Secrets of catching attention revealed! 1,072 ‘context words’ disclosed!

July 2, 2015 9:34 AM

The 1,072 Words That Will Forever Change How You Write Headlines. As some publishers struggle to grow their web traffic, one company believes increasing the ratio of some words in headlines could draw in readers. Researchers at native-advertising company Sharethrough say they have narrowed down a thousand words in the English language (pdf) that are proven to elicit higher emotional engagement. The research released today builds on a previous study published in March from Sharethrough and Nielsen.
posted by TheLittlePrince (46 comments total)

I wonder how quickly the emotional engagement produced by the words will decline when they start being used everywhere?
posted by sotonohito at 9:38 AM on July 2, 2015 [10 favorites]

"...and you won't believe what happens next!"
posted by Chocolate Pickle at 9:43 AM on July 2, 2015 [4 favorites]

That was my immediate thought. If advertisers all start using this as their standard playbook, anything using these words will quickly start to read like obviously an ad.

This seems like the kind of thing that only works when it's still relatively unknown.
posted by tocts at 9:44 AM on July 2, 2015 [2 favorites]

What's with the asterisks? (with no footnotes)
posted by MtDewd at 9:45 AM on July 2, 2015

is it about Africa? Then the top 5

Ignite
Boost
Revolutionize
Unleash
Empower
posted by infini at 9:45 AM on July 2, 2015 [2 favorites]

I think the asterisks are to denote words where pattern matching is possible. So reevaluat* is not a word but they are saying that reevaluate and reevaluating etc would be OK. The same with reflect* as reflected, reflects and reflecting. Could be wrong but that's how it reads to me.

Also typing reeveluat* without a hyphen felt wrong.
posted by diziet at 9:49 AM on July 2, 2015 [1 favorite]

1,000 words? That's not exactly limiting what can be said, nor communications magic. And I found my eyebrows rising at the idea that "the longer the better" for headlines, up to 28 words. That's sounds like anecdata of the moment rather than some universal truth.
posted by chavenet at 9:50 AM on July 2, 2015 [1 favorite]

Ignite
Boost
Revolutionize
Unleash
Empower

Isn't this the plot of Fury Road?
posted by griphus at 9:52 AM on July 2, 2015 [13 favorites]

David "The Father of Advertising" Ogilvy foresaw this kind of approach to writing headline copy ("Our business is infested with idiots who try to impress by using pretentious jargon."). Here are two of his timeless lessons:

"The two most powerful words you can use in a headline are FREE and NEW. You can seldom use FREE, but you can almost always use NEW – if you try hard enough."

"Never use jargon words {...}. They are hallmarks of a pretentious ass."

He also never forgot that research was more important than copywriting or that selling was what the client was paying for.
posted by Doktor Zed at 9:53 AM on July 2, 2015 [5 favorites]

Some publishers struggle to grow their web traffic, one company believes increasing the ratio of some words in headlines could draw in readers.

Publishers hate him! See how to boost your web traffic with this one weird trick!
posted by Sangermaine at 9:55 AM on July 2, 2015 [2 favorites]

Also typing reeveluat* without a hyphen felt wrong.

New Yorker headlines will start using reëvaluat* in their headlines. No hyphen, but you get a diaresis thrown in for free.
posted by chimaera at 9:58 AM on July 2, 2015 [4 favorites]

Uh couldn't you just do 72 words and a picture? #ClickthroughGenius
posted by Mrs. Pterodactyl at 10:01 AM on July 2, 2015 [3 favorites]

No hyphen, but you get a diaresis thrown in for free.

NEW FREE DIARESIS
posted by chavenet at 10:02 AM on July 2, 2015 [9 favorites]

Isn't this the plot of Fury Road?

Chrome
Witness
Valhalla
Warboy
Lovely
posted by leotrotsky at 10:03 AM on July 2, 2015 [3 favorites]

I remember when these cryptic, manipulative clickbait headlines first started appearing on Upworthy and BuzzFeed-esque sites a couple of years ago. I remember saying, "man, I hope this doesn't catch on as the new standard for Web headline writing; that would be obnoxious as fuck". Aaand here we are today.

(Also: "learnings". Die in a fire.)
posted by escape from the potato planet at 10:04 AM on July 2, 2015 [1 favorite]

This is wrong. A tabloid newspaper gets by with a vocabulary of barely a couple of thousand core words (a recent count of the total number of different words used in a copy of the UK Sun was 8,000). So 1,000 words for writing ads is basically more than you'll ever need in total.

"People read what interests them and sometimes it's an ad." - legendary ad man Howard Gossage.
posted by colie at 10:05 AM on July 2, 2015 [4 favorites]

Thinking about the idea of emotional engagement vs too frequent use of words.

There are words which have been categorized as beautiful and there are words which have been categorized as ugly.

It makes sense that there might be which we do find attention grabbing.
posted by TheLittlePrince at 10:06 AM on July 2, 2015

That's sounds like anecdata of the moment rather than some universal truth.

Welcome to online marketing research, the truthiest of the statistical disciplines. People will accept absolute gibberish as gospel if it ticks the right buzzwords and strikes the right pose of authoritative confidence.
posted by escape from the potato planet at 10:11 AM on July 2, 2015 [1 favorite]

Higher Engagement Rate with Longer Headline Wordcount!

I have a couple former editors who always nagged me about keeping headlines short and 'concise' and who changed them if they got 'too long' who I want to see that. And yes, the grammatically correct but running-on sentences in the article with all the semi-colons, emdashes and parentheses is a STYLE, not a flaw.
posted by oneswellfoop at 10:25 AM on July 2, 2015 [2 favorites]

The UK Daily Mail (world's number one news website) has been quietly pioneering the 'long' headline, which seems to work well online. Example from a few months back:

"Businessman who ripped out a man's heart and ate it 'to prove he was not gay' apologises to victim's family and admits he 'exceeded the boundaries of self-defence"
posted by colie at 10:36 AM on July 2, 2015 [1 favorite]

Awaiting Randall Munroe's thoughts
posted by aspersioncast at 10:37 AM on July 2, 2015

Wonderful! I could always use a new batch of words to feed into spam-filters/heuristics/'sentiment analysis'.

Odds are good that people big into clickbait/SEO are going to emphasize using these words specifically in higher density. Since they're known in advance, it'll be easier to filter them out. (Note that this is from a native advertising firm, so they're already working to produce ads disguised as articles with a thin veneer of plausible "If you saw the tiny grey-on-white text which says sponsored, you'd clearly know this was an ad". This just makes them stand out more clearly.)
posted by CrystalDave at 10:55 AM on July 2, 2015 [1 favorite]

Awaiting Randall Munroe's thoughts

Discovering the Secrets of Understanding the World!
NEW Explanations using the Best Words to Make Everything Easier!
posted by straight at 11:06 AM on July 2, 2015

I wonder how quickly the emotional engagement produced by the words will decline when they start being used everywhere?

Yeah, see econometric endogeneity, the (secret) demon of research on marketing.
posted by advil at 11:18 AM on July 2, 2015

Odds are good that people big into clickbait/SEO are going to emphasize using these words specifically in higher density.

SEO doesn't work like that anymore (at least when optimizing for Google, the dominant Search engine). It's really all about persuading readers to actually read your content. So you are going to see more clickbait titles, for sure, but people, possessing brains, will make their own choices on what they click. The Google algorithm, generally speaking, no longer serves results based on "keyword density" and has not done so for about 5 years.

Yours truly,

An SEO Hack.
posted by Nevin at 11:27 AM on July 2, 2015 [1 favorite]

It's really all about persuading readers to actually read your content.

I'm a copywriter and clients want to pay less for this than SEO.
posted by colie at 11:31 AM on July 2, 2015

The greatest headline ever written used none of these words.
posted by mr vino at 11:38 AM on July 2, 2015

Will someone link to the GreaseMonkey script that deletes these 1072 words from every web page, once it's been written? Thanks.
posted by benito.strauss at 11:42 AM on July 2, 2015

I find this graph problematic. Warning: long and nerdy stuff ahead.

First, there is the suggestion of a straight line gradient, when a better interpretation seems to be no effect for the first seven data points and a virtually flat line for the final three data points.

Secondly, eight out of ten of the data points were sampled at a similar rate of context words in headline suggesting very poor experimental design.

Third, there are no values in the y-axis and no suggestion of the intersect which can disguise virtually no effect. The range could be 1% difference or 100%.

I imported this into Photoshop and identified where each x-axis data point was by translating the percent scale to pixels. The graph is presented at a largish size which allowed for a good resolution. A single pixel difference worked out to be 0.023%.

The first group was sampled at 12.4%. Assuming a little imprecision in my method and the only number that would make sense, this is 12.5%, one out of eight of the words.

The first six groups therefore fall into a range 12.5 to 14.6%, the seventh 15.8%, the eighth 17.3%, the ninth 24.4% and the final one 29.5%.

Beyond sampling the outcome again and again at a very narrow range (12.5 to 14.6%), you run across the problem of how do you get six different percent calculations into this range.

Specifically their data points were calculated at:

12.4%
12.8%
13.7%
14.0%
14.2%
14.5% For all of these, reasonably, plus or minus 0.1% (five pixels)

So, how do you get six different percentage points from 12.5 to 14.6?

Assuming your headlines are 21 words or fewer, you have these choices:

1/8, 2/16 = 12.5% (note these two options don't give you different data points)
1/7, 2/14 = 14.3%
2/15 = 13.3%

So only three data points so far. Moving onwards:
3/20 15.0% (a little over range)
3/21 (already used, 1/7)
3/22 13.6%
3/23 13.0%
3/24 (already used, 1/8)

Two more data points in the range, but we still have only two data points between 13.6 and 14.6 and they have four.

4/27 14.8%
4/28 (same as 1/7)
4/29 13.8%
4/30 (same as 2/15)
4/31 12.9%
4/32 (same as 1/8)

At 29 words we finally have enough points within the range between 12.4 and 14.6%. However, we are still missing a data point between 13.6 and 14.6.

To keep a long story from being even longer. In order to get four different points from 13.6 to 14.6 these are least wordy options:

1/7: 14.3%
3/22: 13.6%
4/29: 13.8%
5/37: 13.9%
6/41: 14.6%
6/43: 14.0%

Did they really write headlines this long? The above options assumes they must have been comparing 3 buzzwords in a 22 word headline with 6 in a 41 words headline (unless, of course, they even wrote longer headlines). I think a more likely conclusion is that they did not present the data the way it came out.
posted by dances_with_sneetches at 12:07 PM on July 2, 2015 [4 favorites]

You lost me at "immediateness."
posted by musofire at 12:08 PM on July 2, 2015

I know. "Immediateness" is such a stuffy, latinate word. If they had said "right-now-itude" maybe I would have kept reading.
posted by benito.strauss at 12:20 PM on July 2, 2015

Is anyone else intensensly bothered they went over 1024? Noone? Okay...
posted by I-Write-Essays at 1:31 PM on July 2, 2015 [1 favorite]

You can seldom use FREE, but you can almost always use NEW – if you try hard enough

An executive at a company was talking to his marketing people, and said, "New is an old word. Find a new word."

Well, they did. The word was "improved".
posted by Chocolate Pickle at 2:23 PM on July 2, 2015 [1 favorite]

It's really all about persuading readers to actually read your content.

I'm a copywriter and clients want to pay less for this than SEO.

It's a tough sell to clients, for sure. In the past we would have written 3-4 generic blog posts about whatever per month, but over the past 6 months we have moved to doing one really good longform post of about 1500 words.

Clients don't understand it initially and would rather see more shorter posts, even if engagement is low. What we have found though is not only are the longer posts getting more engagement, they are also attracting more organic search traffic.

I don't think the client always has to determine what the price is going to be. The trick is to educate the client, or find clients that understand what you're trying to do.

Of course, there are a number of technical SEO considerations - schema markup, site architecture, site speed, redirects - that are worth paying for. Those aren't even spammy at all but are truly "Search Engine Optimization."

But the days for keyword stuffing are thankfully over.
posted by Nevin at 2:29 PM on July 2, 2015 [3 favorites]

The replacement for 'new' is 'all new'.

Come on guys let's copywrite here.
posted by colie at 2:37 PM on July 2, 2015

This seems like the kind of thing that only works when it's still relatively unknown.

ALL advertising is like that. Someone figures out a new way to deliver their message and initially it's extremely successful. Others notice and do the same, and they dilute and ultimately destroy it, after which everyone starts looking for new ways to deliver their message.

The first web pop-behind was extremely effective.
The first billboard was extremely effective.
Early radio ads and television ads were very effective.
Direct mail used to be pretty good, too.

It's a classic example of "spoiling the commons".

The problem for advertisers is that most advertising is intrusive and unwelcome, and when it becomes overwhelmingly intrusive people start to ignore it, or even to actively hate it, and to figure out ways to avoid it, at which point it becomes counter-productive leading to decline in sales rather than improved sales.

The only real solution here would be for advertisers to practice restraint, but that's hopeless.
posted by Chocolate Pickle at 2:51 PM on July 2, 2015

Important LifeHack!
posted by jefflowrey at 3:06 PM on July 2, 2015

This seems like the kind of thing that only works when it's still relatively unknown.

I think these insights are pretty timeless, actually. There is nothing really new here. But the links are a good resource for beginning writers, or writers (like me) who are trying to improve their game.

I think the very worst thing you can do when creating "clickbait" is to not deliver. The content has got to be as compelling as the title.
posted by Nevin at 3:29 PM on July 2, 2015 [1 favorite]

I don't hate this. One of the things I do in my job is interpretive writing (think museum labels) and also program descriptions. I get a lot of raw drafts of program descriptions and often have to rewrite them. When I rewrite them, I am almost always rephrasing the content to feature words that appear on this list.

I didn't have any graphs or rules to guide me, but I started out in journalism and picked up a lot of vivid writing tricks there. I've also had a fair amount of training on the communicative art of writing for audiences in settings like museums. We know a lot about how people read and what keeps them reading. Some words are flat-out more powerful than others, and always will be. That doesn't make them jargon. Writing succinctly to communicate information and ideas is extremely hard. it's easy to laugh at it, and it's very hard to do it. Just because the most dumbed-down version of this kind of thing has made us numb doesn't mean the underlying ideas aren't useful.

The long version of this list is something I'm going to refer to and share with colleagues.
posted by Miko at 5:32 PM on July 2, 2015 [4 favorites]

Web-based advertising is particularly vulnerable to the problem of getting people to hate you, because the web browser belongs to the customer and there are dozens of ways for him to control what gets fetched and what doesn't. I read somewhere that these days something like half of people using the internet use some sort of ad blocker, and it's the more hardcore users who do it. Which is a problem because they're the most desireable targets for advertising. (20's and 30's, technically sophisticated, and usually better paid than average.)

And the vicious circle has long been evident. The ads are growing larger, more intrusive, and more obnoxious, which induces more and more people to install ad blockers just to get rid of them. As effectiveness of the advertising declines, the advertisers are more and more desperate and the make the ads even more obnoxious.

Miko, it doesn't matter to me whether you can make your link-bait headlines more attractive by using words from this list. They don't work on me anyway because I never see them.
posted by Chocolate Pickle at 5:43 PM on July 2, 2015

Right, I'm not writing link-bait headlines, as you may have noted had you read my comment.
posted by Miko at 5:48 PM on July 2, 2015 [2 favorites]

Ad blockers don't work when articles with catchy titles (think Buzzfeed) are shared on Facebook.
posted by Nevin at 6:25 PM on July 2, 2015

Ad blockers don't work when articles with catchy titles (think Buzzfeed) are shared on Facebook.

This is why I love things like Social Fixer, so that articles like that get filtered out of my feed. (And anyone who finds something which works around that too often quickly gets their shares hidden, so eventually it gets back to just the things which I want to see)

Which, I think speaks to the above point about how annoying people works particularly badly on a medium where the viewers have technical control. At this point, I use an ad/malware-blocking HOSTS file as a baseline of protection, uBlock for my general adblocking beyond that (Wonderfully efficient), and then uMatrix as the pinnacle of specifically white/blacklisting cookie/CSS/image/plugin/JS/XHR/frames on a subdomain level.

In other words, if you're running Google Analytics, I don't exist to you. If you're running Optimizely, I don't exist. If you're running Disqus or Qualtrics or anything similar, I don't exist. HTTP referrers? Stripped. You don't get to track me, you don't get to infect my computer, you don't get my eyeballs unless I explicitly grant them. Anything which smacks of manipulation/deliberate attention-getting (like the above work) generally gets treated as such.
posted by CrystalDave at 8:15 PM on July 2, 2015

Does Social Fixer remove the George Takei posts everyone is sharing? Or cute elephant videos?
posted by Nevin at 9:32 PM on July 2, 2015

If you want it to, definitely. (Mine does, your mileage varies, of course. Maybe you want those. There's room for everybody.)
posted by CrystalDave at 9:37 PM on July 2, 2015

Generally the "content" or copy I'm creating for clients is not spam or clickbait. For example, I work as an editor for a large news site. I manage the daily newsletter digests, and I also handle "social media marketing" of the various news stories. So the challenge is to persuade people to open that newsletter (which they subscribed to in the first place) in their inboxes, or click on a Tweet link.

With the newsletter subject lines, I did a lot of testing. What I noticed was that edgy, clickbait-style subject headers resulted in fewer newsletter opens. Generally, people are opening according to subject. "ISIS", "Israel" and "Japan" do well for some reason.

So this is a very interesting subject for me.
posted by Nevin at 12:21 PM on July 3, 2015 [1 favorite]