The main purpose of Cohaico is to let you quickly see what the whole twittersphere thinks about a particular movie or gadget.

At First

We look at all tweets and extract the ones that contain a reference to a particular object (e.g. "Moneyball") and then we show those tweets to you.

Alas, there are many people on Twitter and lots of them have something to say about Moneyball. So we started looking at all of these tweets and showing you only a select few, which we think are the most interesting ones. And if you sign in with your Twitter account - we show you what your friends said first.

We also showed how many mentions are there for a particular object - so you can tell how "popular" it is.

That's not good enough either. Many times you just want to see a bottom line of what do "people in general think" about something.

So Later

We started analyzing the sentiment within each tweet: how "positive" or "negative" it is.

Example:

"I just watched Moneyball and it's a masterpiece. I loved every moment." 

Positive.

"I just watched Moneyball and it sucks bananas. I'd rather have my hair picked one by one than watch this crap".

Negative.

And what we did after analyzing sentiment is show you how many of the total mentions for Moneyball are positive.

So know you could (in theory) know how good Moneyball is according to Twitter.

The problem with this turned out to be two-fold.

Because different things have different numbers of mentions, it was hard to compare one thing against another.

Say you have "Moneyball" with 1,200 positive out of 8,000 and "Real Steel" with 2,880 positive out of 17,500. Yeah if you're good with math you can compare those two and figure out that Real Steel is slightly more "positive", but it's not that easy.

Also, just by looking at one object, it's really tough to decide wether 1,200 positive out of 8,000 is good or bad.

So, we could have just shown it in percentage, right?

Wrong. The other problem was this: turns out that in addition to tweeting positive and negative things, people also tend to say things that are... neutral.

Example:

Going to watch Moneyball with my BFFs oogie and shoogie.

And not only that, turns out that most tweets are neutral.

So using percentage to display how good something is was tricky. You may take something like Moneyball having 1,200 out of 8,000 and show 

Moneyball = 15% positive

But people are used to certain scales when it comes to "quality / ratings", these scales are usually 1-10, 1-100, 1-5 etc.

So something like "15%" sounds pretty bad. But in reality, if you look at all movies on Twitter, it turns out that 15% positive out of all tweets is very good.

So what did we do?

What we did was take all of these percetages for all the objects in a particular vertical (say "movies") and "normalized" them on a 1-100 scale. So the movie having the highest percent of positive mentions (say 40%) will be "100" and the movie having the lowest percent will be "1".

And using this normalized scale we now show how positive a particular object is:

One more thing

But showing a normalized score on a 1-100 scale is not enough still. What you also may want to know is how is this object positioned against others in the same category. That is why, we also show the position in category (e.g. "#2 in movies").

This quickly gives a complete "bottom line" for something, based on the things people tweeted about it.

Clicking the position in category line, leads you to the "most recommended" view for this category, to let you explore further, and find more great movies and gadgets.

We've been talking a lot about how a lot of domain-specific conversation on the internet is moving towards "social" channels like Twitter and Facebook and how websites will need to tap into those conversations to show the most relevant content to users.

Here's another example I noticed today:

The Weather Channel is now showing tweets about the weather for any US city or ZIP code. Smart!

What's an older conversation and more basic thing to talk about than the weaher, right?

Do you know of other examples of using Twitter conversations for a specific topic on 3rd party websites? Let us know in the comments.

This is the first in a series of posts where we will describe how we implemented different features in Cohaico and the various design decisions we made.

Star This

After signing in, you can quickly bookmark/remember any item by going to its page and clicking on the large star at the top of the page:

For example: the page for Colombiana

We used an "empty"/"inactive" feeling star icon to indicate that this can be clicked (and then supposedly indicated by "filling" or "activating" the icon).

If still unclear - we also show the tooltip "star this" when mouse-hovering over the icon. We don't want to show the text next to the icon at all times because we want to keep the page as clean as possible.

When you click the "star" icon - the item you're looking at becomes "starred" and accesible via the "starred items" icon in the menu:

Sharing

We also wanted to make "starring" an action that encourages users to share what they like or interested in with their friends on Twitter.

But we didn't want this to be automatic or mandatory - you should be able to star things without sharing.

Having people share stuff to their friends is also on obvious interest of ours, as we want our users to help us spread the word about Cohaico.

So we wanted the UI for sharing to be at once:

• Eye-catching to have people notice it
• Non-obtrusive to let people who don't want to share go on with their business
• Require little action of the user to hide
• Require little action of the user to use (=share)

What we did is to have a tiny "sharing" box with prepared text float across the screen and "park" for a while on top of the item image, then gradually disappear.

If the user finds it interesting and moves her mouse over it - it stops from disappearing and grows in size to become more usable:

We think it's a good compromise that satisfies all our goals.

Good new brothers and sisters:

You can now watch movie trailers right there on the product page of Cohaico!

Boom, Bam, Kaching.

Peace out.

Imagine this: you're sitting with your buddies, having lunch.

One tells about the movie she watched last night on Netflix. Another tells about the cool new gadget that just came out.

The third is just making weird noises with his soup.

But that's unrelated, back to the stories. This is discovery at its best - discover what's new and cool.

We want to enable the same experience on Cohaico, by processing your Twitter feed and extracting cool new stuff mentioned by your friends, then showing it to you on the home page (and category page) as "Recent in your network".

How do we decide what goes in there and in what order?

Well, it's a bit of marshmallow soup, unicorn farts and pure magic. But the crux of it is:

"Recent in network" is organized by products. So we need to first choose what products to show. We do this by grouping all tweets by your friends by product and taking the 5 most fresh and interesting tweets for each product.

"Interesting" in this case is determined as a combination of:

• How close the friend is to you - this is based on the number of interactions you had on Twitter (replies and mentions)
• Original tweets (as opposed to replies and retweets) are considered more interesting
• Longer tweets are considered more interesting

This "interestingness" is then mixed with the tweet's freshness (how recent it is).

Then, for each product, we sum the "interestingness" of the 5 top tweets by your friends to get the interestingness for the whole product.

We take the 5 most interesting products and we show the 2 most intersting tweets for each.

What's up next?

Once you start visiting Cohaico often (and our Google Analytics is very grateful for those page views) you may start seeing the same products come up time and again. That's because some things (a.k.a. the iPad) are popular all the time. People just can't seem to be able to shut their mouths about them.

What we will do is modify our algorithm to take into account what you already saw on your "recent in network" feed and show you stuff that's popular now but wasn't popular in the past.

What do you think? What kind of things would you like to "discover" on Cohaico?

Bottom line: 4G is the new mobile phone communication standard that is faster than the previous generation - "3G".

However, the current "4G" networks and phones are actually a bunch of different technologies that are not fully implementing 4G but are rather a transitional step towards 4G.

The full 4G standard is called "LTE Advanced". This standard will be finialized this year with networks will become available in the coming years.

Current Distribution

Verizon was first to launch a 4G network, using "LTE" technology. LTE stands for "Long Term Evolution".

WTF? Well, it's part of a bunch of standards created by "3rd Generation Partnership Project" (3GGP). The 3GGP is a work group in charge of mobile communication standards. They're basically specifying a mutually agreed upon protocol that lets carriers (like Verizon) and electronics makers (like Apple or Samsung) create devices that work together. So a Samsung phone it's guaranteed to work on Verizon etc.

So LTE is the name for the standard that's after "3G". But it doesn't fully implement what is called "4G", so technically, it's incorrect to refer to LTE phones as "4G". Confused yet? :)

There's more. AT&T is using the HSPA+ standard as a transitional step towards LTE, and is referring to HSPA+ as "4G".

Sprint is using Mobile WiMAX and branding it as "4G".

The thing to remember is that all of these technologies are faster than 3G but slower than what "true 4G" (LTE Advanced) will be.

Verizon's LTE network is currently available in about 100 cities and 80 airports. AT&T is planning to deploy LTE in 2011 with little further details available.

In Europe, many carriers are currently experimenting with coverage in most countries.

In Japan, the largest carrier NTT Docomo is currently rolling out LTE and it is already available in largest cities.

What Is It Good For?

Bottom line: it's faster. On current LTE networks - download speed of up to 12Mbps and upload of up to 5Mbps.

Supporting Cell Phones

The first "4G" phone in the US was the HTC Thunderbolt, on the Verizon LTE network. Some other devices that came out since:

On AT&T: Motorola Atrix and HTC Inspire.

On T-Mobile: HTC Sensation.

On Sprint: Motorola Photon.

LTE Tablets

Recently, Verizon was the first to launch an LTE tablet - a version of the Samsung Galaxy Tab, and an LTE laptop - an HP Pavilion.

In the near future, many laptops will come equiped with 4G communications technology which means always connected and no more looking for a wireless hotspot.

Hola Amigos!

Today we launched a new feature on Cohaico:

You can subscribe to receive email updates with new products and movies mentioned by your friends on Twitter.

Just sign-in as usual using your Twitter account, then type your email on the home page:

We won't spam you (one email a week, tops!) and you can always unsubscribe with one click.

Have a good weekend!

Hey there. You know what time it is? It's

BETA LAUNCH TIME!

(Hater, you love the colors?)

After working in not-so-stealth mode for the last 3 months we're very (very-very) happy to invite you to try out Cohaico beta.

Why do I need this?

As we talked in a previous post, we all deal with #firstworldproblems like

• What's a good cell phone to buy?
• What's a good comedy to watch?
• What's a great place to go snowboarding in Italy? 

Many times - a quick opinion by a friend is enough.

But who has the time to call 20 friends just to find out who watched "Bridesmaids"?!

Enter The Dragon Cohaico

Let us fix that for you.

Guess what - your friends are already sharing everything they know about stuff. They do it on social networks like Twitter.

But if you get on Twitter - you'll just see what was said in the last few hours or so. It's hard to see what all your friends said about a particular thing.

This is what Cohaico does.

We collect everything that people are saying on Twitter and aggregate it around specific things. Like the iPad 2. Or "Bad Teacher". Or the Canon 7D.

What's Next

We have a ton of cool shit in the pipeline that lets you find and interact with the knowledge that's out there on your social networks, so this is just the tip of the iceberg, but we just couldn't wait to get the product in your hands. So go ahead - check it out and let us know what you think using the "feedback" button.

Two months ago I started my own company for the first time, after "getting ready" for many years.

My preparation consisted of gaining a lot of hands-on experience in software development, management and product design and working as an early employee on two other startups.

Not surprisingly, I learned a ton during the last two months and of course I realized that there's a lot I don't know. I can proudly say I am way out of my comfort zone (as in "you can't see the comfort zone with a Hubble telescope from where I am at") , which is pretty much where I expected to be at this stage.

While there are many little things I realized and most of them I am still digesting, here are three quick observations:

Startup founders are "old" and married

If you follow techcrunch and watch movies, you kinda get the idea that there's a clearly defined "profile" of a startup founder: 20-something male college dropout, hacker, party hard, playboy kind of guy.

Not true.

Yeah, some are like that, but the "common" type is more like 30-something, married (or in a long steady relationship), with kids.

What's going on?

I figured out you have to do your 24/7 startup when you're young, before you have all these other responsibilities.

I missed the other way around: you can do your startup after you got all these other life-things figured out: find a girl, get an apartment, make a baby.

Perhaps it's also easier to do a startup after you've been through a few startups as an employee and have some cash stashed on the side from those other startups.

(Update 2011/7/4: Right after blogging this, I stumbled upon this: "People over 35 have recently launched 80% of startups" http://www.businessinsider.com/people-over-35-made-up-80-percent-of-entrepreneurship-activity-during-the-recession-2011-7)

Execution != Code + Product

I always firmly believed that ideas are 5% and execution is 95%.

I still think that. It's just that I got the 95% part wrong.

I thought that "execution" means great product design and great software development.

In fact, those are maybe 50%.

The rest of it? Sales.

Products that succeed? They are the ones that got the crazy distribution by getting people to use their service, other companies to partner with them etc.

On the outside, this often looks like "destiny" or "force majore". You built such a kick-ass product that people are just magically drawn to it.

Bullshit. Nothing happened by itself. Every successful product came to be because somebody, deliberately, carefully and thoughtfully kicked ass doing sales and did it better than the competition.

This isn't "sales" in the narrow meaning, but I am very comfortable calling it sales, as in "something you do to get people to use your product". This may be 1-step and direct as lemonade or long and widing like a road Paul McCartney likes to sing about, but it's sales all the way baby.

And if you think you kick ass as coder and product guy, please be aware: sales is by far the hardest part.

Traction > Team

Common knowledge in raising funding is that investors are mostly looking for 3 things:

1. Team
2. Market Size
3. Traction

Most advice talks about Team being the number one factor in the decision to invest in a company.

I get the feeling this is not true. These three are definitely the most important ones, but Traction beats Team.

I've seen quite a few cases where a mediocre team with great traction gets funded while a very strong team without traction doesn't.

This makes sense in a time of an investment bubble like now: many teams are first-time founders and shitloads of companies are started.

There aren't enough great "proven" teams out there for all the money looking to be invested.

Traction on the other hand? If you have it - proving is easy.

What are your non-common-knowledge startup realizations?

The web is like drugs. We are "users" and everything we write is "user-generated content" (UGC).

UGC was at the center of "web 2.0" - blogs, reviews on Amazon and IMDB, Flickr, YouTube. That was all the rage in the early 2000's.

Before that - only 1% of the people talked and 99% listened. When web 2.0 came - many more started talking.

And then came Facebook and Twitter. This had 3 consequences:

1. Now practically everybody talks and not just listens.

2. Content became shorter (as did our attention spans).

3. Any written UGC outside of Facebook/Twitter started slowly dying.

Lets look at the last point closer. The first victim of Facebook and Twitter were blogs. Fewer people read them and fewer people write them. Blogs are now consolidating and becoming exactly like traditional forms of media - few are talking, the rest are listening. The opposite of "UGC".

Soon, there will be a second victim - the UGC on various sites that have a UGC platform "on their premises". Today, websites in every "vertical" - shopping, movies, travel, allow users to write comments, reviews and ask questions. The problem is - people are becoming reluctanct to create content on these "3rd party" websites. The reasons are:

1. People don't want to bother with creating another account on another website.

2. People care about who listens to what they're saying. They want the significant people in their lives to read what they write. This "audience" is on Twitter and Facebook. This is where my content will be exposed to people who matter to me. Not on some other website.

And so, UGC on 3rd party websites will "migrate" to Facebook and Twitter. Websites will no longer display UGC that resides in their own databases but rather content that was generated and resides on Facebook/Twitter. Even more: as people still want to "react" wherever they are on the web (not just on Facebook.com) - websites will allow users to submit content directly to Facebook/Twitter from within the 3rd party website. In essence, Facebook/Twitter will change from being "applications" to being ubiquitous "communication channels", used everywhere across the web.

There are signs of this migration already:

1. Facebook provides a commenting system that can be embedded on any website.

2. Part of the deep Facebook integration that's being rolled out on Bing lets you comment on web results - these comments go directly back to your Facebook stream.

The benefit is that relevant and personalized content will now be available everywhere. If my friend says something about a movie on Rotten Tomatoes - I will see it on IMDB. People who read are guaranteed not to miss relevant content. People who write are guaranteed maximum audience.

Tools need to be built to enable this future. These tools will extract relevant content from Facebook/Twitter and make it available in the right context everywhere on the web. They will also easily enable sending content back to Facebook/Twitter from everywhere on the web, with the original context attached to it.

Stay tuned for more on these tools...