Sculpting anchor text percentage through disavow?

[ExploreConsulting]

Hi there, should less-than-optimal links be preserved, if those links contribute to a more attractive anchor text percentage profile?

I'm working on a client who spun a bunch of articles, using keyword word anchor text. No surprise, the strategy worked great up to the penguin update.  About 90% of the client's links come from these spun articles.  The other 10% of links are naturally occurring, quality links.  Furthermore, these quality links are also keyword rich.

Now, it occurs to me that if I remove / disavow the links coming from the spun articles, I'm left with the 10% of quality, anchor text rich links.  I'm concerned that Google will see this percentage as too high, and lower the rank.

Furthermore, I have a vague memory of watching some YouTube video, where an ex-Googler says that your brand name should be about 60% of your anchor text, and everything else lower.  Finally, when I examine the anchor text in links coming into the ranking sites, they have 5-15% anchor text density on their keywords.

So, I feel a bit of a contradiction: I should clean up all of the crappy links from the spun articles, but then that risks having only the keyword rich anchor text links active? Therefore, I'm considering leaving some of the crappy links active on non-relevant keyword text, such as the good 'ol "click here" link.

Also, before answering this, I can already predict some of the answers on philosophical grounds: those crappy links from spun articles are not natural and garbage, so get rid of them.  Fair enough, but I'm also interested in an answer on only the dimension of what will produce the highest rank for my client?

[AlanBleiweiss]

Was the hit from Penguin, or a manual penalty?  If it was not a manual penalty, then in theory, you might be safe enough to keep some of those to maintain some diversity.

I would caution you though that there's no way to know what threshold exists for how many need to be cleaned up in order to address the penalty vs. how many can remain while working on obtaining higher quality links.

This is further complicated by the notion that if it was not a manual penalty, some of the losses could be to current on-site failings that were caught up in other algorithm changes before, around the same time as or immediately after Penguin.

For example, what if there were 5 problems with the on-site SEO, and the Penguin update caused a "trigger" due to link anchors?  And what if it turns out that you might only need to do some link clean-up but simultaneously also do some on-site work?

There's just no way to know in advance.  Especially without a full evaluation across the board.

Very interesting concept though.

And for the record, there truly is no secret percentage formula regarding brand instances in anchors.  With hundreds of factors to SEO, one site could have only 20% anchors with the brand in them and still have higher trust than a competitor site that has more brand anchors but weakness in other signals.