When numbers become meaningless and dangerous

I’ve just been looking at the latest stats from comScore (Via Techcrunch), and the statistics for Facebook‘s arrival as the fourth biggest site in the world illustrated for me why site stats can become both meaningless and rather dangerous.

For starters, the numbers of the top sites are so big that we don’t really have any way of guaging them – as Eddie Izzard explains using the examples of mass murder (some NSFW swearing).

But the big problem with numbers like these is that they can become very dangerous, due to the tendency for people to quote them as law, and rely on them:

Venus Blindfolded by Gastev on Flickr (CC Licence)

Venus Blindfolded by Gastev on Flickr (CC Licence)

Reasons to worry:

  • Monitoring services like comScore and Compete can only track online traffic to domains – no clients and no apps. A particular problem at the moment for Twitter, but growing for all sites.
  • A certain percentage of users are always unquantifiable thanks to cookie deletion etc, or end up showing up different times on different computers
  • In the comScore example, Google, Microsoft and Yahoo sites are bundled up into their respective companies, so you can’t tell what comes from Google Search and what comes from Orkut or Google Maps.
  • No accounting for OpenID, Facebook Connect etc.

But essentially, the big risk is:

  • Unless you’re one of the top 20 or so sites, the total number of users of Facebook, Myspace, Bebo etc won’t matter much – you’ll still be able to get 1000, 10,000 or 100,000 fans/friends. What really matters is what you want to achieve, the relevancy of the network, and how you work at building valuable relationships rather than numbers.
  • And rather than numbers, look at the interactions, or for the business minded, how many people actually buy something…

But on the bright side:

Besides the fact we can accept social networking reaches almost as far as the internet with Facebook getting 340 million uniques per year, the fact that Wikimedia Foundations sites clock in right behidn it at 303 million uniques also shows the undeniable value of crowdsourcing user generated input if it’s done relatively well.

And maybe the combination of my three blogs will crack the top 10 next year!

Monitor the UK weather via Twitter mash-up

There’s a fantastic Twitter mash-up to capitalise on the UK fascination with weather – even if most of us will only see about an inch of snow at the moment.

#UKsnow Tweets

#UKsnow Tweets

Recommended via @joannejacobs, it works by Twitter users posting a Tweet with the hashtag #uksnow and the appropriate postcode. For example “#uksnow NG9 3/10″

You can see it working by clicking on the image above or visiting the site of the creator, Ben Marsh, but I don’t know how long it will be running for – you could ask him on Twitter at @benmarsh!

Chatting with the man behind Twilert – the new Twitter alert service

Twilert is a new service which aims to bring the ease of Google alerts to the Twitter world. It’s up against Tweetbeep, which has offered alerts for a while, and both offer filtering by location. Twilert does have the edge on flexibility around the selected keywords, specifying both the username of the sender or the recipient of Tweets, and even offering some basic postive/negative attitude reporting.

So it seemed a good idea to find out more about Twilert, and especially how it might differentiate itself further in the future, by speaking to Dan Leach, who is behind the service.

What was the main inspiration for Twilert? Was it something you personally needed for monitoring Twitter in your day job?
The inspiration came about a month ago when I was looking through the Google Alerts I have setup for my various clients (I work in PR and marketing). A lot of the information was dated and I wanted to find a way of seeing what people were really saying and thinking about the brands and products I represent. As an obsessed Twitter fan I wanted to find a way of tracking conversation and opinion on the site without having to sit in front of a Twitter Search page all day. And so, Twilert was born.

What’s the main advantage over other monitoring services?

There are limited options available to people if they want to monitor “tweets”. Aside from the aesthetic differences between Twilerts and its competitors, the two main advantages include:

  • Full customisation of alerts: Twilerts options reflect exactly that of Twitter’s Search service which mean alerts can be tailored by keywords, author, recipient, location, link-location, and attitude (positive, negative, neutral). This means you can filter out irrelevant tweets from your alerts.
  • Ease of use: Twilert doesn’t require you to have a Twitter account, nor does it require any technical knowledge of Twitter or search. True story: to ensure the site was as user-friendly as possible, the test subject I used throughout development was my Mum. If she can use it then anyone can!

You’ve obviously built in some quite specific ways to filter messages: by location proximity, whether they include links, and by positive or negative attitude. How are you calculating the attitude of Tweets?

The attitudes of tweets is calculated by Twitter’s emotion algorithm (created first by Summize) which uses certain phrases and words that suggest a positive, negative or neutral phrase and maps them against keywords in the tweet.
It is by no means 100% accurate and will continue to evolve, however it will provide a decent enough snapshot of whether people are speaking positively or negatively about whatever you are tracking.

With such a comprehensive attempt at filtering is there anything you haven’t been able to include?

We have included everything that Twitters allows through its API. With the limited information that is provided with each tweet (author, recipient, location, content) it is difficult to filter them anymore than is already possible on Twilert.

The site was built by Codegent – if you funded the build, does this mean you have plans to recoup your money by monetising Twilert? Or by utilising the data on popular terms etc?

Monetising the service is a long way in the future – Twilert is less than a week old so our focus is 100% on providing a high quality service to our users. However, we will be exploring extended functionality that could be implemented for enterprise users. It is worth stressing though that the basic alert service will always remain free to users.

Twilert is definitely the weapon of choice for anyone looking for regular automated emails which compile your reports within potentially pretty specific criteria. It’s also one of the better looking 3rd party applications for Twitter, and being designed by someone working in PR and Media, it should be well placed to capitalise on the influx of brands and agencies looking to

I’m not the only one questioning Pepsi’s Unfriendlyfeed

A few days ago I wrote about the efforts by Pepsi and Coca-Cola to engage their consumers in different ways – Pepsi chose social media, whilst Coca-Cola chose a loyalty reward scheme. And yet they are both making mistakes big enough for me to post ‘How Coke and Pepsi are wasting their online strategy‘.

It’s not surprising that other people are also questioning the policy, like Todd Jordan with his post ‘Pepsi – Are you listening‘ (Found when he mentioned it on Twitter – @Tojosan). Interestingly his post appears to have been picked up by Josh Karpf from Pepsi, commenting as Josh.

It’s good that that someone at Pepsi has picked up on the post, but Josh is repeating the same things many people have already heard:

‘We’re listening, Todd–and making efforts to do just what you are saying.’

‘We need to find a better way to aggregate and share fans’ passion for the company.’

‘As for the Friendfeed room, it was never intended to live as a standalone communications platform for PepsiCo. It’s one of many “outposts” we have launched, and intend to launch going forward as part of our ongoing digital plans. We are actively taking in feedback from across the web and starting to join in on conversations outside of the room, which you may have noticed.’

‘We do need to moderate comments to some extent to make sure profanity is removed. However, we do not moderate at all based on things we do or don’t like.’

‘We are going to introduce more new voices into the room from inside and outside the company very soon. I agree that we need to be more engaged in fan-based communities beyond ones that are launched inside the company. You surely understand that this is a first step for us; and we are moving towards becoming far more open, inclusive, and “closer” to our consumers.’

All pleasant enough, and I’m sure Josh is a nice enough guy. And with people like Steve Rubel involved, you’d hope they’ve got an idea of where they can improve.

But it really doesn’t take that much effort to start making improvements right away – like opening the Friendfeed Room up to everyone. I don’t want more voices inside and outside the company. I want all voices by anyone who has anything interesting to say – and that’s what Pepsi should want too!

They can still moderate, either pre, or preferably post comment. But how on earth do they expect to get closer to a community by dictating strict topics for discussion once every few days? It’s like walking up to the community and shout “You will engage, You will engage, You will engage!” over and over and over.

Funnily enough, although the FriendFeed room is quiet, Twitter has 11 messages mentioning Pepsi in the last 20 minutes via Twitter Search. And in the whole front page there’s not a single message containing profanity.

None of page 2 or 3 either. Mainly because people are treating each other like adults!

And look Pepsi! Look at a blog search for Pepsi and Friendfeed!

  1. Pepsi are you listening? by Todd Jordan
  2. How Coke and Pepsi are wasting their online strategy by Me!
  3. Why blogger outreach can fail by Virginia Nussey
  4. Pepsi’s social media challenge by Jason Lee Miller (focusing on the lack of new suggestions generated)
  5. Pepsi asked for my thoughts by CC Chapman

Responding in the comments of the first or second post is fine, but by the third and fourth I’d be looking to make some immediate changes. After all links 5 and 6 were posted on November 12, 2008. Mine was on November 28th, and Todd’s was today. That’s almost a month without any obvious changes.

Maybe it’s time?

Pepsi can by schnaars (CC licence)

Pepsi can by schnaars (CC licence)