin Server/Application Server. This server hosts the web site's administrative interface and any of the more intensive processing tasks (there won't be many). The idea is to tune the web server to handle a heavy load and let the less "visible" aspects of the site degrade out of site.

I created the instances of both of these servers but didn't get far in setting them up. That is tonight's task. It looks like I'll be: configuring and securing both servers, setting up administrative and monitoring tools (phpMyAdmin, Nagios, Munin, Webalizer, AWStats, etc), and setting up my "one-click" build process. By the end of the night I'm going to have a basic, hello-world style landing page in place.

Permalink | Leave a comment »

14 Day HN App » Day 1: The Ground Work

Thu, 18 Nov 2010 08:28:00 -0800

Well, last night was Day 1 of my "14 Day Hacker News App". I'm going to post what I got done the previous day and what I am planning on doing that day/night. Last night (aka Day 1) I got the ball rolling on Rankoro.us:

Reserve a domain name. Well, in this case two domain names. I Reserved Rankoro.us and Rankorous.com through GoDaddy.com. I'm leaning towards Rankoro.us but I like to reserve the equivalent .com whenever I'm using a non-.com domain.

Set up social network accounts. I set up a Facebook Application and Twitter account incase I need them down the line.

"Finalized" wireframes. Ok, so MS Paint isn't really the Wireframe method of choice but it's quick. I made a few rough sketches and sent them off to a buddy of mine who owes me some design work.

Transfer DNS Nameservers/set up DNS. I hate GoDaddy.com for everything except the cost of registering. I've been trying out dnsimple.com (started by a Hacker News reader) and so far it's making life a little simpler. I pointed the nameservers at my dnsimple account and set up a few basic DNS records for Rankoro.us and Rankorous.com.

Set up Google Analytics and Google Webmaster accounts. Not ready to use these yet but I set up and verified both using the DNS record option so they will be ready to go when needed.

Added Rankoro.us and Rankorous.com as aliases for my Google Apps account. Easiest way to get mail set up for a new domain--just make it an alias of a Google Apps account (I have the $50/yr premier version, well worth it) and set up some MX records (back to dnsimple.com) and it was done..

Create an account with ConstantContact.com to hold my mailing list. I don't plan on sending any emails for (exactly) 2 weeks but I'll use the account to create my pre-launch mailing list. Even after your trial period expires, you can still add to your mailing list. Great for those projects that never seem to launch.

That's all. Tonight I'll be setting up hosting using my Rackspace Cloud servers, setting up my initial project (it's going to be php/mysql), and get a landing page set up to start collecting some email addresses for my prelaunch.

Permalink | Leave a comment »

Idea-to-Launch in 14 days: Participating in Hacker New's "Launch an App" Month

Wed, 17 Nov 2010 07:39:00 -0800

I'm a bit of a Hacker News lurker and have been really interested in the movement encouraging its readers to launch a new app in the month of November. The idea spurred a lengthly conversation, a Facebook group, its own meta-app and numerous follow up posts. The two small sites I put up this month (StartupBites and Is McRib Back) hardly qualify as "apps" and I didn't have any bright ideas so I figured I'd just watch from the sidelines.

Well, last night I came up with a new idea for an app (or new take on an old idea). I think I can get it done in two weeks so, what the hell, I'm in. It's (tentatively) called Rankorous and is an app for creating and sharing Top 5 lists with friends. I'll do my best to blog the next two weeks with updates and articles about what I'm up to.

Note: This is going to (hopefully) be supported by advertisements and affiliate income so instead of "One Paying Customer" I'll set my goal as "some adsense/affiliate income".

Permalink | Leave a comment »