You should follow me on Twitter

http://dustincurtis.com/you_should_follow_me_on_twitter.html

by tdavis • 17 years ago

263 71 17 years ago

dustincurtis.com

alexgartrell• 17 years ago

I wonder if it wasn't, in some way, correlated with the length.

More characters means more like to be seen.

dcurtis4• 17 years ago

I tried other longer but dulled phrases:

"Please follow me on twitter here."

"I have an account on twitter."

"I'm @dcurtis on twitter."

"Follow me at http://twitter.com/dcurtis

They performed poorly.

delano2• 17 years ago

You ought to include those in the data set.

ironkeith2• 17 years ago

I've read in the past that telling people to do something isn't as effective as telling them that everyone else is doing it. I wonder how a phrase like "most people who like my blog follow me on twitter" would fare?

siong1987• 17 years ago

how about "Follow me on twitter here. I have x followers."?

PStamatiou• 17 years ago

Similarly, I wonder how some automated text like "11,399 people follow me on Twitter, you should too" would work.

forgotmypasswd• 17 years ago

Number of followers is great for determining spamminess of an account, but mentioning the number of followers makes the message seem less personal. I would guess this would reduce its effectiveness, but I'm sure a quick test would provide harder evidence :)

jseifer• 17 years ago

Or the Cialdini route: "Can you follow me on twitter because I have X followers?"

HoneyAndSilicon• 17 years ago

But some people are going to respond negatively to large numbers (or at least very large). e.g.: if I see "I have 28,437 followers" I'm likely to think: okay, this is probably some Perez Hilton type person.

But if the info could be tailored more (e.g., "2,437 HN subscribers that follow me") I would be more interested. In other words, not everyone finds following the most massive crowd a sign of value.

timf• 17 years ago

I don't have the source or exact words on hand but I read there were A/B tests done on phrasing in infomercials. The phrase "operators are standing by" fared overwhelmingly worse than "if the lines are busy, keep trying" which implied that they were typically so busy with orders that they could hardly staff enough operators.

[EDIT: thanks to ironkeith for the Cialdini link ... it was a book by him and it was in a section devoted to the "Social Proof" research]

ironkeith2• 17 years ago

It was an article referencing research by this guy: http://en.wikipedia.org/wiki/Robert_Cialdini

byrneseyeview• 17 years ago

Here are the other 49 tricks, briefly summarized:

http://www.moskalyuk.com/blog/yes-50-scientifically-proven-w...

(This link is via HN, naturally: http://news.ycombinator.com/item?id=613402)

antiismist• 17 years ago

I've read the book, as well as Influence. Influence is a far stronger book because it does a better job of describing the theoretical underpinnings of why the 50 tricks work.

Some of the tricks are really narrow and so they aren't all that useful themselves (e.g. to write post it notes and put those on interoffice forms and surveys for a better response), but with the theories you can make better extrapolations.

10ren• 17 years ago

With the one about towels (number 2), I wonder if people behaved better because they thought they were being monitored; and, if so, whether some of them decided not to stay in that hotel again.

ehsanul2• 17 years ago

I wonder how to best apply this to job-seeking.. Hmm...

dcurtis4• 17 years ago

I thought about this. But there's a fine line between being tacky and being efficient. When I read bandwagon phrases like the one you propose, I always cringe. It's just not genuine.

philh• 17 years ago

It would still be interesting to see the results.

I have a similar reaction to "you should follow me on twitter". Before knowing about this test, it possibly made me a little less inclined to like/trust you than otherwise. (I think it did, but I don't know how to measure that.)

I wonder if that's a common reaction: maybe there's an uncanny valley-type effect, where more efficient phrasings become turnoffs? Does the high conversion rate come at a cost of slightly alienating a subset of your readers?

blader• 17 years ago

This is fantastic.

Just curious - did you run any tests for statistical significance?

timr2• 17 years ago

Assuming a truly random sample of 5000 unique, non-overlapping viewers for each experiment, the standard error for the experiments runs from .3% to .5%.

The true value for any given sample is pretty likely (> 95%) to be within +/-2 standard errors. So in this case, the difference needs to be more than 0.6%~1.0%, depending on the experiments you're comparing. In other words, these look like they're significant differences.

(For reference SE ~ sqrt(p*(1-p)/N) when N is small relative to the population size)

huhtenberg• 17 years ago

Do you exclude traffic and clicks received from HN ?

jganetsk• 17 years ago

The experiment is still going on! This guy wants to see what the clickthrough rate is given the elaborate description of the exeperiment above the link.

How Hofstadterian!

seldo• 17 years ago

Generally I'm against using the word "here" as link-text, since it doesn't provide any Google-juice to the relevant terms. But in this case the only other possible link texts might be "follow me" or "Twitter", so it's probably okay.

guicifuentes• 17 years ago

This is not about this article in particular, but there has been a lot of news about Twitter these days here in HN but really, do they deserve this relevance and attention?

Elepsis• 17 years ago

I realize you said you weren't talking about this article in particular, but really, the fact that this happens to mention Twitter is entirely peripheral to the point.

axod• 17 years ago

One thing this doesn't measure is quality. Only quantity.

Don't quite know how you can measure that though.

jbenz• 17 years ago

I'd like to see how many of these visitors who clicked through each link actually converted (as in, signed up as a follower).

herdrick• 17 years ago

Right. I'd like to believe the people you really want reading your stuff aren't so suggestible. But I think is research for commercial sites.

alexandros• 17 years ago

I wonder if the wording has a qualitative effect on the people who tend to click on the link though.

abossy• 17 years ago

Why did you link the word "Twitter?" My impression upon seeing that is that the link would take me to twitter.com, not perform a follow action.

I would wager that the third message would fare better than the fourth (of the examples you gave) if you rephrased:

You should follow me on Twitter

to:

You should follow me on Twitter

dcurtis4• 17 years ago

I tried exactly what you describe.

The conversion for "follow me" as the link during my one test was 9.22% compared to the "twitter" link, which was 10.09%.

I'm not sure why it performed (very slightly) worse. It might have to do with the link being in the middle of the sentence (you have to go back and click the link) rather than at the end, where you finish the sentence and click the link.

derefr• 17 years ago

Or perhaps people thought they really were going to get the Twitter home page, but then thought "eh, may as well since I'm already here..."

scott_s• 17 years ago

Out of how many clicks and visitors? The difference of 0.87% could easily be within measurement error.

edw519• 17 years ago

Pardon me for the blasphemy, but if I wasn't already following your writing and didn't know better, this whole post would strike me as a little "mail order sleezy".

Instead of spending valuable resources tweaking the wording of your links, why not just channel that same energy into providing excellent content, and let the rest take its course.

That's why I follow you in the first place.

sgupta• 17 years ago

Tweaking the wording of your links is time well spent. Imagine if the same testing methodology was applied to the Sign Up button on a B2B app. You could drastically increase conversions with just a few small modifications!

petercooper3• 17 years ago

Precisely. It's so sad some people focus only on getting traffic to their sites and then don't focus on the conversions. If people focused on conversions and experience more than traffic, the Web would be a lot more pleasant.

DanielBMarkham3• 17 years ago

We're just going to have to get you on the Twitter bandwagon, edw519. (grin)

Resistance is futile.

petercooper3• 17 years ago

There are a lot of good books, good content, good products, good Web sites, good Twitter accounts, good whatever, out there that don't get the attention they deserve. Covering strategies and tactics for getting people engaging with you and your content is not a waste of time.

Adopting strategies like these can result in a higher mean quality of content being popular since smarter, more cynical people can learn marketing isn't all stupid and that the way to defeat the dumbing down of society is to market quality stuff better.

DanielBMarkham3• 17 years ago

Ok.

I posted an article with questions about Twitter earlier today.

So now I have ten or twenty followers.

I mean this in a nice way, but -- so what? What do you do when you get followers? Sell them stuff?

Seems like twitter followers are like blog readers. It's great to have them, but you have to act like they're not there and concentrate on content quality. Or maybe I missed something.

bts• 17 years ago

I don't the link's particular destination here matters too much. His main point pertains to the increase of clickthrough rates of (any) link by changing the language of and around its text.

DanielBMarkham3• 17 years ago

I guess my point would be why, as a reader, would I want to follow anybody? Sorry for changing the point of view to the seller, but I'm wondering what's in it for the buyer? What's the transaction that's taking place here?

Given that piece of information, applying some of the practices of persuasion mentioned in other posts starts to make sense, but without understanding the transaction, I'm not sure what his point is. Just to get people to click a button? Surely there's more to it than that. (Although I _do_ get the A/B kind of discussion, just wanted to move the discussion towards a "what's in it for me" vein if I could)

EDIT: Instead of all this A/B crap, how about just honestly tell the reader why they want to do this? Is that such a crazy idea?

simonw• 17 years ago

Ask them questions. I've used Twitter for everything from finding good bars in Bristol to getting someone to take a photo of the front door to a cold war bunker (twitpicced to me within 20 mins of the request for a picture).

TravisLS• 17 years ago

I suspect (with no evidence whatsoever) that the increase from linking "here" was because people felt clicking the link would complete the action rather than just take them to twitter.

I wonder what the effect would have been had you used: You should follow me on twitter.

jodrellblank• 17 years ago

He opens with "I spend a lot of time thinking about how to improve user experiences", so where in this are you thinking about the user experience?

Why do I want to be influenced to follow you on Twitter?

jeffesp• 17 years ago

In this case I would say the goal is improved user experience through further contact with the author. As in, how can I increase the ease of doing something that _I_ want the user to do? It is debatable whether this is actually improved experience, but I think that's how the UX people look at these things.

dcurtis4• 17 years ago

If I can phrase things to drive people more efficiently through interfaces, it will help them have a better experience.

wooby• 17 years ago

In Nielsen's "Designing Web Usability" I remember him frowning on the use of "here" and "click here" as links. Since around 2000, when I first read the book, I heeded his advice.

  For more information: click here
  To sign up, go here

were cardinal no-nos, if I remember correctly.

But it looks like everybody else uses this technique, and the average web surfer is more likely to click give the redundant area to click on.

Maybe people are just more comfortable with "click here" when the links lead to an undefined word, like Twitter. Would it be true to say that most people on the web don't know what Twitter is? Do most of Curtis's visitors know what it is? If most of his visitors don't know, are they clicking on "here" because it's a more comfortable link given Twitter is undefined? While each of his test cases vary in wording, the first 3 are all alike in that twitter is the only word hyperlinked.

My theory is that people are less likely to click on hyperlinks if the hyperlinked text or word is something they don't know about. If you're talking about a brand name, service, or anything not general knowledge than you're better off with "click here."

markerdmann• 17 years ago

It would be interesting to see a table showing each phrase you tested along with its clickthrough rate.

Confusion2• 17 years ago

I like his articles, but this one is silly. 4 Messages on 4 blogposts and he's drawing conclusions? That little data doesn't warrant any conclusions: too many uncontrolled variables. Results might as well be random. Audience, and expected clickthrough rate, per post topic may differ, the effect of repeating the implorement several times kicks in and if I bothered to give it some more thought, I could probably come up with ten other reasons that could explain the results. That was even possible after the first set of measurements in my physics graduation experiment; let alone in sociology/psychology.

zackattack• 17 years ago

i wasn't under the impression that it was 4 messages on 4 blogposts? i thought it was a universal include? but then it would only additionally account for archived posts..

Confusion2• 17 years ago

I only count 14 articles on http://dustincurtis.com/index.html, of which the first bunch doesn't contain any of these Twitter 'invites'. Neither does his blog. Am I overlooking something?

joubert• 17 years ago

His blog has appealing typography and layout.

paraschopra• 17 years ago

I cannot stop myself from doing a plug for my startup Wingify (http://www.wingify.com/) here :)

We have an A/B, split and multivariate testing platform ready to use. With features such as running tests specific to a visitor segment (say organic) and having multiple goals (say CTR and time spent), the platform is quite powerful. So, if any of the fellow HNers wants to setup a similar experiment, I would be really glad to create an account right away.

By the way, it would also be interesting to see how would call to actions rank if you segment traffic into these buckets: regular blog readers, new visitors, visitors from twitter. My hunch is that traffic from twitter would have behaved quite differently from regular blog readers.

bearwithclaws• 17 years ago

Dustin continues to impress me. Wondering had he try more 'counter-intuitive' suggestions like:"You must not follow me at Twitter"?

onreact-com• 17 years ago

How about: "People who follow me on Twitter live longer, happier lives and have three times as much sex as the average citizen! Follow me now!"

yan• 17 years ago

Are these numbers correlated with numbers of visitors to your blog? If your visitors went up by 10% during a month and the amount of your followers went up by 10%, it doesn't necessarily mean it was because of wording...

jlongster• 17 years ago

Anybody else feel like you should get a reason for following someone's twitter? "You should follow me because I'll make you pancakes every morning."

axl• 17 years ago

Seems to be a bit of a statistic maniac. http://dustincurtis.com/about.html

You should follow me on Twitter

Related Stories