I think there may be a couple of other strong reasons why the last two were clicked so often. "I'm on Twitter" and "Follow me on twitter" are commonly used phrases and read like advertising. On the other hand, "You should follow me on twitter" is interesting. It makes me ask why? I assume that you are promising content as valuable as your blog so I'll follow you and see. Then, "You should follow me on twitter here" is even more interesting. Now I'm asking why, and why there? "Here" is a little mysterious so I click to see if you've got something special going on. That's my interpretation but without seeing how the other similar phrases did it's hard to say.
The context is important too since it appears right after the post, the phrases that look like they are continuations of your post will do better.
pmcelhaney
· 5 months ago
I found this blog because someone tweeted "You should follow me on twitter here: http://bit.ly/UHnWk."
I followed the link because I was curious why someone I'm already following would want me to follow them. I assumed "here" was either a different Twitter account or some kind of Twitter add-on.
I hope you'll try the experiment again, but test the conversion rate after someone clicks the link.
Paul Stamatiou
· 5 months ago
Great experiment and thanks for sharing your findings with us! I'll have to do something like this on my blog.
Could it be possible that merely increasing the length of a sentence increased the number of clicks its last word received? That's the simplest conclusion to be draw from your tests, that it's not the substance of the content but the size it takes up drawing more attention to itself.
What happens if you do this (slightly longer sentence increases click-throughs even more?): You should follow me on twitter by clicking [link]here[/link]
Or this (longer, but link buried): You should follow me on twitter [link]here[/link] to get my latest updates.
webwright
· 5 months ago
+1. Would love to see a value statement with some length:
"I occasionally post cool design nuggets <link>here</link> on Twitter" or somesuch.
dcurtis
· 5 months ago
That's a good idea. I'll try that in the future, once traffic normalizes, although I am pretty sure the results would be tainted now with this article being on the site.
Here are the character counts in each sentence, with the increase in characters over the base sentence. Not the same, but I feel like they're too similar to be ignored. I suspect you might even get a closer correlation if you look at the area of the sentence, as rendered in the browser.
I'm on Twitter 14 Follow me on Twitter 20 +40% You should follow me on Twitter 31 +52% / +121% You should follow me on twitter here 36 +16% / +157%
You could do some tests with font size, as well as with longer but non-imperative sentences, if you wanted to try and rule this out entirely. Google Analytics' Website Optimiser would be a handy way to test all the different permutations.
Noxn Xon
· 5 months ago
I will follow you... Master...
chrispbacon2
· 5 months ago
You have way too much free time my fine feathered friend..
JoseCasquet
· 5 months ago
You've convinced me!! I couldn't help but follow you on twitter!
brandonatgithub
· 5 months ago
I enjoy following you, but I imagine your traffic increased during the measurement period, as well.
Steven Kovar
· 5 months ago
There is a great industry in wordsmithing to encourage greater click-throughs and conversions. You'd be surprised how some of the smallest changes, like replacing a comma with a semi-colon, in something as small as two sentences can drastically change its effectiveness.
I've also heard (this is not first hand information) that linking to something with ambiguous text like 'here' is bad because it can be inaccessible for disabled users who are just jumping through links - they don't know the context of the text unless they read around it. I'm not sure this is true as I have never used JAWS or the like, but it sounds plausible.
I think linking like: "You should _follow me on twitter here_" would be more effective and accessible. Did you try that, Dustin?
k
· 3 months ago
that´s why you should always use "alt" and "title" attributes.
Mike
· 5 months ago
I wonder, did you try any with the word “because” in it, as in “You should follow me because I am awesome”
The book “Psychology of Influence” (great book: http://www.amazon.com/Influence-Psychology-Pers... ) showed that having “because…” greatly increased your chance of persuading the person to do what you want. (Eg, “Can I cut ahead of you because I am in a hurry")
Eythor Westman
· 5 months ago
What's your sample size?
Joe Conyers III
· 5 months ago
This is a great experiment. I really enjoy your style of writing and sense of editorial presentation.
I'm curious how context played with it, I imagine referrers and different articles had various conversion rates. I wonder if that sample size for your 10 articles and countless referrers is big enough. Very interesting study, let us know if you continue to play with it.
ryanpsims
· 5 months ago
Interesting, but like another commenter, my guess is that by increasing the length of the text, you increase the prominence of the message. Making sure it's seen (and then understood). You might have seen a similar increase in click-through by increasing the size of the target (fitt's law).
berkichi
· 5 months ago
Thank you ! :D
A concerned researcher
· 5 months ago
Consider the issue of confounding. Confounding is when another variable is linked to both the potential cause (changes in wording and structure) and outcome (click rate), which actually causes or greatly contributes to the association. If a confounder is present and you did nothing to control for it, you can get a false positive, such that you think your dependent and independent variables are related, but it was really the confounding variable the whole time. So in this example, what other things would cause people to click the link instead of or in addition to your changes? One example is other media attention that occurred at the same time or even worse that snowballed.
An infamous example of this is when researchers thought that yellow fingers caused lung cancer because they forgot to control for that pesky confounding variable - smoking.
Dustin, read about research study design and statistics to improve future work.
Readers, always question how data are produced.
melody
· 5 months ago
Indeed... interesting findings, but hardly quantifiable in the absense of a control.
kurthaug
· 5 months ago
Dustin, don't worry about researching "study design and statistics." While what "a concerned researcher" says is absolutely true, just DOING it is far more important in the world of click-through analysis than STUDYING it. The RESULT is more important than the CAUSE. We're not RESEARCHING the cause of lung cancer here, we're measuring for RESULTS in a highly fluid, almost infinitely multi-variabled, 4 dimensional environment where if you stop to think about it TOO much before acting, you will totally miss the opportunity through overanalysis.
You're being "stealthy" enough that I'll follow you on twitter...<g>
rutkmi
· 5 months ago
I'm with you Kurthaug -- it's not a completely "elegant" test but it was actionable and produced a desired outcome -- more clicks! Just do it Dustin! :)
Peter Rigano
· 5 months ago
I am not a statistician, but based on his description, this sounds like an adequate test design. My reading of his description was that visitors to the site were randomly shown one of the above statements, and the experiment was continued long enough to achieve a sample size of ~5000 for each group.
Given a specific statement ("Follow me on twitter", "I'm on twitter", etc.), each user's choice to click is parametrized by a Bernoulli random variable. Thus, his clickthrough rate, p, is the sample average of these n Bernoulli variables, which, according to the CLT, will be approximately normally distributed with mean p and variance p(1-p)/n [where p=clickthrough rate, and n=sample size]. Confounding isn't an issue in this case, since we have random sampling, i.e. people are randomly assigned to treatment and control groups.
Given that setup, we can use a two-proportion z-test with unequal variances to test against the null hypothesis of no change in clickthrough rate for any pair of statements. Doing this test for the change from "I'm on twitter" to "Follow me on twitter", gives a z-score of about 5.5, which rejects the null hypothesis at the 99% level. The change from "I'm on twitter", to "You should follow me on twitter here" yields a z-score of 14.5, thoroughly rejecting the null again. [Note: Both scores assume sample sizes of 5000, as stated in the post]
More variables would allow you to make even more inferences about clickthrough rates. However, Dustin's simple experiment certainly appears to produce statistically significant results.
Julie Hamwood
· 4 months ago
I think this is a great test with some very eye-catching results, however, I'm not sure if it is an adequate test design. It isn't clear to me from Dustin's text whether all the different "follow-me" statements were equally randomly available throughout the few months of running the test, or if different "follow-me" statements were introduced across the months.
If the "follow-me" statements became more forceful and personally identifiable over the life of the test; and if, at the same time, there was an increase in number of people using twitter and/or existing twitter users increased the number of people that they follow, then Dustin's positive click-through results could just be reflecting positive twitter uptake and following patterns generally.
Either way, I love that Dustin took the time to explore this.
Were these tested in consecutive order? Twitter followers seem to increase non-linearly (at least at times).
Were you tracking clickthroughs, or new followers? Current content of your stream could obviously be another nuisance.
dcurtis
· 5 months ago
They were tested at random. I didn't change the twitter stream at all during the tests. There was no usual traffic. I was counting clickthroughs, not followers.
Bryan Sebastian
· 5 months ago
This was a really cool experiment and the results are very interesting. Thanks for sharing Dustin, great work!!!
larrylegend
· 5 months ago
Hi Dustin,
Like some of the others here, I'm not so sure about your experimental procedure.
Since each change resulted in an improvement, it's possible that the passage of time has more to do with the change in clickthrough than the wording.
For example, as time went by, maybe more people got into Twitter since you started, or maybe, as word spread around about your articles, your rep has grown bigger and people were more eager to follow you. I'd suggest, as a rough way of looking into this, returning temporarily to "I'm on twitter" or "You should follow me on twitter" and see if the clickthrough rate drops back down or not.
Larry
Tim Inman
· 5 months ago
Thanks, Dustin. You should follow me on twitter here.
feniix
· 5 months ago
I think you are missing a lot of variables. For instance, time, you say you ran this test for a couple of hours in the afternoon PST, that is around 2 pm central time and 3 pm eastern. Then, depending on the location and demographic of your test sample you can do another set of different assumptions. That is, people following your blog/site/twitter located zones other than PST have more free time or are more willing to get distracted between 2/3pm and 4/5pm. Of course mine is just a blind shot, but it would be interesting for the sake of "sociological study" to know the order in which you set up the different phrases. Was the context (the article) always the same associated to the different phrasing of your "link to twitter"? I mean, I have no doubts there are factors in the wording and phrasing that change the acceptance rate of different products and contents, otherwise publicity/advertising industry and as well political campaigns/propaganda would not have any to succeed in what they try to achieve but it would be interesting if you could do a better measurement of your test sample.
kurthaug
· 5 months ago
MORE variables you missed, Dustin:
* Throughput slowdown due to uneven global server load over the test period, including TweetWhale sightings across geographies * Cross-tabulation for penetration of high-speed internet against the worldwide traffic correlated with twitter acceptance in the various regions * Browser/platform compatibility issues across both time and space to eliminate any potential bias towards iPhone users using Tweetdeck or VistaPro users on Chrome (without current versions of Adobe plugins, of course).
And MOST importantly,
* The Coriolis effect evident while watching your PPC ad budget swirl down the drain while you conduct Sensitivity Analysis (SA) AND Uncertainty Analysis (UA) to ensure eradication of any Type 1 or Type 2 errors resulting from faulty assumptions about the relevance of each of the myriad possible variables.
Get it?
JUST DO WHAT WORKS.
Carry On!
Myles
· 5 months ago
"Each of the permutations I chose was randomly selected..."
I _think_ that means they were all done at the same time, but each page load would show a different one.
Guest
· 5 months ago
Interesting experiment, and thank you for sharing your results.
To offer a different perspective on this, being commanded to follow somebody wouldn't work for me at all; for myself (and I suspect a lot of other people) a polite request would be the deal-breaker.
I suppose it all hinges on the type of people you are hoping will follow you, and why. Personally I'd rather have a few hundred followers who actually read my tweets and are likely to re-tweet me, than twenty thousand followers who have little or no interest in my product and have just added me in the hope of boosting their numbers. As the saying goes, 'horses for courses'.
Vlad Kaltenieks
· 5 months ago
Another valid point here is that with the last two you have used the wording with „You”, which is an important impact booster, as suggested by recent studies in neuromarketing.
People listen better when someone talks to you instead of at you. This is also supported by the fact that the old brain is self-cantered and egoistical. Which also means that people don’t really care about your product, they care what it can do for them.
I am wondering how much the click through rate would increase if you would include the benefit statement in that sentence.
Joel Bruick
· 5 months ago
Interesting. If this is true, though, then you should get at least a tenfold increase in clickthroughs with "You have absolutely no choice -- at least not if you ever want to see your family again -- but to immediately begin following me on Twitter here."
Anonymous
· 5 months ago
There's a bit of a mismatch between your data and the replacement images. The text is: a. "Follow me on Twitter" 4.70%"
zafarali
· 5 months ago
Interesting! I should follow you on twitter and you should follow me on twitter here http://twitter.com/zafarali
Tom
· 5 months ago
since you called me an average person, i will never follow you !
johan horak
· 5 months ago
You should do more studies sharing them here
Peter
· 5 months ago
Very interesting data about click-throughs - but did you get the same increase in followers? I.e. did the % of people clicking through and deciding to follow you stay constant, or decrease?
kurthaug
· 5 months ago
THAT might be a useful correlation. Click-throughs and sign-ups are 2 differnt calls-to-action.
"You have a 50/50 chance of winning $1,000,000 if you follow me on Twitter here" might have relatively lower click-thoughs, but a higher ratio of sign-ups...
Danielle Touma
· 5 months ago
how many more click throughs did you get after writing the article?
:)
kurthaug
· 5 months ago
This would be an interesting "finger in the wind" metric. Would anyone be surprised if his actual followers increased ten fold after this post? Is anyone cynical enough to think that THIS may have been his biggest play for increasing followers all along?
The nitpicky variables everyone is chattering about on this thread (positive and negative) are probably DWARFED by the traffic and signups directly resulting from just starting the conversation.
Hmmm.... probably something to be learned there...
Suzanne
· 5 months ago
I'd like to have seen the results for a better semantic link: "follow me on Twitter" instead of "here" or "twitter", both of which are somewhat ambiguous, though "here" does partially complete the call to action by providing a specific (though still ambiguous) location.
"Twitter" implies it's the main site, even though the context suggests it may be to your profile. (Suggestions are by their very nature, ambiguous, not specific direction.) "Follow me on Twitter" is a complete action that makes sense to anyone who knows what Twitter is.
(Chris said something very similar.
EricFriedman
· 5 months ago
Great experiment - thanks for sharing this.
Erik
· 5 months ago
The number of people interested in Twitter has changed during the past few months too. This change in sheer numbers of people is likely contributing to changes in your click-through rate and it is difficult to determine exactly what that influence could be.
So, while this experiment is interesting, it's certainly not something I'd use to determine the effectiveness of invitations/commands to follow on Twitter.
stevebrewer
· 5 months ago
Language is important, yes. At Expedia back in the day, when packages launched, we tried a tab that said "Packages", one that said "Vacations" and one that said "Vacation Packages" - the latter increased BOOKINGs ~30%.
Not sure I agree with the conclusion, though. I think 'comprehension was improved' is a more likely explanation than your 'Mind Control' conclusion. One would need to do something a lot more scientific to figure this out definitively.
Obviously it would be quite hard to track the conversion rate, but I'd be interested to see if the landing page makes any difference to the conversion.
Interesting experiment none the less!
Bruno Ribeiro
· 5 months ago
Like Mike (13th comment) I think you should try with 'because'. First giving a strong reason ('because I'll share value information'), then with a weaker argument ('because I'm awsome' like Mike suggested, and then with an unrelated reason ('because I like strawberries' or something). It would be interesting to see if the research made by Langer still holds true in the web.
Been a social psychologist myself I must say that is a nice piece of investigation. Cheers.
Web Design Mumbai
· 5 months ago
Hi Dustin, you did a good analysis but I think you didn't really learn from your own experiment.
I wonder why did you not add the words "here" while providing links in the following sentence at the bottom of your page :)
"You can learn more about me on my life progress page and on my Posterous."
Elsa
· 5 months ago
I don't have Twitter. Never the less after reading the article I almost clicked 'here' in the very last sentence. It was involuntary how I moved to click. Very spooky and powerful.
Will Parker
· 5 months ago
Have you tried putting the call-to-action in a natural location within the sentence instead of using 'here' links? Where possible, I try to use link text that describes the action that will result when the link is clicked.
For example, I'd go with "You should follow me on Twitter." with the link styled in bold with a high-contrast color.
User sees: Shiny, candy-like link text User feels: Hey, I could click that. User thinks: Maybe I should click that … OR if you've done a good setup… Damn, I'll just FOLLOW him<click!>.
The more insidious reasoning behind this comes from the Ferengi Rules of Digital Acquisition :
"Once you've uttered a call to action, never give the customer a chance to think twice about obeying your command."
ileana
· 5 months ago
I am surprised that the link on the word 'here' worked best. After all these years of being told that links have to be on words that convey meaning, this is an interesting find.
In the past, we have also found that the CTA 'click here' generates greater click-through on our interactive banners.
k.h. schmidt
· 5 months ago
funny, i'm more likely not to follow someone if they use that sort of language. my gut reaction is this: don't tell me what i should/n't do! i'll do it if i damn well please.
Pierre Bastien
· 5 months ago
You mentioned hyperlinking on the word "here" improves the clickthrough in scenario 'd'. Hypothesis: that helps because it's clearer you are linking specifically to YOUR twitter page. When your hyperlink is on the word "twitter", people might think you're linking to the main twitter.com site, which is a less valuable link and therefore gets less clickthroughs.. One way to test this: compare "I'm on Twitter" against 'I'm on Twitter here".
fnord
· 5 months ago
Is your twitter account as awesome as your site-design?
Former Rep
· 5 months ago
I love stuff like this. I used to do medical/pharma sales and I was always impressed much improved my response rate was with customers when we called them to task or provided them a callout or action. Thanks for posting
edwenn
· 5 months ago
Interesting stuff. Could you also do a survey somehow on how many people went blind trying to read this page. White on black does for me every time & now I can hardly see the keys on my keyboard (also white black) as your post has temporarily burnt itself into my brain and is overlaying everything I look at.
billig hjemmeside
· 5 months ago
reminds me of the importance of designing layout/buttons in webshops. Nice article
jamescarlos
· 5 months ago
Yes master...
tucker
· 5 months ago
any study on the use of badges instead of wording? It seems to me that man has an undeniable desire to click on buttons when they see one.
Porta
· 5 months ago
No, I wont.
alan36
· 5 months ago
how do you do? Follow me on twitter dont know where im going but i will when I get there.
The mad designer
· 5 months ago
I agree. Sometimes you actually have to force the reader into heading your call to action by directing them specifically. Thanks for the great read.
Christopher Spence
· 5 months ago
This is awesome, thanks for sharing!
Bruce Collier
· 5 months ago
inyeresting test. I wonder how it would have been different if you'd tried "_follow me_ on twitter".
Because 'follow me' is the pertinent action, wheres a hyperlink behind 'twitter' just implies that it's only a link to the twitter homepage.
Jodi Schneider
· 5 months ago
Repetition also makes a difference. So does the context of the post. What order did you use these?
Did you try a randomized experiment, where the text at the bottom of the post varies?
Mike D.
· 5 months ago
Good stuff. Two things:
1. The "here" part is effective because when I see the word "Twitter" underlined, I reflexively think it's just going to take me to the front page of Twitter, which I have no desire to visit. Whereas "here" suggests to me "just click THIS link and you will instantly be following me". That's a big difference.
2. Although I fully agree with your results, I have a problem putting them into use because I *don't* think that most people should follow me on Twitter. If you want to, go ahead, but I feel like if I issue a command like that, I'm making an implied contract with the reader that following me will be worth their time. And it probably won't be... :)
nickmerwin
· 5 months ago
sudo follow me on twitter here
Jim Bob
· 5 months ago
I clicked on follow, actually, after reading 15 posts on the blog, and deciding that it would be a good follow.
M
· 5 months ago
Ordering people to follow you on Twitter really works! It sounds like obey or die!!! :P BTW good experiment!
Sunny
· 5 months ago
I am sorry but I just can't believe the click through rate you're giving here. You cannot have this level of click with human on a text link at the bottom of a page. You must be mis-measuring something. How do you measure click-rate ? How many distinct clicks did you get on each message ? Regards
Fabricio Teixeira
· 5 months ago
Have you tried disccounting natural twitter interest growth during these months? And have you considered people's natural interest growth about what you have to say, after the AA.com episode?
Kalim Fleet
· 5 months ago
followed - "but not because you told ME to with a DIRECT COMMAND appended by a LITERAL CALLOUT" - lol - wait, what?!?
ChazUK
· 5 months ago
Wow, even after reading that once I got to the end of the article I read the twitter link and almost clicked it!
I normally don't even bother reading it.
Connie Malamed
· 5 months ago
Thanks for sharing this. It's powerful. Suggestion: I think in your graph above, you should find another way to depict the increase, because it is cognitively natural for us to interpret a DOWN arrow as a decrease and an UP arrow as increase. As things increase, they pile higher and higher. So your arrow should be flipped around. If the up arrow can't work with what your trying to communicate, then an arrow that runs left to right is also a metaphor that people understand as meaning an increase. Connie Malamed, Author of Visual Language for Designers
dcurtis
· 5 months ago
I thought about this. The reason it points down is because you read down. I felt the flow of the article from top to bottom was more important than the abstract idea that things that increase go up.
Joel Stransky
· 5 months ago
I love to see results like this. As long as it doesn't lead to an extreme. Adobe's bug list has a curt little message that was supposed to increase sign ups but only turned me away.
"If you were logged in you would be able to see more operations. "
did you consider adding a benefit after the directive. For example, i.e.,"you should follow me on twitter here to get great tips on design, etc."
Igor Schwarzmann
· 5 months ago
Although I think your findings are quite fascinating, there is something you didn't take into account: I'm guessing, that the regular readers of your blog are influenced by your invitation to follow you on twitter over time. That said, I don't think that's the only reason you have experienced an increase of the clickthrough rate, but it might be a partial aspect of your numbers.
clussman
· 5 months ago
Any tests where you linked the entire call-to-action? I would expect that to garner higher clickthroughs.
Stephen Tive Pirrie
· 5 months ago
How about "Follow me now bitches!"??
Már
· 5 months ago
Dustin, did you try the sentence "You should _follow me on Twitter_" with the words "follow me on Twitter" as a link text?
If so, how did that sentence+link-text fare?
Már
· 5 months ago
Dustin, did you try the sentence "You should _follow me on Twitter_" with the words "follow me on Twitter" as a link text?
If so, how did that sentence+link-text fare?
Már
· 5 months ago
Dustin, did you try the sentence "You should _follow me on Twitter_" with the words "follow me on Twitter" as a link text?
If so, how did that sentence+link-text fare?
Brian Christiansen
· 5 months ago
At UIE, our research on "the scent of information" (http://bit.ly/4ob42) we've found that links described by 7-12 words is the sweetspot. (We discuss this in a podcast here: http://bit.ly/kvsSR ). So links of a certain length are almost certainly part of your click increases.
Would be really interesting to see if you tracked which users actually follow from each of these different links and see how long they stick around / how much they interact with you and/or retweet. Are users who respond better to more direct guidance different from users who respond to the passive statement?
Penny Chambers
· 5 months ago
you got my "follow" and you made me laugh. well done.
Penny Chambers
· 5 months ago
you got my "follow" and you made me laugh. well done.
dmerms
· 5 months ago
Does anyone know to implement this using google website optimizer? I do this, but then it asks to place code on twitter for conversion. How do you do that?
Eatup Drinkup, Dave
Matt Daniels
· 5 months ago
Loving it.
In jest: have you considered testing black text on white background? Every piece of research that I've read says that reverse text decreases credibility.
Nonetheless--awesome optimization
phoenixseoguru
· 5 months ago
What about?
You should follow me on twitter NOW!
or
Follow me on twitter and win a million pesos, here.
teddyzetterlund
· 5 months ago
I know the article is not so much about the link placement as the phrase itself but I'm not very fond of the fact that this suggests "here"-links. Do you think the statistics would change notably if you were to make "me on twitter" or "follow me on twitter here" to a link instead?
mahalakshmi
· 5 months ago
This indeed is a good observation. After reading your blog I feel it is not only the power of language that made people follow you but also the quality, richness and freshness of content you provide in your blog.
I am going to follow you on twitter now :)... with the belief that you will continue to provide great content. All the best
Craziest Gadgets
· 5 months ago
neat experiment. you should send me cash here.
4psd
· 5 months ago
that's quite interesting
Andy Mabbett
· 5 months ago
Link text should - not least for accessibility - be meaningful and unique on the page. "here" is certainly not the former; and often not the latter. In the sentence "You should follow me on Twitter here", better link text would be "follow me on Twitter here".
[Apologies if this has already been covered - I don't have time to read the 97 preceding comments!]
Name
· 5 months ago
Bleck.
Name
· 5 months ago
Do you find these persuasion techniques shameless; and nothing new, but as old propoganda and evil? I also find this extremely unscientific. I do admit, I am quite skeptical and cynical in most things. I suppose if we as designers desire to earn a paycheck from our clients, we need to expect and help them earn money, and that means getting conversions and click-throughs and all of these other great marketing buzz-words. *sigh* I long for the days of snake-oil salesmen and carpetbaggers; there was such an art to it; but I reckon in retrospect, the Web 2.0 phenomena will be regarded as such decades from now. Simplicity will come back and arrogant self-promotion and greed will fade. Long live art.
Rob O'Daniel
· 5 months ago
It follows that the same would likely hold true for encouraging readers to subscribe to your blog's RSS feed. I'll hafta try rewording the RSS banner in my blog post footer... I guess I'd like to increase Twitter followers as much as the next guy, but I'd be REALLY interested in upping my blog's subscriber count.
I know this can be accomplished via Twitter too, but thusfar, the bumps my blog has seen in Twitter-sourced visitors has not translated to a corresponding uptick in subscribers.
Abounding Media
· 5 months ago
This is a reminder for everyone in design and media that we need to think through every part of the process. We know words are powerful, but it's not until we see a simple experiment like this that we begin considering slight alterations on common items. Thanks for the eye-opener.
Chuck
· 5 months ago
Cheers for sharing your results @dcurtis, fascinating stuff! We've taken your advice on board (http://www.designbyfront.com/about/)... lets see how it goes :)
Loryn Jenkins
· 5 months ago
Dustin, like Chuck, I've copied your wording onto my blog. Although, I think we all would be better off in copying your action—testing—than copying your results.
BTW: Your tweet about monetizing this blog formed the basis for my most recent blog post. The first five hours of answers are captured for posterity: http://loryn.me
Hamant Keval
· 5 months ago
Hi there - Found you through Micheal Campbells Tweet ! Awesome
Great little article. Thanks for shareing. Amazing how even little tweaks make such fantastic improvement in CTR's
Take care Hamant
God
· 5 months ago
it's incredibly sad the lengths you web 2.0 bandwagoners will go just to feel some self-worth. pretty much everything you folks make is just another excuse to design a pretty page without any substantial content (just like your thoughtless boutique redesign of AA.com that you felt could pass for a true enterprise level site design)
how's that for your ego?
Dan Denney
· 5 months ago
To test this a little bit, I simply tweeted it with "You should read this" and a bit.ly link. It doubled or tripled the clicks of my last ten design related tweets... Cool stuff
jared salzano
· 5 months ago
I think it's because "I'm on Twitter" requires the readers to think about following you in the first place, and we aren't given a link.
"Follow me on Twitter" Makes us think about following you, instead of just realizing that you have an account.
"You should follow me on Twitter" is more interesting, and more welcoming, not just advertising your Twitter account.
When you give the link, it only makes it easier, so we don't have to go out of our way to find you, we can just click a link.
And of course, I'm now following you on twitter. You should follow me on Twitter here. ")
Damian Copeland
· 5 months ago
You could possibly improve further with:
"Follow me on Twitter for improved sexual prowess" or "If you don't follow me on Twitter, I'll eat your kids"
Typegeek
· 5 months ago
Great post, great design and great experiment. Thanks for sharing. I have a link at the top of my page and I've just changed the wording as suggested. But perhaps I should put it at the bottom of each post as more of a call to action that can be completed after my visitors have read a post.
Patrick Haney
· 5 months ago
This is all interesting of course, but calling it a "controlled study" is a stretch. You have no idea if the increase in clickthroughs is due to just the text changes you made, or if other factors influenced the number of clicks on your Twitter link. In fact, concluding that the text "You should follow me on twitter here" with the link on the word "here" is the best option for getting a user's attention is just dangerous. It's not an accessible solution, as "here" says nothing about where the link goes.
I find it odd that you don't address the fact that your first bit of text uses a capital "T" on "Twitter", but the other text options stick with a lowercase "t." And what about the fact that this text is within a paragraph and not on its own? Did you try including a Twitter icon at all to see if that got any more attention than just text?
Microcopy is extremely important on the web, and how you write it can definitely affect how a user will react to your website or web application. That seems to be the point of this write-up, but the way you've gone and "proven" it isn't quite as solid as one might think.
Name
· 5 months ago
You should better say “You should follow <me on twitter>”. “Here” has the same meaningless message as “click here”.
I didnt read all the comments - for me the line height is much too narrow.
But I recomment linking the complete line - maybe this will give the click through rate another boost.
imtired
· 4 months ago
Fuck Twitter.
Matt
· 4 months ago
Great post - interesting experiment. Do you think that this can be translated to call to actio buttons that get people to sign up to things? I guess I can only try and see what happens. Thanks for the ideas.
J_P
· 4 months ago
Have you considered the growth in Twitter use over the period of time you recorded these percentages? Any chance of comparing clickthrough rate with a friend who has stats from the same period of time?
Mário Rodrigues
· 4 months ago
Well this is dam interesting because since im a web entrepreneur depending in CTR to generate as much money as i can i really find this interesting because YES there is an underlying pattern of human behavior that we most of the times don't even realize since preschool we are tough how to obey we are born to obey and follow rules thats how 93% of us work. And it would seem only natural for us to follow unconsciously a command instead of a request, i believe this technique if used properly and studied like you did could become a trend, or a curse if unwisely used.
And although most of us don't like to admit we were interested somehow in that sentence, and give banal excuses like " ohhh i just clicked because in twitter blablabla or i though it could have something different" the thing is you CLICKED. it worked.
5 star experiment.
ranndino
· 4 months ago
It is a well known fact (that can be observed all around us) that people who tell others what to do in a way that assumes authority get what they want. Self centered jerks get all the hot girls and are successful in business. Nice guys really do finish last.
Maz
· 4 months ago
I think you missed a chance to increase usability by increasing the number of words included in the hyperlink. I'd be curious, for example to see how the following faired:
"You should follow me on twitter."
I still think using "here" as your link is a usability nightmare, particularly when it's employed in multiple places throughout a page. It makes scanning links exceptionally difficult. I would conjecture that the reason it worked in this case is because of the isolation of the sentence and link.
naidle
· 4 months ago
Thanks for putting this simple, yet effective experiment together.
Derrek Pearson
· 4 months ago
Please, please, please increase the leading in your comments. It's hard enough to read white on black but to cram it together on top of that is painful. Great article.
meaghano
· 3 months ago
Ha. Dustin Curtis = social media expert.
That is really interesting. I think it's because it sounds more conversational and confident, whereas, Follow me on Twitter is on every goddamned person's blog.
PCM
· 3 months ago
So this is your useful post, that obliges us to "follow you on Twitter"?
Nice graphics though. Much more interesting than the text.
patrickoneil
· 3 months ago
Make "follow me" the link--delete "here". You will see more success.
Name
· 3 months ago
You should lick my balls here.
Mike Caskey
· 3 months ago
I think the main points are 1. You're testing, and 2. You're curious
This is what rocks about the Internet as a medium. It's interactive, and everything can be measured. When you run a split test, it's like asking your users a question, and having them answer (most truthfully) with their actions (usually clicks).
One question I would ask in your next experiment is whether the "here" effect was usability-related, or caused by another linguistic or psychological property.
What I mean is, when I'm scanning a page, I look for links (without thinking about this) and seeing "here" (with the visual emphasis of link treatment) at the end of a sentence, makes me wonder what's going on... and of course in order to find out, I have to read the sentence.
My point is, you might just be making more people aware of your call to action, by way of forcing interest, and this might not be usability-related at all. Definitely something to test, IMO.
BTW, awesome page. I mean... fantastic stuff here! The graphics, layout, typography, content... it's all there.
Mike Caskey
· 3 months ago
Sweet article. Have you set up your experiments to measure interaction between your sentence segments? If not, I recommend trying on a multivariate split test, where all possible combinations of each of your variables is served up for each of your visitors, and the effects are measured. I like that your experiment was all based on words. Have you also tried layout, color, contrast, or other design variants? Also, I question whether the "here" effect was usability-related... when it might just be another effect of sentence structure + word emphasis, resulting in the creation of a question your user's mind as they skim the regions of your page... seeing "here" at the end of the sentence makes me wonder what it is, so I'm forced to go back and read the rest of the sentence.
Gagan Diesh
· 2 months ago
I would be curious to get a bit more quantitative info on the total number of people coming to your site--percentages don't really reveal a true picture unless the sample size is both large, and diverse enough to arrive at logical conclusions.
venkat
· 2 months ago
Great analysis will use follow me on twitter here under my blog posts
paul_carey
· 2 months ago
Fascinating stuff, but I can't help but wonder that the choice of copy would have resulted in a different bunch of people signing up.
Specifically, I believe some people would respond negatively to being told what to do, mentally replying 'I'll be the judge of that' when presented with 'You should...'.
I can imagine a scenario where a higher total signup rate would result in a lower actual signup rate of the type of people you're interested in.
Mike Caskey
· 2 months ago
paul_carey, you've opened up a whole 'nother dimension of analytics/optimization! The same pattern of thought applies to sem/seo, ppc, and other ways of generating traffic, so it applies to the pre-conversion conversation too!
AnkitaaSinghvi
· 2 months ago
You really should follow me on my twitter; which is AnkitaaSinghvi You canbe my little stalker :O Go on. I know you want to.
Mario
· 2 months ago
You could put the link in all the text, not just in "Twitter", change the link color to blue - frequently associated with links -, and remove "click here". I've noted that the link with light gray is not that clear.
Webdesign Berlin
· 2 months ago
Nice experiment. But for SEO is linking "here" not the best way. But especially for Adwords landing pages it's a great tip.
Name
· 2 months ago
Maybe: "Follow me on twitter here."
jonjo
· 1 month ago
How many visitors did you test each of these options on?
Toby
· 1 month ago
Making the sentence much longer (and therefore more visually prominent) increased clickthrough? Did you try taking this to it's logical conclusion and removing all other elements on the page, like the article itself? I find 1000pt red bold type wrapped in the blink tag helps too!
There is a point to my facetiousness - while these effects are interesting in isolation, and useful to know about, they are just a small part of the workings of the total design. Dustin, I'm sure you are well aware of this, I'm just concerned that a certain type of person may read this and deduce that 'longer sentences = better clickthrough, THEREFORE lengthening all sentences on my website is purely beneficial'. This is not the logical deduction
Alan Taylor
· 1 month ago
This is a great, quick study which highlights what many people know but forget: words count in many ways. I wonder whether capitalising the call to action has any effect?
Создание сайта
· 1 month ago
I love it!
Charlieq
· 1 month ago
Will your tweets hurt my eyes as much as this page?
JJenZz
· 1 month ago
Not sure if someone has already suggested this but did you try the phrase 'You should follow me on twitter' with the words 'follow me' as the link, instead of the word 'twitter'?
Linking the word 'twitter' may suggest that it just links through to twitter, or a page defining twitter rather than your specific twitter page. *shrug*.
Very surprised at how much clickthroughs were increased by adding the word 'here'. Great read!
mckelvaney
· 1 month ago
Is this not a really 90's way of displaying links? Thats like putting on a site, 'Look at our great prices <link>here</link>.' It is moving the link away from the context. Personally, I don't know if I would want people following me who click on links with the anchor being 'here'. Despite what the metrics say.
dvc
· 3 weeks ago
Stupidest. Thing. I've. Read.
Ever.
easilyamused
· 1 month ago
These are great points, It also shows a great call to action. Users like to be told what to do next.
Interesting. I wonder if adding a similar command would work...
You should give me money on paypal here.
François Cassin
· 3 weeks ago
Hi,
First I'd like to congratulate you for the great article and experiment. However I was wondering if you checked the growth in conversion rates versus the growth of twitter user during the time it took you to make this experiment ?
If the user growth is superior or equal to your conversion rate growth over the same period of time maybe it just means that more readers were on twitter so more users decided to follow you, regardless of the wording you used ?
nice and useful study! your rates are based on which scale of clicks? hundred? thousand? do you think it is only related to twitter or it can be generalized to all kind of links or social platforms like: you should follow me on slideshare here: http://www.slideshare.net/ereteog ?
Super Carly
· 3 weeks ago
I really like reading others' experiments, and I lurrrve testing stuff like this!
I read in Maria Veloso's web copy book that writing 'because' will increase positive responses. Fact. It's been tested.
The pretty well known story:
'Can I cut in line because I need to use the photocopier.'
(well....so did everyone else in the queue!!)
I'd love to see you test this again with 'you should follow me on Twitter, because xxx.'
I think I'll go for 'you should follow me on Twitter, because I am super.' Fits nicely with my moniker.
I have seen follow me on twitter but the you should part enticed me. Now how does this work with jobs? Hire me! You should Hire Me! You should Hire Me Today! I like it and might use that at some point in the future. Thanks for the great ideas. I will be following you on twitter
gelbendorf
· 3 weeks ago
but of course... the commanding firm call for action works very well, a picture of a good looking woman that reflect authority would probably improve the rate, but... My personal reaction was "who the ____ you think you are telling me what to do...". But I still follow you on twitter, you are a witty talented kid.
eliburford
· 1 week ago
Stop hassling me.
I just started following you before I read this article ...
Thanks for doing the research and publishing it!
IwasFramed
· 1 week ago
I dare you to use popscreen.com to manage your online videos...but you won't of course. just a sample of my Olivia Munn videos http://www.popscreen.com/pl/837/Olivia_Munn Damn I love your posts...why don't you offer me some advice on UX for popscreen, I'm all ears.
All Comments
"Don't follow me on Twitter"
"Don't you dare follow me on Twitter"
"Ok, I dare you to follow me on Twitter"
"I double dog dare you to follow on Twitter"
You shouldn't follow me on Twitter http:/twitter.com/tweeaks
The context is important too since it appears right after the post, the phrases that look like they are continuations of your post will do better.
I followed the link because I was curious why someone I'm already following would want me to follow them. I assumed "here" was either a different Twitter account or some kind of Twitter add-on.
I hope you'll try the experiment again, but test the conversion rate after someone clicks the link.
You should follow me on twitter here: http://twitter.com/stammy :-)
What happens if you do this (slightly longer sentence increases click-throughs even more?):
You should follow me on twitter by clicking [link]here[/link]
Or this (longer, but link buried):
You should follow me on twitter [link]here[/link] to get my latest updates.
"I occasionally post cool design nuggets <link>here</link> on Twitter" or somesuch.
Here are the character counts in each sentence, with the increase in characters over the base sentence. Not the same, but I feel like they're too similar to be ignored. I suspect you might even get a closer correlation if you look at the area of the sentence, as rendered in the browser.
I'm on Twitter 14
Follow me on Twitter 20 +40%
You should follow me on Twitter 31 +52% / +121%
You should follow me on twitter here 36 +16% / +157%
You could do some tests with font size, as well as with longer but non-imperative sentences, if you wanted to try and rule this out entirely. Google Analytics' Website Optimiser would be a handy way to test all the different permutations.
I think linking like: "You should _follow me on twitter here_" would be more effective and accessible. Did you try that, Dustin?
The book “Psychology of Influence” (great book: http://www.amazon.com/Influence-Psychology-Pers... )
showed that having “because…” greatly increased your chance of persuading the person to do what you want. (Eg, “Can I cut ahead of you because I am in a hurry")
I'm curious how context played with it, I imagine referrers and different articles had various conversion rates. I wonder if that sample size for your 10 articles and countless referrers is big enough. Very interesting study, let us know if you continue to play with it.
An infamous example of this is when researchers thought that yellow fingers caused lung cancer because they forgot to control for that pesky confounding variable - smoking.
Dustin, read about research study design and statistics to improve future work.
Readers, always question how data are produced.
Be a NINJA, not a SAMURAI!<g> http://www.shurikensystems.com/page14/page5/pag...
You're being "stealthy" enough that I'll follow you on twitter...<g>
Given a specific statement ("Follow me on twitter", "I'm on twitter", etc.), each user's choice to click is parametrized by a Bernoulli random variable. Thus, his clickthrough rate, p, is the sample average of these n Bernoulli variables, which, according to the CLT, will be approximately normally distributed with mean p and variance p(1-p)/n [where p=clickthrough rate, and n=sample size]. Confounding isn't an issue in this case, since we have random sampling, i.e. people are randomly assigned to treatment and control groups.
Given that setup, we can use a two-proportion z-test with unequal variances to test against the null hypothesis of no change in clickthrough rate for any pair of statements. Doing this test for the change from "I'm on twitter" to "Follow me on twitter", gives a z-score of about 5.5, which rejects the null hypothesis at the 99% level. The change from "I'm on twitter", to "You should follow me on twitter here" yields a z-score of 14.5, thoroughly rejecting the null again. [Note: Both scores assume sample sizes of 5000, as stated in the post]
More variables would allow you to make even more inferences about clickthrough rates. However, Dustin's simple experiment certainly appears to produce statistically significant results.
If the "follow-me" statements became more forceful and personally identifiable over the life of the test; and if, at the same time, there was an increase in number of people using twitter and/or existing twitter users increased the number of people that they follow, then Dustin's positive click-through results could just be reflecting positive twitter uptake and following patterns generally.
Either way, I love that Dustin took the time to explore this.
And you should follow me on twitter http://twitter.com/juliehamwood
Were you tracking clickthroughs, or new followers? Current content of your stream could obviously be another nuisance.
Like some of the others here, I'm not so sure about your experimental procedure.
Since each change resulted in an improvement, it's possible that the passage of time has more to do with the change in clickthrough than the wording.
For example, as time went by, maybe more people got into Twitter since you started, or maybe, as word spread around about your articles, your rep has grown bigger and people were more eager to follow you. I'd suggest, as a rough way of looking into this, returning temporarily to "I'm on twitter" or "You should follow me on twitter" and see if the clickthrough rate drops back down or not.
Larry
For instance, time, you say you ran this test for a couple of hours in the afternoon PST, that is around 2 pm central time and 3 pm eastern.
Then, depending on the location and demographic of your test sample you can do another set of different assumptions.
That is, people following your blog/site/twitter located zones other than PST have more free time or are more willing to get distracted between 2/3pm and 4/5pm.
Of course mine is just a blind shot, but it would be interesting for the sake of "sociological study" to know the order in which you set up the different phrases.
Was the context (the article) always the same associated to the different phrasing of your "link to twitter"?
I mean, I have no doubts there are factors in the wording and phrasing that change the acceptance rate of different products and contents, otherwise publicity/advertising industry and as well political campaigns/propaganda would not have any to succeed in what they try to achieve but it would be interesting if you could do a better measurement of your test sample.
* Throughput slowdown due to uneven global server load over the test period, including TweetWhale sightings across geographies
* Cross-tabulation for penetration of high-speed internet against the worldwide traffic correlated with twitter acceptance in the various regions
* Browser/platform compatibility issues across both time and space to eliminate any potential bias towards iPhone users using Tweetdeck or VistaPro users on Chrome (without current versions of Adobe plugins, of course).
And MOST importantly,
* The Coriolis effect evident while watching your PPC ad budget swirl down the drain while you conduct Sensitivity Analysis (SA) AND Uncertainty Analysis (UA) to ensure eradication of any Type 1 or Type 2 errors resulting from faulty assumptions about the relevance of each of the myriad possible variables.
Get it?
JUST DO WHAT WORKS.
Carry On!
I _think_ that means they were all done at the same time, but each page load would show a different one.
To offer a different perspective on this, being commanded to follow somebody wouldn't work for me at all; for myself (and I suspect a lot of other people) a polite request would be the deal-breaker.
I suppose it all hinges on the type of people you are hoping will follow you, and why. Personally I'd rather have a few hundred followers who actually read my tweets and are likely to re-tweet me, than twenty thousand followers who have little or no interest in my product and have just added me in the hope of boosting their numbers. As the saying goes, 'horses for courses'.
People listen better when someone talks to you instead of at you. This is also supported by the fact that the old brain is self-cantered and egoistical. Which also means that people don’t really care about your product, they care what it can do for them.
I am wondering how much the click through rate would increase if you would include the benefit statement in that sentence.
and you should follow me on twitter here http://twitter.com/zafarali
"You have a 50/50 chance of winning $1,000,000 if you follow me on Twitter here" might have relatively lower click-thoughs, but a higher ratio of sign-ups...
:)
The nitpicky variables everyone is chattering about on this thread (positive and negative) are probably DWARFED by the traffic and signups directly resulting from just starting the conversation.
Hmmm.... probably something to be learned there...
"Twitter" implies it's the main site, even though the context suggests it may be to your profile. (Suggestions are by their very nature, ambiguous, not specific direction.) "Follow me on Twitter" is a complete action that makes sense to anyone who knows what Twitter is.
(Chris said something very similar.
So, while this experiment is interesting, it's certainly not something I'd use to determine the effectiveness of invitations/commands to follow on Twitter.
Not sure I agree with the conclusion, though. I think 'comprehension was improved' is a more likely explanation than your 'Mind Control' conclusion. One would need to do something a lot more scientific to figure this out definitively.
Obviously it would be quite hard to track the conversion rate, but I'd be interested to see if the landing page makes any difference to the conversion.
Interesting experiment none the less!
Been a social psychologist myself I must say that is a nice piece of investigation. Cheers.
I wonder why did you not add the words "here" while providing links in the following sentence at the bottom of your page :)
"You can learn more about me on my life progress page and on my Posterous."
For example, I'd go with "You should follow me on Twitter." with the link styled in bold with a high-contrast color.
User sees: Shiny, candy-like link text
User feels: Hey, I could click that.
User thinks: Maybe I should click that
… OR if you've done a good setup…
Damn, I'll just FOLLOW him<click!>.
The more insidious reasoning behind this comes from the Ferengi Rules of Digital Acquisition :
"Once you've uttered a call to action, never give the customer a chance to think twice about obeying your command."
In the past, we have also found that the CTA 'click here' generates greater click-through on our interactive banners.
Because 'follow me' is the pertinent action, wheres a hyperlink behind 'twitter' just implies that it's only a link to the twitter homepage.
Did you try a randomized experiment, where the text at the bottom of the post varies?
1. The "here" part is effective because when I see the word "Twitter" underlined, I reflexively think it's just going to take me to the front page of Twitter, which I have no desire to visit. Whereas "here" suggests to me "just click THIS link and you will instantly be following me". That's a big difference.
2. Although I fully agree with your results, I have a problem putting them into use because I *don't* think that most people should follow me on Twitter. If you want to, go ahead, but I feel like if I issue a command like that, I'm making an implied contract with the reader that following me will be worth their time. And it probably won't be... :)
BTW good experiment!
You cannot have this level of click with human on a text link at the bottom of a page. You must be mis-measuring something.
How do you measure click-rate ? How many distinct clicks did you get on each message ?
Regards
And have you considered people's natural interest growth about what you have to say, after the AA.com episode?
I normally don't even bother reading it.
Connie Malamed, Author of Visual Language for Designers
"If you were logged in you would be able to see more operations. "
see for yourself.
http://bugs.adobe.com/jira/browse/FP-1976
If so, how did that sentence+link-text fare?
If so, how did that sentence+link-text fare?
If so, how did that sentence+link-text fare?
Interesting experiment, thanks for sharing.
Eatup Drinkup,
Dave
In jest: have you considered testing black text on white background? Every piece of research that I've read says that reverse text decreases credibility.
Nonetheless--awesome optimization
You should follow me on twitter NOW!
or
Follow me on twitter and win a million pesos, here.
I am going to follow you on twitter now :)... with the belief that you will continue to provide great content. All the best
[Apologies if this has already been covered - I don't have time to read the 97 preceding comments!]
I know this can be accomplished via Twitter too, but thusfar, the bumps my blog has seen in Twitter-sourced visitors has not translated to a corresponding uptick in subscribers.
BTW: Your tweet about monetizing this blog formed the basis for my most recent blog post. The first five hours of answers are captured for posterity: http://loryn.me
Great little article. Thanks for shareing. Amazing how even little tweaks make such fantastic improvement in CTR's
Take care
Hamant
how's that for your ego?
"Follow me on Twitter" Makes us think about following you, instead of just realizing that you have an account.
"You should follow me on Twitter" is more interesting, and more welcoming, not just advertising your Twitter account.
When you give the link, it only makes it easier, so we don't have to go out of our way to find you, we can just click a link.
And of course, I'm now following you on twitter. You should follow me on Twitter here. ")
"Follow me on Twitter for improved sexual prowess"
or
"If you don't follow me on Twitter, I'll eat your kids"
I find it odd that you don't address the fact that your first bit of text uses a capital "T" on "Twitter", but the other text options stick with a lowercase "t." And what about the fact that this text is within a paragraph and not on its own? Did you try including a Twitter icon at all to see if that got any more attention than just text?
Microcopy is extremely important on the web, and how you write it can definitely affect how a user will react to your website or web application. That seems to be the point of this write-up, but the way you've gone and "proven" it isn't quite as solid as one might think.
But I recomment linking the complete line - maybe this will give the click through rate another boost.
And although most of us don't like to admit we were interested somehow in that sentence, and give banal excuses like " ohhh i just clicked because in twitter blablabla or i though it could have something different" the thing is you CLICKED. it worked.
5 star experiment.
"You should follow me on twitter."
I still think using "here" as your link is a usability nightmare, particularly when it's employed in multiple places throughout a page. It makes scanning links exceptionally difficult. I would conjecture that the reason it worked in this case is because of the isolation of the sentence and link.
That is really interesting. I think it's because it sounds more conversational and confident, whereas, Follow me on Twitter is on every goddamned person's blog.
Nice graphics though. Much more interesting than the text.
This is what rocks about the Internet as a medium. It's interactive, and everything can be measured. When you run a split test, it's like asking your users a question, and having them answer (most truthfully) with their actions (usually clicks).
One question I would ask in your next experiment is whether the "here" effect was usability-related, or caused by another linguistic or psychological property.
What I mean is, when I'm scanning a page, I look for links (without thinking about this) and seeing "here" (with the visual emphasis of link treatment) at the end of a sentence, makes me wonder what's going on... and of course in order to find out, I have to read the sentence.
My point is, you might just be making more people aware of your call to action, by way of forcing interest, and this might not be usability-related at all. Definitely something to test, IMO.
BTW, awesome page. I mean... fantastic stuff here! The graphics, layout, typography, content... it's all there.
Specifically, I believe some people would respond negatively to being told what to do, mentally replying 'I'll be the judge of that' when presented with 'You should...'.
I can imagine a scenario where a higher total signup rate would result in a lower actual signup rate of the type of people you're interested in.
You canbe my little stalker :O
Go on. I know you want to.
Did you try taking this to it's logical conclusion and removing all other elements on the page, like the article itself?
I find 1000pt red bold type wrapped in the blink tag helps too!
There is a point to my facetiousness - while these effects are interesting in isolation, and useful to know about, they are just a small part of the workings of the total design.
Dustin, I'm sure you are well aware of this, I'm just concerned that a certain type of person may read this and deduce that 'longer sentences = better clickthrough, THEREFORE lengthening all sentences on my website is purely beneficial'. This is not the logical deduction
Linking the word 'twitter' may suggest that it just links through to twitter, or a page defining twitter rather than your specific twitter page. *shrug*.
Very surprised at how much clickthroughs were increased by adding the word 'here'. Great read!
Thats like putting on a site, 'Look at our great prices <link>here</link>.'
It is moving the link away from the context.
Personally, I don't know if I would want people following me who click on links with the anchor being 'here'.
Despite what the metrics say.
Ever.
You should give me money on paypal here.
First I'd like to congratulate you for the great article and experiment.
However I was wondering if you checked the growth in conversion rates versus the growth of twitter user during the time it took you to make this experiment ?
If the user growth is superior or equal to your conversion rate growth over the same period of time maybe it just means that more readers were on twitter so more users decided to follow you, regardless of the wording you used ?
Just wondering :)
And my blog: http://lateral-adventure.blogspot.com/
your rates are based on which scale of clicks? hundred? thousand?
do you think it is only related to twitter or it can be generalized to all kind of links or social platforms like:
you should follow me on slideshare here: http://www.slideshare.net/ereteog ?
I read in Maria Veloso's web copy book that writing 'because' will increase positive responses. Fact. It's been tested.
The pretty well known story:
'Can I cut in line because I need to use the photocopier.'
(well....so did everyone else in the queue!!)
I'd love to see you test this again with 'you should follow me on Twitter, because xxx.'
I think I'll go for 'you should follow me on Twitter, because I am super.' Fits nicely with my moniker.
Peace all
My personal reaction was "who the ____ you think you are telling me what to do...".
But I still follow you on twitter, you are a witty talented kid.
I just started following you before I read this article ...
Thanks for doing the research and publishing it!
just a sample of my Olivia Munn videos http://www.popscreen.com/pl/837/Olivia_Munn
Damn I love your posts...why don't you offer me some advice on UX for popscreen, I'm all ears.