Crepuscular Light

But don’t you want to know WHY???

Exxx — Fri, 23 Jul 2010 09:42:39 +0000

So, you have a web analytics tool. You’ve done your research and you’re happy with the solution you’ve chosen. And you’re using your lovely web analytics tool as a campaign tracker. Only. That’s your one use. To see how well your campaigns are performing. But you don’t look at why their performance is at its current level. You just track them so that you have figures for Upstairs. Why?

This is a pet peeve of my own; people using web analytics solutions to track one landing page or to look at one report. Obviously, if that’s what someone would like to do, then more power to their elbow, but the vast amount of data and insight that is possibly being wasted by this under-usage seems like a tragedy.

What you don’t know if you track one landing page:
- what visitors did next (unless you have an exit link on the page to track)
- how to segment based on behaviour
- how to target based on behaviour
- how to improve your acquisition strategy
- WHY

Small-picture tracking is wasting opportunities – granted, this may be a usage dictated by a lack of resources, parcelled web properties, teams in silos and so on, but looking at campaigns in isolation, without looking at other influences and impacts is squandering optimisation prospects.

To be fair, working for an analytics vendor as I do, there is sometimes an in-house disconnect between the usage we had in mind when creating a feature/function/report capability and the actual way in which a customer will use it.
For example, we have a lovely array of out-of-the-box reports of which we’re very proud. We once had a customer ask if we could add an option to remove all of them as they just took up real estate in the UI after they’d created all of the custom reports they actually wanted.
We have a report called Last Visitor Details (shows individual visit paths) that we never considered to be incredibly useful because it can’t be used for trending or statistical analysis. We have customers who ONLY use this report. Therefore, I am aware that my bias may be unfounded and that if any part of a tool’s usefulness can be exploited, there are wins but..but..don’t you want to know WHY?

The Analysis Exchange – making angels smile

Exxx — Thu, 08 Jul 2010 17:15:40 +0000

I recently completed an Analysis Exchange project with Change for Kids and James Phelps. The purpose of the Analysis Exchange is to provide “on-the-job” training for those interested in web analysis while offering free analytical insight to non-profit organisations. Each project is made up of a triad of oozy giving; the student, who does the analysis; the mentor, who provides ongoing feedback and direction; the organisation – which benefits from gaining insight into current online efforts and getting recommendations for optimisation.

And the best part of all? It’s fun!

The process takes two or three weeks and begins with the goals for the project being determined between mentor and organisation. The student then steams ahead with an analysis based on these goals, tapping the mentor for advice and feedback along the way. At the end of the period, the findings are presented to the organisation and then everyone breaks for ice-cream before grading each other on their participation (a highly exciting back-to-school feeling).

Change for Kids works with New York City public schools to provide music, arts and literacy programs to underprivileged children. Colin Smith and Michael Quinzio, who participated in the project on the organisation side were engaged, enthusiastic and open, which made the project tick along very smoothly and James Phelps was always on hand to help.

So, if you’re a non-profit organisation and you want to get valuable insights FOR FREE – take a look at this page on how the Analytics Exchange could work for you.

If you’re a mentor, and you want to help budding analysts and the web analytics industry, then learn more here.

If you are looking for more experience in web analytics while working with good causes and making angels smile, then you need to read this.

Oh, and GIVE CHANGE FOR KIDS LOTS OF MONEY! That is all.

Help Change for Kids to win a grant to create a literacy center

Follow Change for Kids on Twitter

Find Change for Kids on Facebook

Follow James Phelps on Twitter

Is your bookshelf naked?

Exxx — Wed, 23 Jun 2010 15:35:37 +0000

I recently finished reading Social Media Metrics: How to Measure and Optimise your Marketing Investment by Jim Sterne and it’s a thoroughly cracking read. It’s full of useful resources and case studies and observations on companies who have attempted to maximise their marketing potential by using social media (or soshal meedja, as we’d say in Ireland).

Chapters like “Getting the Message – Hearing the Conversation” may be especially important for those who are using social media because they think it’s the budget airline of marketing. I.e. you get the message out and you use the general public as sales reps and support teams (assuming you’re getting it right).

I indulge in some mild-mannered brand analysis and social “listening” regarding the product I work with (Yahoo! Web Analytics) and I have indeed found that being “heard” will assuage the fears or furies of most users of social media. For example, I can recall one Twitter exchange in which a customer was complaining about the user interface being down. Because I was looking out for tweets, comments etc, I was able to reach out and direct them to our status page and explain that we knew about the problem. The customer was delighted – which is great, considering that the problem still existed. But merely being heard and having the frustration answered made that customer feel better supported and less stabby-stabby.
I’ve also noticed that some people react to such reaching out with complete silence. I’m not entirely sure how to take that one. For example, if someone asks a general question and I answer it (in an official capacity from an official account) – it’s sometimes ignored. Is that “Don’t stalk me” or “I only speak to ‘real’ people” or “Well, I shouldn’t have had to ask in the first place”? It doesn’t give me sleepless nights, but I suppose it is human nature to look for a response in a conversation. Any theories?

This book also set me thinking about reputation management where social media is involved. I recently blogged about how much about “you” can be found in a simple search engine search, but what about having your profile hijacked? I’m not talking about having your account hacked and messages being sent out from your email account. Nor am I talking about your Facebook status being updated by someone else – I’m talking about the freedom that exists for others to use your name for a purpose you didn’t intend. For example, what’s to stop someone tweeting “RT @myname I’m not working right now” (or something more damaging)? What’s to stop one of my Facebook “friends” tagging a picture of someone blurry doing drugs with my name and having that appear on the newsfeed of all my “friends” and their 3,000 “friends”? And this can be extended to your company, so if you have a social media strategy, it needs to include vigilance.

Once I finished Social Media Metrics, I was hungry for more and I’m currently reading “Reading Virtual Minds, Volume 1: Science & History” by Joseph Carrabis which has caused me to miss my stop on the bus home twice so far, as I crossly informed the author. This is an unusual and fascinating book in that it discusses the investigation of human behaviour online (with the assertion that online behaviour is not offline behaviour online) and human communication in a technical context. This book had me from the beginning when, in the author’s forward, he discusses presenting to a crowded room from a laptop provided by the organisers. Ascending to the podium, he looks at the laptop and says “Oh-oh. Not Joseph’s laptop. Joseph not familiar. Very scary. Ooh”.

How could you resist reading further?

This book is full of “That’s it EXACTLY!” moments. For example, when describing the ways in which social media communities are created, the question of out-sourced support desks is raised. We’ve all heard the stories about call centres in India being staffed with help technicians who have been taught localised slang and cultural norms. The reason behind this is that even if you are being helped efficiently, your faith in the system and the process will be weakened if you cannot identify in some way with the person on the other end of the line. You send out Signal X, you expect Response Y. If you get Response A, you’re all at sea.

This reminded me of a recent call to my Visa providers when my card had been suspended for unusual online activity (which, funnily enough, was my membership payment to the Web Analytics Association from Hungary!). I’m not a phone person as I like to either engage people face-to-face or via textual exchanges (yes, I’m one of THOSE). However, because I live in Hungary and have limited contact with Irish folk, I happily spent 5 or 10 minutes on the phone to the Irish call center with many’s the merry “begorrah and begosh” because I was so comfortable in what I perceived to be a familiar environment. The rest of my day was spent loving Visa because of that chat with Mary in Dublin.

This is just the tip of the iceberg of enthralling insights in “Reading Virtual Minds”.

So, understand this, your bookshelf is naked without ”Social Media Metrics” and “Reading Virtual Minds”. Cover it up, for goodness sake. Will no-one think of the children?

Who do you think you think you are?

Exxx — Thu, 03 Jun 2010 14:03:42 +0000

Lately, I’ve been playing with the new release of Twitalyzer (serious analytics for Twittererers), which features an intriguing “Personality” report. Twitalyzer uses TweetPsych to assess tweets and create a profile based on social and psychological cues. Fascinating stuff – of course, the question is not

“What does my online behaviour say about me?”
it’s
“What does what I say allow people to perceive about me?”.

Now, I’m one of very few Emer Kirranes out there as far as the internet is concerned. With this in mind, I feel obliged to be very careful about what I share and say online. I’m also vaguely contrary about giving away my information generally – for example, Facebook thinks I’m ten years younger than I am (though, to be fair, so do I), and I suspect that Google thinks I’m male. Having said that, I have noticed that quite a bit of “me” is discernable from a rummage around search engines, a rake through my Tweets and a quick rifle through my blog posts and comments. But perhaps it depends on how hard you look.

I put together the word cloud below out of search engine results from Yahoo!, Google and Bing on “Emer Kirrane” in order to profile the person that emerged (click on the image to see full size).

Apart from the fact that it appears I didn’t exist before my present employment, the persona that materializes is largely professional and betrays little in terms of personal likes, dislikes, hobbies etc (unless of course you take “masquerading” out of context). Yet the fact remains that I can see “me” buried in there. And when I look at my Personality report on Twitalyzer or my profile on PsychTweet, there’s still quite a bit of “me” to be found. It’s controlled, and it’s limited, but I’m in there.

So, if you could combine a psychological profile built out of an overall online personality with your web analytics data, rather than just guessing from browsing behaviour and referrer information, imagine the targeting possibilities! Given that with something like Yahoo! Web Analytics (at the risk of a shameless plug), you can tell the age, gender and inferred interests of visitors (aggregated) if they’re logged into part of the Yahoo! network, it can’t be a dream that’s impossible to dream – can it?

Attribute THIS!

Exxx — Wed, 19 May 2010 14:01:01 +0000

Here’s the scenario – Ben gets up in the morning and switches on the television. He sees an advert on the goodness of milk. The music in the background of the advert reminds him of an old advert from years ago about baked beans. He now wants baked beans. Then he remembers that it’s 6am and he doesn’t eat breakfast so he showers crossly, dresses grumpily and begins his day faceitiously (he’s not a morning person).

At lunchtime, he runs into an old friend – let’s call him Jerry. They pop into a café where Jerry orders a venti soy caramel frappichino extra whip and Ben has a coffee (He’s old-fashioned). When they sit down at a table, Ben notices a poster advertising milk from the same campaign as the advert he saw in the morning. As he has just discovered that Jerry and he have little left in common after all these years, Ben mentions that the milk campaign is using the music from the old baked beans advert from years ago. This leads them to discuss the terrible music (“it’s just noise”) being used in the current baked bean advert.

They don’t stay in the café for long – they both realise that the friendship has run its course.

On the way home from the café, Jerry pops into a shop for a loaf of bread. He spies a tin of baked beans. He remembers the old advert from years ago and smiles. He buys the tin of baked beans. He hates baked beans – but he’ll only remember this later when he discovers that there’s nothing else to eat in the house.

So, the question is – how do you attribute the sale of this tin of baked beans? Who or what should ultimately get credit? Is it the producers of the milk campaign? Is it Ben? Is it whoever came up with the original baked beans campaign that made the link between product and music so enduring? Is it all of the above? And if so, how do you weight one against the other?

If I was monitoring the baked beans campaign, I would no doubt pat myself on the back for the current advert, even though it in no way contributed to this sale – and in fact, as a stand-alone advert, is only hurting the brand.

Traffic attribution can be an art of lateral thinking – how do you do it?

Here comes the science bit – upgraded your YWA tracking code?

Exxx — Wed, 14 Apr 2010 10:54:07 +0000

Do you use Yahoo! Web Analytics? Yes? How clever of you! (me? biased? pfff!)

Have you been using it since the days of IndexTools? Yes? How loyal of you!

Have you upgraded your tracking code to the latest version? No? Not sure? Well, let me help you out a little.

When IndexTools became Yahoo! Web Analytics in 2008 and completed the rebranding process in 2009, the tracking code was overhauled. So, the current version is version 5 and the previous version is version 4. Support for version 4 ends on July 30th 2010, so if you want to continue to benefit from support and updates, then you might want to think about upgrading.

Not sure which version you have? The difference is that Yahoo! now hosts the .js file for tracking and version 5 uses setter functions as opposed to global variables.

I.e. Old looks something like:

So, log in to your account, go to the Installation section and copy the code there to get started, get the Installation Guide and away you go!

Moses’ Ark – Textual Fluency and Boubas

Exxx — Tue, 06 Apr 2010 09:03:07 +0000

My (rather dishy) other half is currently studying for a degree in psychology and issues of the magazine “The Psychologist” regularly arrive on our doorstep. The February copy arrived with an arresting yellow cover and the words “Read this issue. You will find it an engaging, informative and quality read”. I felt compelled to do as I was told and found a fascinating article by Hyunjin Song and Norbert Schwarz.

Consider the following. You’re thinking about taking a holiday this year and you want a destination that will be hot enough for lazing by the sea in May. You have a couple of brochures. You would base your decision on your own general knowledge of the countries represented in the brochures and your specific requirements for the desired holiday, right? What you would not do is base your decision on the font in which the information was provided to you. Surely not. And yet…

Recent experiments have shown that font influences the fluency with which information is processed. The amount of effort required to read can inform the choice of the reader. An experiment similar to the following was run – showing two sets of readers the same text in two different fonts:

When the readers were asked to estimate the time it would take to do this task, most of those presented with the task in the clearer font (in this case Arial) thought it would take almost half the time other group estimated. I.e. the time and effort taken to read the text influenced the readers’ feelings towards the task itself.

Similar experiments were run regarding more commercial subjects. Given two product descriptions, again taking two groups and two fonts, readers were able to make decisions about product choice more easily if they could process the text more easily.

In the internet world, we know that a clearer font means a clearer message, but think about the other risks you take if you use “difficult” fonts for stylistic reasons. You not only muddle your message, you influence the way in which your website visitors understand and view your product, your company, your website processes. Ease of understanding breeds a sense of familiarity. In a shop, this could lead to repeat purchases of a product. Online, this could mean repeat visits.

Here’s another fascinating tid-bit. In their article, Song and Schwartz refer to a study done in 2006 by Daniel Oppenheimer and Adam Alter which noted that companies with easier to pronounce ticker symbols (like KAR) performed better than companies with difficult to pronounce symbols (like RDO) on the first day of trading! The thought was that the more pronounceable symbols seemed less risky.

Moving from font into visuals, have a look at the images below.

If you were told that one was a Bouba and one was a Kiki, which one would you think was which?

Most people would decide that the image on the left was a Kiki and that one the right was a Bouba. This is true across languages and age groups and was first observed by Wolfgang Köhler in 1929. This leads to the suggestion that object naming is not random, which leads to the thought that combining names with “inappropriate” shapes online could cause a subconscious clash for the website visitor.

The majority of people will answer “two” when presented with the question “How many types of each animal did Moses take on the Ark” despite the fact that they know that the Ark was Noah’s, not Moses’. Fluency processing and visual cohesion should not be a nice-to-have in website design, but a fundamental part of ensuring that your website is engaging with people.

The Tale of the Littlest Ninja

Exxx — Wed, 17 Mar 2010 12:33:02 +0000

Last year, as part of a surfeit of general enthusiasm for all things WA-related, I entered the Web Analytics Association (WAA) Championship. This competition is a great idea and allows web analysts and wannabes to get their investigative hats on and get cracking on some analysis of the WAA site.

Not surprisingly, I did not win the grand prize of a free pass to eMetrics or a lovely top-5 badge for this blog, nor did I expect to. I entered for the fun of it and thoroughly enjoyed writing my entry, which you’ll find re-printed here. I did, however, feel that all entrants, bar the two winners were given a bit of a dressing down in the “Fatal Flaws” list (the number of winners was reduced from 5 to 2 as a result of all the shoddy entries). I for one am quite convinced that I come under the “innappropriate tone” category…

Now, before you read on and get all over-heated and upset, please remember that I am Irish and we are a facetious people. I can see how I might come across as scathing, but I just looked for more leadership and community from the WAA and to be honest, I think that the new-look site largely reflects many of the things I advocated in my entry. I hope no-one is insulted, that is not my intention. I’m merely reprinting for fun, frolics and a bit of mischief on St. Patrick’s Day. And you don’t have to list out the flaws to me – this has been done.

The Tale of the Littlest Ninja

Once upon a time, in the land of Analyticadia, there was a village called Measureville. In this village lived a little ninja, who was known to his friends as Little Ninja. Little Ninja was happiest when counting and slicing and dicing and analyzing, which set him apart from his friends, who generally liked to poke things with sticks and spit at lizards. One day, a wandering minstrel came to Measureville with tales of a mighty contest for ninjas everywhere. Little Ninja was excited at the thought of playing and competing with like-minded martial-analysts and immediately set off for the site of the competition – the Castle of the Web Analytics Association.

After many adventures best told over a cold beer, Little Ninja arrived at his destination and set to work using all the tools that had been provided to him. Eventually, bleary-eyed and frankly a little the worse for wear, he emerged from his work hut and went to the Gurus to present his findings.

Little Ninja said: “If you are provided with top-notch free analysis, act on it!”

Little Ninja had noticed that insight gained from previous Web Analytics Association Championship entries had not been acted upon. For example, a Predicta ninja had noted that “some URLs even contain user email information” – if this had been acted upon, the 2009 Championship would not have been delayed – an incident which had caused sadness throughout Analyticadia.
The Web Analytics Association, a mystical organization of world renown, counted some of the most talented and respected ninjas on its membership scroll. Why then, when given recommendations and free analysis, had changes not been implemented? This seemed to weaken the root of the perceived value of membership.
“The Web Analytics Association must be a paragon of preach-practice” cried Little Ninja, “Those who seek wisdom and knowledge should be able to come here and find guidance and leadership!”
Little Ninja paused for breath after this impassioned speech. He noticed one of the Gurus was eyeing him rather warily. Nevertheless, he pressed on with steadfast determination (referred to by his friends as stubborn pig-headedness).

Little Ninja said: “If you know what people want, give it to them!”

Little Ninja had always believed that the mystical words and phrases used by “The Visitors” when employing the magic of a search engine were an important key to their intent and desires. These nomadic folk were not looking for what they knew they would find but for what they wanted to find. They had an expectation in their minds and would keep searching until they found something that answered their needs. Or give up – they were an inherently lazy people, increasingly spellbound by instant gratification.
Little Ninja scratched a graphical representation of the phrases used by The Visitors on the wall of his work hut and noticed that it looked like a creature with a long tail. This “Long Tail” was bringing almost half of The Visitors to the Web Analytics Association’s Castle and those Visitors were largely looking for news and education. However, the rooms of the Castle full of news and education took The Visitors the longest to get to, on average.
“Build better signs to the rooms of Edification and Intelligence”, shrieked Little Ninja, dribbling slightly in his excitement, “The Visitors want knowledge and that is why they come to this Castle! These rooms should have pride of place and the contents should be refreshed more regularly! The Blog Room is hidden deep within the Castle, a scarcely visited jewel! Share the knowledge! Build a map! Let the wishes of The Visitors be satisfied!” At that, Little Ninja fell to the floor in a trembling heap at the feet of the Gurus and had to be revived with some smelling salts and a stiff brandy. One of the Gurus wrinkled his nose disdainfully at this unseemly display.

Little Ninja said: “A ninja’s home should be his castle!”

Little Ninja picked himself up from the floor and grew pale and quaked before the stern glares of the Gurus. But he was determined to go on.
He had noticed that one of the most popular doors to the Castle was the one used by ninjas on the membership scroll who had forgotten their magic password. While gadding about the land of Analyticadia, Little Ninja had discovered that the number of ninjas on the membership scroll had grown 400% between 2005 and 2008 so getting ninjas to enroll was clearly not a problem but they were obviously not seeing the WAA Castle’s member rooms as places to regularly visit, and as a result were often having to be reminded of their magic passwords.
“The Castle should be the first stop of the day for ninjas!” yelped Little Ninja, his eyes crossing with the effort to continue speaking under the bemused gazes of the Gurus. “Ninjas should not see the Castle as just a place to pick up discounts for ninja gatherings or badges for the doors of their own castles. They need reasons to pay the tithe to renew their membership – especially those who can’t make it to the gatherings!” Little Ninja gulped nervously and wondered if he should continue with his recommendation that ninjas have clearer news and collaboration areas in the castle, perhaps allowing endorsed white scrolls from member ninjas and so on. The Gurus were looking a little uneasy at this point, however, so he settled for what he hoped was a reassuring wink. The middle Guru harrumphed crossly.

Little Ninja said: “Knowing is not acting!”

Shuddering violently and wilting before the glowering Gurus, Little Ninja gulped and drew himself up to his full height, which was not the least bit impressive.
“I have nothing but respect for the Web Analytics Association’s castle”, he humble-mumbled, “but the ninjas of Analyticadia get lost there and are not getting the value they should from an amble about the castle grounds”. Completely spent, Little Ninja slid to the floor and closed his eyes. As the life left his body, he thought of the words of Oscar Wilde, and painfully squeaked (in an effort to have wise last words) “No object is so beautiful that, under certain conditions, it will not look ugly”.

If we track them, do they not tremble? Advocating “Vempathy”

Exxx — Mon, 22 Feb 2010 10:23:24 +0000

I’m, naturally enough, a believer in visitor tracking. I think it’s perfectly fair to try to understand the visitor experience in order to work on both improving that experience and to maximise a website’s profit/message potential. When I first began working in the industry and fully understood what kind of data could be collected, I remember finding it quite ominous and unsettling. Now that I know how the data is used, I don’t bat an eyelid. However, I think it is important to remember that the vast majority of visitors we track are those outside the industry – if you put yourself in their shoes, you see a much darker picture of targeting and analysis. But if education is all that stands between distrust and acceptance, why are we not trying to teach? Perhaps we should try a little visitor empathy, or “vempathy”.

I interviewed and surveyed a number of people recently to ascertain how they felt about tracking generally. This was spawned by the fact that when someone asks me what I do (which results in explaining what my company does), I generally find them to be both bored and terrified – a very remarkable combination and one which illustrates the fact that many visitors who claim to distrust tracking will very seldom do anything to counter it. Lethargic horror is probably the most exploited aspect of website visitors.

Most of those I spoke to were shocked to discover what could be learned about them in a single visit and described themselves as suspicious, paranoid and manipulated as a result. I then asked if they would feel differently if this was a site they visited often.
No.
Would they feel differently if they knew that the data was being used to improve their experience on the website?
No.
Would they feel differently if they knew the data would be used to generate tailored advertising?
No.
Would they feel differently if they knew that personally identifiable information (PII) was not being collected and used?
And this was the turning point. There is an assumption that if “I” am being tracked, then “MY” information is being collected. This seems to be the point that needs clarification.

After that, all hell broke loose. Those I spoke to wanted their permission to be sought before tracking was done, but didn’t want a pop-up or to go to a specific page on the site.

The other feeling, from the remainder of the interviewees, was “indifference”, a sense of fairness that if I go to a “free” website, I should “sell something”. Which, I felt, illustrated a belief that “the data is probably used for something sinister, but who cares”? To my mind, this is just as unwelcome a feeling as one of paranoia. It still paints us as the bad guys, but gives us permission to quietly commit our sins.

In Yahoo! Web Analytics, you can see the “interest categories” of your visitors, based on the Yahoo! properties they visit. Eric Peterson, in a fascinating post, has outlined a browser history hack! With tools like these, it is therefore possible to begin building a profile of your visitors as users of the internet, not just as visitors to your website. And this can surely give you incredible insight. But even I, while finding these possibilities dance-in-my-pants exciting, also sometimes find them dance-in-my-pants disturbing. It’s a reflex and I can settle the little voice inside with my knowledge and understanding of the industry. The general site visitor may not have that luxury. If you look up “website tracking” with a search engine, you will see thousands of results from vendors and analysts, but little (outside of from tinfoil-hat-wearing conspiracy theorists) along the lines of “Dear Visitor – here are the reasons, without the fancy blurb, why we track your visit and this is what we use the information for”.

Should we be taking greater responsibility for educating website visitors? Should we work on our “vempathy”?

Playing “Twitter Whispers” with your message and attribution

Exxx — Tue, 19 Jan 2010 15:54:30 +0000

You’ve probably heard this grammar joke (this is the polite version):

A panda walks into a café. He orders a sandwich, eats it, then draws a gun and proceeds to fire it at the other patrons.

‘Why?’ asks the confused, surviving waiter amidst the carnage, as the panda makes towards the exit. The panda produces a badly-punctuated wildlife manual and tosses it over his shoulder.

‘Well, I’m a panda’, he says. ‘Look it up.’

The waiter turns to the relevant entry in the manual and, sure enough, finds an explanation. ‘Panda. Large black-and-white bear-like mammal, native to China. Eats, shoots and leaves.’

Now, I’m a fan of grammatical pedantry generally, but the idea that a message can be so easily transformed by the slightest change is fascinating. For example:

Let’s eat, Daddy!

Let’s eat Daddy!

For want of a comma, this childish entreaty for food becomes something far more sinister.

With this in mind, consider the impact of grammatical changes, truncations, translations and other modifications that happen every day on Twitter when someone “retweets” something you’ve written (if you’re on Twitter). Generally, retweets are considered to be a good thing – you’ve written an article, you state that fact on Twitter, and the news is circulated around the online community. However, outside of “pure” retweets, your message is changed in some way every time it is passed on. This could be positive or negative – there could be no impact – but consider the potential lack of control over your original message.

Here are a couple of examples:

TheRickWilson: Good riddance to bad rubbish RT @LizMair: Laters, Tim Kaine!

MisSMarG3La: Im gonna have to disagree. RT @JMFW_: bacon and waffles > ..eggs

billiferlove: Freakn ridiculous!!!!!! RT @chargers: Chargers have signed Head Coach Norv Turner to a three-year contract extension.

Regardless of the original intent, a comment could add weight and colour to your message. It is possible that @LizMair was bidding a cheerful “Good Day” to Tim Kaine. For all we know, she is acquainted with Tim Kaine and drinks had been arranged for later in the day. But with the addition of “Good riddance to bad rubbish”, one would assume that @LizMair’s message is the same as @TheRickWilson’s (which is possible, but can you be 100% sure if the only thing you are looking at is the above tweet?).
What if that was your message? Perception of what you intended by what you wrote can easily be coloured.

Of course, this can be a source of tremendous promotion.

_artlover: Wow! RT @petapixel Cool nature shot – A roll cloud over Uruguay:http://j.mp/5GXCvA

Carly24Dani: I’d buy it RT @gossipteens: Taylor Swift Planning Her Autobiographyhttp://bit.ly/74NfYz

lilt82: Buy it RT @tifamusic: RT @soshifty: @tifamusic @ward21music Tifa’s new single “If I Could Fly” available on iTunes now: http://bit.ly/5Frlm6

The addition of a comment can make your message more persuasive. If someone I follow closely on Twitter uses a word like “Wow” as a comment on a retweet, the likelihood that I will pay attention is increased. In addition, the habit of trying to attribute a tweet to all previous retweeters can give the impression of reach in a message.

So, I decided to experiment. Yesterday, Eric Peterson posted the following tweet on Twitter:

erictpeterson FREE PDF COPIES of Web Analytics Demystified and my KPI Book for EVERYONE: http://bit.ly/5AleMT #measure

Safe in the knowledge that this would probably get quite a bit of attention in the retweet department, I decided to try to stalk it through the Interweb to see how it would evolve as time progressed. I didn’t expect the message itself to change much, given the “free books for everyone” nature of the tweet, but I was interested in the potential for change and the impact of retweet habits.

Click to see full size

During the first hour following the tweet’s publication, the number of modified tweets quickly exceeded “pure” retweets (i.e. retweets without any change or addition of comment to the original). In addition, the tweet swiftly lost its “#measure” hashtag, which is used by those in the analytics community to tag analytics-related tweets. This was generally caused by the desire of retweeters to add comments to the message, for example “Download Now”, “downloading“, “Cool”, “Great contribution” etc.
While all of this positive reinforcement is wonderful, note that now there are less tagged instances of this tweet so those who simply monitor the tagged tweets stream could miss out.
Some of those who retweet without the hashtag may not have web analytics followers, so the message could stop there.

Within the first hour, the URL leading to the book download page also began to change. This could be a problem if you tag your URLs on the fly for campaign data capture in a web analytics tool. Twitter attribution is notoriously difficult given this particular tendency towards change. Someone clicks on the URL you have provided but then does not include your campaign parameter when they re-introduce the URL into their promotional tweet. Coupled with the loss of author-attribution (in this case, tweets without mention of @erictpeterson), this can also make the tweet tricky to manually track as it moves and changes. Many “direct access / no referring URL” visits could be attributed to Twitter “travel” – this can be difficult to gauge and could theoretically play havoc with your online campaigns.

Within four hours, the #measure hashtag had all but disappeared from non-pure retweets and the rate of URL change increased. The tweet began to appear in foreign languages (paraphrased, not translated), and further instances of tweets with completely changed text and no author attribution began to appear.
Within 24 hours, there was little sign of the original tweet. The original message was sent in a new tweet with different text the next day, which can account for much of the disappearance, but bear in mind the potential implications.

Remember that each time you tweet, you are running the risk of playing Chinese Whispers and having your message distorted by each retweeter. Do you know exactly where your tweets end up when you set them free?