stats

Why it doesn’t matter if not all user-created content is great…

January 23, 2012 By Dan Thornton

Youtube users are currently uploading an hour of content every second, or 60 hours every minute.

Assuming 0.25% of all content being uploaded is great content, that’s 3.6 hours of amazing videos every day. That’s 25.2 hours of great content per week, with the average TV viewing in a UK household somewhere between 20 and 30 hours per week.

Pretty amazing, and also why the follow-up attempts to enact laws such as SOPA and PIPA will occur with regularity in the U.S, and the influence of the U.S will be increasingly felt on every country around the world which might be encouraged or persuaded to enact such laws.

It isn’t about piracy. It’s about copying, creating and the disruptive effects we have all had.

Twitter turns 4 years old today!

March 21, 2010 By Dan Thornton

At 8.50pm on March 21st, 2006, Twitter co-founder Jack Dorsey sent his first Tweet to start the microblogging service which currently defines all others.

Along with Evan Williams, Biz Stone, and tens of million people worldwide, Twitter has grown into one the most notable social networks on the planet. Recent stats have pegged online visits at 75 million, content at 1.2 billion tweets per month, and mobile usage up 347% since 2009 with 4.7 million mobile users in January 2010.

We’ve seen Tweets from space and from the American President. Brands have used it to communicate in different ways, and to make money. And as a news and information source, first-hand accounts have included plane crashes, the terrorist attacks in Mumbai, and natural disasters including Haiti and numerous earthquakes around the world.

American forst fires gave rise to the use of hashtags, whilst users also began the now widespread practice of @ messages. Customs have become established such as #followfriday, a huge ecosystem of third-party clients and developers have grown, and third party advertising systems have been allowing users to monetise their content.

Individual users now have millions of followers and have posted tens of thousands of messages.

For a more indepth look at the start of Twitter, check out this post from co-creator Dom Sagolla – How Twitter Was Born. Or check out this post from Jack Dorsey on Flickr.

Got any plans to celebrate #twitterday ?

Why it’s dangerous to compare print figures to website stats

April 14, 2009 By Dan Thornton

Although hardly newspaper/print apologists, both John Duncan and Martin Langeveld have posted interesting articles trying to compare the print/online split in newspaper readership in number terms. Duncan comes in with online having 17% of page impressions on Inksniffer using the Guardian as a case study, while Langeveld posts that only 3% of newspaper reading happens online.

While I totally agree that it’s easy to overestimate the online figures in comparison to print products, and both articles are good reality checks, I have to say that I think comparing print and online readerships directly in this way is equivalent to comparing the number of people who drive cars with the number of people with vowels in their name.

And touting the eventual figures is very dangerous.

For starters, the readership of print titles rests on research figures for average shared readership of titles. For instance, the metrics John Duncan quotes are:

From 2007:

Average daily UK uniques for Guardian website: 270576 (after discounting overseas readers etc).

Average UK sales of Guardian/Observer: 310788

But then the UK sales figures is multiplied by 3 to take into account shared readership, becoming 932,364, on figures available by the Guardian.

Meanwhile Langeveld refers to an engagement study from the Newspaper Association of America conducted in February 2006, based on 4594 respondents to a survey.

Now shared readership definitely happens, and without being able to actually see what people do, rather than what they claim, it’s impossible to be totally accurate.

But…

If you’re taking shared readership of print products into account, then surely you’d also need to factor in people reading newspaper website content without ever being logged as a visitor to the site?

That includes people blocking cookies, people using RSS, people reading reposts of newspaper content (Great example of the spread of multimedia news by Martin Belam by the way), people reading content via aggregation sites and site scrapers etc, etc.

And by the time you’ve taken into account all the vagaries of print readership figures (which aren’t a bad guide to something so difficult to measure), and then taken into account the vagaries of online measurement (Less inaccurate, but still pretty fairly vague), and using data and research from 2+ years ago (But that’s probably the most recent readily available) it starts to be apparent that quoting a an exact figure is pretty irrelevant – especially when some people will undoubtedly take it as gospel.

After all, two years ago, Facebook didn’t have 200 million users, Twitter had just launched, there was no iPhone, there was less broadband penetration in the UK, there hadn’t been events like earthquakes or Mumbai to highlight realtime information, etc, etc.

And there’s a big elephant in the news room: Whoever said that print newspaper readers were guaranteed to only be getting their online news from newspapers?

I can get digital news on my mobile or my PC, via text,audio or video, and via social networks, blogs, websites, link aggregators, RSS, podcasts, videocasts, and from global sources. Whether or not print titles are only seeing a small percentage of their print readership visiting them online is less relevant, than how many of those readers are getting news content online from any source.

So what can you do?

When it comes to looking at the situation now and for the future, the numbers are far less important than looking at data trends. I’d much rather base a theory or business strategy on a few years of data showing a rise in one area and a fall in another. The numbers are rough guides to point towards when the trends are in the same area, but that’s all.

Just to reiterate, I don’t want to criticise John and Martin for doing what is a useful, if flawed, exercise to highlight caution in assuming that online readership is bigger than it really is, or that print readership is smaller than you might think. As I tried to comment on the Nieman Labs site (sadly it vanished into cyberspace after I submitted it), it’s the way the information is being presented that worries me.

Twittertise improves measures for corporate Twitter accounts

October 6, 2008 By Dan Thornton

Twittertise is billed as a method of advertising on Twitter, but personally I see it more as a tool to measure the Return on Investment for corporate Twittering – something which arguably has a lot more value than a basic advertising tool. Owner and creator Jon Steinberg got in touch via Twitter, (@jonsteinberg), but it took me a little while to get some questions to him, and I’m glad it did, because this post now coincides with a new release which answers some of the questions I had about the value of the service.

In it’s simplest terms, Twittertise is a free service which builds on Bit.ly url shortening to offer some statistics on referrals through to your website – but the latest release starts to go further than the Bit.ly tool by offering some useful graphing capabilities for overall referrals and individual messages – and graphs are always useful for corporate reporting!

Jon was kind enough to answer some questions about the thinking behind Twittertise, how brands including Comcast, Nortel and the American Cancer Society are using it, and how you could be using it for your company:

Twittertise and Whalewhisdom are both applications by Thursday LLC – can you tell us a bit more about the company, where it’s based, and how many people are behind the applications?
“Whalewisdom and Twittertise are completely separately. I intermingled a Vimeo account by mistake. I’m an investor in Whalewisdom. Twittertise is wholly owned by me via Thursday LLC. I hired a developer named Gearoid Morley in Canada to build it as a fee for service project.”

Twittertise is powered by Bit.ly – what’s the relationship between the teams behing Bit.ly and Twittertise?

“I just use the bit.ly API. I am friends with the people behind bit.ly and it is a wonderful team and product. But I have no business relationship with them. “

Twittertise offers scheduled posting to allow Tweets to be written far in advance of being published – how do you see this being used for a medium which often leads to conversation?
“For this question, I’d point to this blog post: https://jonsteinberg.com/post/48188357/why-i-created-twittertise-i-had-twittertise. “

The website mentions Comcast as one of the major brands using Twittertise. Presumably this is in addition to human channels like @ComcastCares? Do you see brands using it for more corporate messaging etc, with human channels supplementing it?
“Yes comcastcares has used it. I envision brands using a single account that they feed with both real time tweets through the web or a client and then supplement them with Twittertise. This will enable them to time and track important messages and notifications that require clickthrough tracking. “

Initially Twittertise combined scheduled posts (possibly based on Twitabit?), with the stats available from Bit.ly. Now the latest release has taken things beyond what was readily available with the graphing tools, have you seen a rise in sign-ups and usage?
“We never used twitabit. The scheduling engine was completely built by Gearoid. Today is the first big push with graphing, so I’m hoping that blogs like 140char can get out the word. But we’ve seen pretty steady upflow throughout. I think graphing was a necessary piece the next step in a cleaner UI.”

With the increased economic pressure to show a Return on Investment from time spent using microblogs (and on social media marketing in general), do you expect to see an increase in people using Twittertise to broadcast offers and events without investing as much time in conversation?
“The beauty is that Twitter requires a counterbalance from its corporate users. Corporate users who simply broadcast without responding and engaging in conversation will find themselves with few followers. The right to use Twittertise while simultaneously maintaining your followers is almost something that is earned by a corporate user. The real time conversation tweeting by a brand is what earns it the right to track important communications where you need to show an ROI on Twittertise.”

Twittertise is free to use at present – are you aiming to monetise by developing the service and offering a subscription-type model?
“At this point, I’m honestly just trying to develop the right product for brands. Once we do that, I think revenue will follow.”

What do you see as the next steps/developments for Twittertise
“Improve the UI and continue to try and onboard major brand users.”

Has there been any examples of a hugely successful message or use of Twittertise so far? Anything that has surprised you, or shown a particularly unusual way of utilising the service? Obviously you’ve highlighted some on the Twittertise Blog.
“I think comcastcare’s use during huricane Gustav is my favorite. Timely messages that needed to be spaced.
https://jonsteinberg.com/post/48213746/comcastcares-using-twittertise-to-send-out-gustav
https://jonsteinberg.com/post/48214167/another-shot-of-comcastcares-using-twittertise
Nortel has also been a consistent and solid user for communicating corporate and product related information and tracking it.
https://twitter.com/nortel
On the non-profit side, I’m proud to see American Cancer Society using:
https://twitter.com/americancancer.”

Thanks to Jon for taking the time to answer a few questions, and I’m definitely planning to follow Twittertise and future releases in detail. One of the biggest problems in social media and social networking is tracking a return on the time invested, and this is one of the tools that will start to make that job easier.

It’s also interesting to note that Jon advises using Twittertise in conjunction with real human tweeting – I’m not a big fan of just plugging in an autofeed and letting it run – it could be really useful to ensure important messages don’t get forgotten or phrased badly in the rush of conversation. And it could also be useful to retweet your most important message when you’ve finished posting for the day to ensure people in different timezones might see your most important messages.

If you liked this post and want to keep up with the latest articles, news, Twitter tools and interviews, why not subscribe?

And if you missed our earlier interviews why not catch up with them now?

An interview with Blippr founders Jonathan C and Chris Heard.

An interview with Posty creator Cesare Rocchi.

Factalicious about Twitter in France

July 3, 2008 By Dan Thornton

Twitterfacts

I went back to twitterfacts because it is awash with data about Twitter, and when I look at the numbers closely I come up with different conclusions.

About ten days ago, they were talking about the use of Twitter in France. It is really interesting, particularly the sentence near the end – Twitter doesn’t have a French version – or a German version, or Spanish version, or a Chinese version. Apparently, there is English and Japanese. So now you know.

So how do people in France use Twitter?

The take-up rate of Twitter in France, unlike Norway for example, is quite low. France has a population a little large than the UK, and it has about 6 000 active users – one per 10 000 men, women, children, and grandmothers.

The blog reports most of the data as averages, but it also produces bar charts, and I was able to look at modes and medians. The reason why I would look at modes and medians is that the data is usually skewed. There are lots of people who signed up and don’t use it much, and then a long tail going up to the heavy users.

So lets look at the modal activity

Following

A staggering 16% of tweeters in France are following no one. The modal number of followings, that is the number if you pulled a user out of a hat at random, is 1-4. The median, the point which marks the 50% least active and the 50% most active users, are following 5-9.

So, 1 in 6 are following no one. And 2 in 6 follow between 1 and 9 people. 3 in 6 follow 9 people or more.

Followers

4% of tweeters in France have no followers at all (no bots in France?). That is, 279 tweeters in France subscribed, and no one is following them.

The modal number of followers (pluck someone out of a hat at random) is 1-4 and the median (50:50 mark) is 10.

So the number of followers is higher than the number of people being followed? That makes sense. The difference is made up by bots from abroad.

Tweets

5%, or 1 in 20 users have subscribed and don’t tweet. I suppose they have abandoned the service.

The modal use is less than 9 ( 1 every 3 days) and the median use (the 50:50 mark) is 25. So, the top 50% of users are tweeting more than once a day, and the bottom half are tweeting less than once a day.

I think for this analysis, twitterfacts excluded people who had not tweeted for over a month or who had been members for less than a month. So for people who seemed to be “in”, the above conclusion is true.

Degree of activity

30% had tweeted in the last week. 17% last tweeted between 1 week and 1 month ago. 36% hadn’t tweeted for a month.

Privacy

12% of tweeters in France run an entirely private timeline.

Conclusions

So what do you make of this? Twitterfacts is optimistic – particularly if you are allowed to use French.

Why can’t you use French – surely you just type it? I just did.

PS If you are teaching introduction to statistics, this would be a great exercise on the central tendency.