Small change in structural data markup unjustly lowers page's SERP?

[1script]

Many of my pages ranking in Google exhibit this strange behavior where they would be ranking for a week or two on page 1, then drop for a week or two to page 2 or even below (sometimes not found at all) , then rank again back on page 1. This has been going on for months now, for about a year I would say, since Summer 2012.

One thing I noticed is that all these "flickering" pages have rating shown on the page and marked up as structural data (http://schema.org/aggregaterating). The more popular the page, the more chances are that someone votes on its content. Up or down, does not matter. What seems to matter is that both the ratingvalue and the ratingcount values change by one digit with each consecutive vote.

Usually, it is almost a guarantee that if you change the page's content, its Google ranking will suffer until the page is re-crawled and re-indexed and its new content is re-evaluated. So, if the new vote was looked at in terms of HTML change, it would be so insignificant compared to the pages' 25K+ content size that I don't think the ranking would change in any way. But what if the change in structural data is looked at differently? I mean, if the page only has two or three votes, an additional vote can change the value by significant 20%+

So, do you guys think this structural data markup change plays a role in the ranking changes or can the continued rank flickering be explained by something less esoteric?

[JD_Toims]

My personal "best guess" is if structured data is involved in the fluctuations it's not the ratings themselves changing that is having the influence, but it's more likely to be Google comparing user behavior differences in the results between "reviews / ratings" type pages and "less interactive" [or what are perceived algorithmically as less interactive] type pages.

I guess another version or what I'm trying to say is:
I would say it's more comparative analysis between two different types of results rather than a "direct influence" from the structured data on any give page. [Hopefully I'm saying what I'm thinking so it makes sense to someone besides me lol]

[1script]

Thanks for your input, JD_Toims. Just to add some more data to this: ALL my content pages have the ranking box, just not all pages have votes. What ends up happening is that the best pages (both in terms of user interactions - voting, commenting, etc., as well as Google ranks) have the most frequently changing structural data simply because with more people reading it, there's a better chance that someone will vote on it.

The pages that don't rank very well to begin with, seem to be coasting along at whatever ranking positions they were (give or take a few of course), and appear to be pretty stable. But the pages that rank best, also fall worst, then come back and the cycle repeats.

I've yet to implement a way to time the structural data change with rank change (and it will be tricky because there will always be lag of unknown length) but from what I can see looking at the stats, there appears to be the link, hence my post here.

I also didn't mean to suggest that there's a direct link per se (better votes mean better ranks). It looks like ANY vote, up or down, would trigger this effect.

[JD_Toims]

I also didn't mean to suggest that there's a direct link per se (better votes mean better ranks). It looks like ANY vote, up or down, would trigger this effect.

Interesting.

Are you experiencing the same type of fluctuations after a "main content" change on the pages or do the fluctuations only happen after user-input-data changes?

[1script]

@JD_Toims: Interesting indeed. The main content does not change often - it's a forum and once a thread "dies down", there's only so much interaction going on there - after a while voting on it becomes the only type of interaction happening there.

I mentioned comments before but now that I've looked at the stats with an eye on the structural data, it appears that the best performing pages (in terms of Google traffic) are NOT the ones with a lot of new comments being added. My best performing pages seem to be the ones where the content has been static since 4-5 years ago.

That's another head scratcher for me personally. I kinda assumed Google likes new content. But it looks rather they "like" that there's new content on the site as a whole but they don't necessarily rank that new content very well, if at all.

[nomis5]

Interesting.

I have ratingvalue but accompanied by a reviewCount not a ratingCount. I probably receive three or four email reviews a day but am so lazy that I only update the affected pages every three or four weeks. I've not noticed that it has any effect so far but will bear your comments in mind from now on.

[JD_Toims]

My best performing pages seem to be the ones where the content has been static since 4-5 years ago.

Fascinating -- Thanks for sharing.
[On my way to scratch my head and think a bit about this one.]

[JD_Toims]

I mentioned comments before but now that I've looked at the stats with an eye on the structural data, it appears that the best performing pages (in terms of Google traffic) are NOT the ones with a lot of new comments being added.

I've been having an e-mail conversation about the topic of this thread and it triggered a thought.

The algo is no longer keyword based for topicality, but rather co-occurring phrase based and above or below a "threshold" set by other pages on the same topic can be seen as "unnatural" or spam.

Is it possible the comments being included on the page for SEs are "skewing" the phrases present on the page(s) so the page(s) are no longer "as topical" as the original thread or are even allowing too many co-occurring phrases in relation to a natural writing style and making the page(s) look spammy to an algo?

[EditorialGuy]

My best performing pages seem to be the ones where the content has been static since 4-5 years ago.

That's true for some of my pages. Overall, the top-performing pages on our site date back a dozen years or more (some to the 1990s) and are updated as necessary to keep the information current. Updating tends to be minimal: Usually it involves changing a price or two and fixing links.

FWIW, I recently added schema.org markup to most of those pages, and it will be interesting to see if it makes any difference in terms of Google rankings and referrals (for better or for worse).