How much % of uniqueness?

This has still been a question for debate.
What is the good percentage to get when you're spinning articles?

I've read some forums where people are recommending a disturbing 30%.
Some proposed to have at least 70% uniquess.

Your opinion?

My OPINION would be 75-80% uniqueness. You never can be too careful. Also, you wouldnt wanna be penalized for duplicate content(very annoying).

Hmm... there's something I've always been wondering.
Say, for example, an article on weight loss tips.

With thousands or even tens of thousands articles on the net that provides weight loss tips, wouldn't there be a chance your article and that particular article will be very similar?

If I were to combine, say, 10 articles and spin them a little, Will Panda see that as an unique article or a duplicate?

I would be shooting for 70 or 80%. It's not worth the risk to go lower. Every Panda update is closing in again and again towards penalizing poor content. Google will, if they haven't already, one day figure out what is really original and what is spun content.

I doubt Google cares about uniqueness %.

From my experience with natural language processing, spinning text is like adding noise to image.

If you add a little noise - modern algorithms can easily recognize the image/text.

If you add too much noise the image/text becomes just that - noise. Algorithms can recognize noise easily.

Thats why i don't use spinners.

I would estimate a uniqueness of 70% or higher is best.

Yeah, I shoot for 70%+. I just take an article from somewhere, have it sentence level spun first (so the original sentence, and then a rewritten form of it).

Once I have that done, I go through and do world level spinning. Takes a while, but well worth it.

70-80%

i would never go under 70% . better still dont spin. just write unique

I also say 70%. Spinning is not fun to do, but better to be too unique rather than not unqiue enough.

If you really must spin, then 30% uniqueness is way, way too low. Then again, the more you spin it the more it loses meaning, so you really want to be closer to 50% than to 100%. I'd agree with the majority in here that 70% is about the best level you can aim for.

Mind you, unless you're really pressed for time I'd look at another way of doing it. Spun content is not really the best thing to use for anything that you're setting any kind of store by. Whenever I am reading content I can always spot something that's been spun. Even doing an original rewrite is better - you can take the sentence and change it around a bit while retaining the meaning, something that you just don't get with spinning software.

Even the better spinners out there are still in the dark ages. A quality spinner would have to have built in algorithms to ascertain the particular usage of a given word or phrase based on the context it's in, and then only select synonyms that matched that usage. Until then auto-spun content will always be easy to spot by the first sentence of gibberish.

In the meantime you run the risk of:

"I will look for you in the woods"

turning into...

"I would appear with yourself at the lumber"

Absolute shite! Going through and correcting that nonsense can take just as long as rewriting it.

Now that my rant is over I'll answer the OP's question.

For my site's: 100% unique hand written
For web 2.0: 100% unique, spun, then manually edited
For mass submission: auto spin on best to 30-40%, then manually edit to 60%+

Well..in my opinion you should have at least 35-40% uniqueness.

BlackSeng said:

With thousands or even tens of thousands articles on the net that provides weight loss tips, wouldn't there be a chance your article and that particular article will be very similar?

If I were to combine, say, 10 articles and spin them a little, Will Panda see that as an unique article or a duplicate?

Click to expand...

Thousands/tens of thousands?... Hmmm...

Do you realise that Google indexes a Trillion pages, and that weight loss is one of the largest niches on the planet. From recent stats I've seen, I'd say your looking at in excess of 100 million indexed pages on weight loss - and probably around 1 billion in the supplemental index.

Uniqueness is an entirely misunderstood topic. Copyscape is useless as it just looks at 3-4 word quanta and looks for multiple instances. It's a quick and easy estimate - but only that.

Do you really think that you can find another way of saying 'eat less calories and you'll lose weight' that hasn't been already been said on those billion pages?

And you need to distinguish between auto-spintaxing and human spinning - as Songer states above...

AutoSpintaxing, while horrible in the wrong hands, CAN produce OK content with a quick manual check afterwards - for MASS submission to article directories etc. There is such a proliferation of bad grammar today - and non-english speaking writers, that this just doesn't matter. And by 'doesn't matter' - I'm not talking ethically/aesthetically - but commercially. Submissions are just place-holders for links. PERIOD. You aren't trying to win a Pulitzer!

Manual spintaxing is the only way forward. But PROPER manual spinning is 3D spinning. ie.

1) Multiple paragraphs and reordering and random substitution.
2) Multiple sentence rewrites and random substitution.
3) Nested braces spinning of sentence chunks and words/word groups.

We call them 'UberCubez' btw, and have used hundreds of them to build multiple large networks (250 sites+ and 1-3k pages each.)

Uniquness needs to be looked at from 2 perspectives:

1. On your own sites
Here you really need to avoid intra-site duplication - as this looks bad to the SE's. Multiple pages saying the same thing on the same site is a no-no in SEO.

2. For public properties
Some inter-site duplication is inevitable. Concentrate on what is appropriate. For articles/2.0's, just get them as unique as possible inside of your time/tools constraints. ONLY do manual writing if you're building large spun-constructs - as this massively leverages your time. The days of writing a single unique article for E-Zine are long gone!