Bouncing in SERPs - If it's not Rank Modification for Spam Detection?

[TheOptimizationIdiot]

There have been a large number of reports of SERPs "bouncing" and settling after changes are made to a page, redirecting, links being added (even nofollow appears to be a possible cause), but the following video seems to imply the Rank Modifying for Spam Detection Patent is not in use: [youtube.com...]

Thread: [webmasterworld.com...]
Patent: [patft.uspto.gov...]

We know bouncing is happening:

All 3 pages had received approximately 50 FB likes, 1 tweet from an account that tweets on this subject, and 1 G+ from an account that shares on this subject.

I then posted 7 blog comments on articles that were on the exact same subject with the same keywords in the title or had close synonyms, with a link pointing to keyword1 and keyword2. I left keyword3 alone. All the links were not keyword laced with 6 being links under my name and 1 a bare url. All of them were "no follow."

Keyword1 dropped from the serps the next day. It reappeared the next day at the top of page 2. It bounced on page 2 for a week then settled at position 13

Keyword2 remained unchanged for a week, bounced to down 1 position for a day then moved up 2 positions before settling back to it's original position at the moment.

Keyword3 remained unchanged during this whole process.

-taberstruths

I've sat and watched bouncing right after changes that stabilize over time.

Me, too. I mentioned that observation too, right after the patent hit the light of day. I'm thinking it may actually be applied, but only in very specific cases. My guesswork, not Google's official statement at all.

-tedster

Above from the Penguin's 1st Birthday Thread [webmasterworld.com].

[After redirecting an EMD] We've not experienced a rise quite so meteoric, but are definitely bobbing back to the surface. 40s last week, 36 over the weekend and 20 this morning.

---

Fell to mid 40s over the weekend. Decided not to touch a thing and hold fast just in case this was some sort of google a/b test. Then we popped up at no.9 this morning.

---

[Less than 1 hour later] Errrr... no.5 now

-overscan

I have gone from page 1 for years and years to nowhere to be found, to page 15, to page 2, page 7, and now whatever page 250'ish is on.


-mihomes

We're hovering around positions 5 - 8 at the moment. We used to be #1 but still many pages to get picked up and redirected.

-lewis1

Above from the EMD Update Thread [webmasterworld.com]

---

And I could go on and on if I "dug through history" here, because "bouncing" in the SERPs has been well documented for years, but I think that's enough background to start, so...

---

[Forgetting for a minute what the name of the algo portion causing it is called]

What are people seeing that causes bouncing?
Personally, I notice them most after changes affecting a large number of pages on a single site.

How long does it last?
Used to be a couple of weeks when I first noticed it, but I've seen it go for over 2 months recently.

How many positions are the bounces?
I've seen from "small" (1 to 5 places) to "all over the place" (gone from the SERPs, then back and forth between pages, then different positions on the same page).

What do you do about it & what impact does your reaction have?
I've found "waiting it out" to be best, because it seems if I make changes either "bounces worse" or the pages will completely drop from the SERPs for a period.

---

[Back to what the name of the algo portion causing it is called]

If it's not "Rank Modification for Spam Detection", what is it then, "Random SERP Position Fluctuation for Finding Reactionary Rank Manipulating Optimization"?

[n00b1]

Matt Cutts seems to imply that this sort of 'rank modification for spam detection' technology isn't implemented. I know he doesn't flat out deny it but he does seem to mention it specifically here - [youtube.com...]

[diberry]

n00b1, that was addressed in the first paragraph of the OP in this thread. ;)

[n00b1]

So it was :D. I have also experienced this (and currently am). I also believe that waiting it out is the best course of action, but not necessarily the easiest. It's so tempting to fiddle around with things.

[tedster]

If it's not Rank Modification for Spam Detection?

My guess - it's a long a slow process of new data integration in the the rankings. Matt recently made a few comments about that kind of thing to several webmasters. What a tangle it sounds like Google has going, eh? No wonder he suggested we will not be able to reverse engineer that algorithm anymore!

I'm looking for a reference right now - don't buy what I say unless I can prove it.

[diberry]

Tedster, let's assume your memory is accurate, since that sounds like a reasonable possibility to explore regardless of what Matt did or didn't say. If this is what's happening:

--Why exactly would that "slow process of new data integration" cause bouncing rather than just a stable shift in rankings? (I.E., my page was #5, but then Google added new, better pages and dropped me to #9.)
--And why wouldn't everyone be seeing the bouncing behavior?

[TheOptimizationIdiot]

My guess - it's a long a slow process of new data integration in the the rankings.

Interesting. By new data do you mean something along the lines of other signals and the bouncing is actually "different scoring" rather than intentional?

So, basically more factors = overall score changes?

And what could those factors be that cause a "bounce" after we make changes when pages were stable before? Something "user behavior" related or "time to 'trust' the change" like a 301 not always being trusted immediately, maybe?

[tedster]

Why exactly would that "slow process of new data integration" cause bouncing rather than just a stable shift in rankings?

Because it's not just data integration, it then involves statistical sampling of various SERPs. There have been strong indications of Google developing something like this since well before Panda. Deep charts of the SERPs have shown wave like fluctuations that look almost like astrophysics or even quantum state equations to me.