Is it possible to measure content

[silentcrowd]

I know its a odd question, but lets say you have to curate articles in bulk amount, kinda like buzzfeed volume. Apart from basic spelling and grammar checks, is there any systematic way of judging content quality.

I was thinking

>> number of links to other news pieces = info depth of the article

>> title/brief-s sentence structure = viral element of the article

>> word-count or sentences per paragraph = measure of readability

Is there any tool you use for content audit? something that goes beyond keyword densities and share counts

or is it an overkill?

 

[Kiopa_Matt]

You forgot the "put women into a frenzy" metric. That's the best one for viral content.

 

[AngryFiver]

You forgot the "put women into a frenzy" metric. That's the best one for viral content.

Today's Facebook Poll: Should there be public hangings for feminists?

Give us your thoughts below.

 

[HostZealot]

I'd say such tool most likely does not exist, yet it is definitely needed. It's inventor will earn $xxx xxx xxx, unless being killed by herds of Indians from other IM forums.

 

[JulianTrueFlynn]

You forgot the "put women into a frenzy" metric. That's the best one for viral content.

Let's be honest as marketers for a moment — I know the big joke is woman are crazy. How many times have you seen these exact images (this is ones that just popped in my head first):

It's less 'women' and just using the 'gender wars' going on to get people into a more emotional state (emotion = sales).

 

[golan]

>> number of links to other news pieces = info depth of the article

>> title/brief-s sentence structure = viral element of the article

>> word-count or sentences per paragraph = measure of readability

Is there any tool you use for content audit? something that goes beyond keyword densities and share counts

I'd say such tool most likely does not exist, yet it is definitely needed. It's inventor will earn $xxx xxx xxx

Beseech for @cardine


.

 

[rish3]

There is nothing that would accurately quantify "quality" per se. This is one of the reasons Google has such a hard time with spam. They can see things that are typically "around quality", but if you wrap a dog turd in expensive foil paper, they think there's something good inside.

There are a few things you can try that would give you part of what you are looking for.

• Readability Metrics, like Coleman-Liau Index or Flesch–Kincaid Grade Level. Open source implementations do exist.
  
• Counting spelling spelling errors (Gnu aspell, etc)
  
• Counting grammatical errors (here's an example)

I don't think this will end up working terribly well though. Readability, spelling, and grammar aren't a terrific indicator of quality. Google does this for a living, with many of the best minds in the space, and they aren't that great at it.

 

[Kiopa_Matt]

It's less 'women' and just using the 'gender wars' going on to get people into a more emotional state (emotion = sales).

Women can make shit go viral a whole lot easier than men. Most men are busy working all day to pay the bills, while their wives fuck around on Facebook and other places, sharing god knows what.

 

[avatar33]

Women can make shit go viral a whole lot easier than men. Most men are busy working all day to pay the bills, while their wives fuck around on Facebook and other places, sharing god knows what.

Not in all verticals.

In the Sports vertical, for instance, take a look at how quickly men share dunks of Lebron or moves of Cristiano Ronaldo. SBNation's facebook page is a good example of that.

Everyone is on Facebook, whether you're at work, at home, on the bus, on the shitter. Saying men are working hard while women are on Facebook is just false. I work hard and yet I still check my Facebook constantly.

 

[silentcrowd]

Women can make shit go viral a whole lot easier than men. Most men are busy working all day to pay the bills, while their wives fuck around on Facebook and other places, sharing god knows what.

I agree, women do tend to share more, we have lolcats and cute baby pics getting shared like crazy, everyday

text content wise, i am thinking its mostly listicles on fashion, diet, green living. shouldnt be difficult to analyze, since listicles tend to have similar sentence structure.

Also I am guessing if one pulls out titles from fashion sites, he would see the seasonal trends repeating in topics. Will have to look into that.

UPDATE:

@avatar33: true sports gets viral with men...but thats mostly scoring videos.... its just sad only gambling clients come looking for content..oh where are those sport merchandisers

 

[silentcrowd]

There is nothing that would accurately quantify "quality" per se. This is one of the reasons Google has such a hard time with spam. They can see things that are typically "around quality", but if you wrap a dog turd in expensive foil paper, they think there's something good inside.

There are a few things you can try that would give you part of what you are looking for.

• Readability Metrics, like Coleman-Liau Index or Flesch–Kincaid Grade Level. Open source implementations do exist.
  
• Counting spelling spelling errors (Gnu aspell, etc)
  
• Counting grammatical errors (here's an example)

I don't think this will end up working terribly well though. Readability, spelling, and grammar aren't a terrific indicator of quality. Google does this for a living, with many of the best minds in the space, and they aren't that great at it.

Hey thanks for 2 two metrics - Coleman-Liau Index and Flesch–Kincaid Grade Level. Looked into those.

I am a bit surprised they dont factor in the paragraph element. Guess I failed to define the readability term properly.

By readability, I meant the chance that the article will be read by an average reader. Also assuming we tend to avoid articles with bigger paragraphs - this means, we are less likely to read an 500 word article written in 2 paragraphs, than one thats consists 4/5 paragraphs.

I agree it is nearly impossible to accurately quantify quality of articles. But my objective here was to curate quality articles. I think some sort of content audit can definitely scale up the process, and also communicate improvement areas to writers, so that some form learning is also embedded in the process.

About Google, I have feeling they are slowly becoming more of an answering machine. So far, they have done some impressive job with technical queries, movies, musics even. Would be interesting to see how they approach more natural queries.

 

[qqbot]

I have something I think would be able to do it. I just haven't had the time and work ethic to put it together yet, plus there's the fact that being able to separate articles based on how "good" they are or whatever does not mean the same thing as separating based on how many clicks they will get.

The problem I find is the whole book/cover thing.

A great article does not always get clicks, and there are lots of horrible, shitty clickbait articles that get a fucking TON just because of the title+thumbnail.

I would suggest just put them up somewhere, throw shitty traffic at it and see what people click as opposed to programming some kind of editor-bot.

 

[silentcrowd]

Its more like identifying improvement areas, and then getting the writers to work on those.

I am guessing by "how many clicks they will get", you mean the shares it will get, I am counting on it more as a proxy factor of quality. That said, share count would also vary by article types, eg- listicle vs opinion pieces. What I am looking for is a mechanism that compares listicles against listicles (same article types)

Clickbaits suck big time, but arent they mostly used to draw readers from social channels? I am rather focusing on the first impact, title and first paragraph(at least), with the assumption if readers get through that part, share count would rise.

I would suggest just put them up somewhere, throw shitty traffic at it and see what people click as opposed to programming some kind of editor-bot.

I am planning to run a test, pull up articles from last 6-month archive from select pop sites, classify article types(diy guides, lisiticles, review, news pieces - havent fixed yet), check share profiles per article types