Keyword specific penalty

How to be certain if there is any?

[selomelo]

I have a small website that performs well in google, both in normal as well as BD indexes for a combination of keyphrases.

However, I have a very strange observation: its serp position for a two words keyword combination displays a very predictable behavior: Although among the top 10 with allinanchor, its SERP position changes periodically in a cycle like this:
first, it comes as # +200 then as # + 300 then as # + 400, and then can be found nowhere. After a while, the cycle repeats again (I think once in a month): +200, +300, +400, nowhere.

Is there any idea as to why this can occur? And if this is some kind of filter or penalty, what might be causing it and how can one find the cause?

Thank you

[tedster]

If the effect is only seen with an allinanchor: search, then as peculiar as this report is, I wouldn't have any concern. I don't think there are any penalties or filters that are exclsive to allinanchor and it's probably some strange artifact of the Google back end.

Are you seeing this in the normal search results anywhere? Does your traffic take a hit in the same cycle?

[selomelo]

Thank you Tedster,

The cyclic behavior is not related to allinanchor. Allinanchor remains stable (always among top 10) throughout the cycle. Only the position in natural serps changes as described.

Right now:
allinanchor: kw combination = # 3
allintext: kw combination = # 3
SERP position: kw combination = found nowhere (not in top 500)

The strange thing is that some other 15-20 keyword combinations containing the same keywords as the above performs well.

And, if any keyword of the combination is misspelled (or rather written in extended letters like "?; instead of "u" the same keyword combination yields a position at #4 not because there is any typo on any pages of the site, but because Google interpretes it correctly.

It is this last point that led me to think that some kind of filter or penalty might be at work. Let me be more specific: When you type "Turkish" + another keyword, site is found nowhere. But when you type "T?" or "turkısh" or "türkıh," + another keyword, the site appears at #4. Google substitutes extended (non-latin) characters by their Latin approximates ("u" instead of ü and "i" instead of ı ("i" without a dot). In this way, Google treats all variants mentioned above as "Turkish" in English, and serves natural results accordingly.

Somehow, these typos (that I discovered when checking my weblogs) bypass the filter, and when queried with such typos, my site comes at #3 or #4.

That is why I find this observation interesting, and therefore wanted to ask your opinion.

As to traffic, it does not effect the site's traffic at all, since its best position is +200.

[tedster]

Sorry for misreading your first post, selomelo. This does sound like a keyword-specific filter or penalty, although why it is in force is hard to say. I've seen something like this several times, but not with the cycling effect you described. In that case it seemed to be related to internal link text being too heavy in those particular keywords -- at any rate, rankings came back after we backed off on the internal anchor text.

Hope someone else will have some more ideas.