Google Traffic Throttling, Ceiling or YoYo - some analysis

[philburton]

Hi guys, found a load of threads on this forum about people experiencing "throttling" in their SEO traffic from Google, and there seems to be some discussion as to whether it exists, so I'd thought I'd share my experience.

I firmly believe G does throttle my SEO traffic.

I own a pet supplies website with about 600 products, and about 700 pages in total. It's been in the G index since August 2010. The site is pretty SEO friendly and has a small about of back links (about 10 or so).

I dont monitor my position in search rankings, but I do pay very close attention to the traffic I receive from G, and these are my observations:

- I only get traffic from obscure or long tail search terms

- Traffic from search terms seems to "cycle" round, for instance, on one day, I will get 1 visitor from "rotastak space command" (a popular hamster home), then I will not get another visit from that keyword for between 2 and 4 weeks.

- The "quality" of traffic being sent does seem to vary. This is a theory that has been posted by other users on this forum - suggesting that G profiles it's users and knows which users are most likely to spend a lot of time on a site, and those that wont. For example, one day, my Pages-Per-Visitor metric will be up around 7-10 for visits from G, whereas another day, they can be below 2!

- Since I had a server issue in November, I NEVER, NEVER, EVER get more than 1 visit per keyword. In November, I was starting to pick up multiple visits from keywords, but my server crashed (nothing to do with me) and ever since, G seems to be punishing me.

- I believe the "throttling" works on a weekly basis, in my case at least. Let me show you why:

Between December 1st 2010 and Feb 23rd 2011 - my weekly traffic from Google has been no higher than 93, yet no lower than 86 - that is remarkably steady, and not really showing any signs of climbing (Yet, Webmaster Tools say during the same period my impressions in SERPS have risen 5 fold).

If we take the same period 1/12/10-23/2/11 BUT this time looking at the daily return from G, it fluctuates from lows of 6 visits per day, to highs of 21 - never ever going higher than 21. Also, every "high" point is followed by a subsequent dramatic drop off in traffic which further adds weight to the "throttling" theory.

- I also agree with those of you in this forum who have identified a "push pull" or "yoyo" theory - whereby when one page gets traffic, another doesn't. I see this, but not with pages, I see it with HOURS of the day.

For example on a Monday I will get traffic spikes at 1pm, 3pm, 5pm and 7pm - with lulls inbetween. Then, on the following day I will get spikes at 2pm, 4pm, 6pm and 8pm, with lulls at 1pm, 3pm, 5pm and 7pm. If you overlay the hour by hour traffic for two days, the yoyo effect is staggering.

-------

Anyway, sorry for the ramble, thought I'd just share my observations with you guys - Im really frustrated because the traffic simply never seems to improve. Any ideas?

I think I've covered all my observations, but if you guys have any questions, feel free to ask!

[WeaselyOne]

Welcome to WebmasterWorld.

Given the small number of links that your site has (10 or so) I wouldn't expect you to receive anything but obscure or long tail traffic.

I would encourage you to spend your time link building (and/or creating engaging original content) which will increase your traffic much more quickly than researching possible issues like traffic throttling.

That's my 2 cents anyway. :)

[BenFox]

I wouldn't draw too many conclusions from such a small data set but thanks for sharing. Having said that I'm not overly familiar with yo-yo/throttling theories.

The first four points are entirely understandable if you're site is relatively small and your weekly traffic "cap" could be a side effect of ranking in #7-#10 for long tail terms.

The hourly yo-yo sounds pretty funky. Is it only on traffic from Google that you get this effect?

[philburton]

@BenFox

Yes, it's only on traffic from G that I see this affect. If I get 20 visits from G on one day, I will see a profile like this:

1pm: 4 visits
2pm: 0 visits
3pm: 5 visits
4pm: 0 visits
5pm: 5 visits
6pm: 0 visits

Then on the following day, I will see:

1pm: 0 visits
2pm: 4 visits
3pm: 0 visits
4pm: 5 visits

You get the picture!

I'm not kidding.

I have been working on my back links lately, as I know that is a problem, but I think this case is just an example of throttling, but on a smaller scale.

[ohno]

Ou sites see the same as you! Traffic spikes, some days it is like a switch is flicked at certain hours! No sales then 4 in 10 mins-normal? Not to me it isn't........

[almighty monkey]

Erm, ok, time for a bit of a reality check here guys. All businesses see this. This is not an SEO thing. When I was 16 and a checkout monkey, sometimes we'd have really busy days and some days it would be quiet. Sometimes there was an obvious reason. Some days, there just wasn't that many people doing their groceries.

It's not SEO. It's not Google. It's just business.

I've run dozens of campaigns across a wide variety of verticals. I have never seen any evidence that Google 'throttles' traffic. Why on earth would they? If they have decided you are good enough to be ranked highly in the first place, why would they swap you out for someone they're a bit more suspicious of?

You've said your traffic is changing, but your not tracking your rankings.

I think the problem is your reading the figures too closely. My advice, read trends quarter to quarter, compare and contrast the last 3 months over the same period last year. If they're up, Yay. If not, you then have enough data to draw actual conclusions. Chasing hourly blips is a waste of time and does not promote good mental health.

[ohno]

Erm, OK, Our sites are 10 years old. Sales have been throughout the day up until the May Day update & now this latest update. If Google does not throttle traffic how on earth do our visitors stay within -+10 (yes TEN!) on most days throughout the week?! I could tell you how many visitors we will have on what days next week. It's also very strange how you can be dead ALL day long then at a set hour sale after sale comes in then switches off as quickly as it came. Other members have seen this pattren since May Day too. Seems too much coincidence to me.

[dataguy]

Interesting analysis, @philburton.

I've got a site with 150,000 pages and a few million backlinks, and I see essentially the same thing.

I run Windows servers with Classic ASP, so I don't have very many options for traffic stats software like the Apache / PHP crowd has. Because of this, I've written my own traffic analysis software using MS SQL to hold the traffic data instead of using server logs.

This has resulted in me being able to do some pretty fancy analysis, very different than what most webmasters can do. For instance, I have a report that shows minute by minute traffic predictions. I can usually tell how much traffic G will send me within a few percentage points. Any deviation from this, and I know something is up. Often I'll see changes in G more than 24 hours before anyone reports seeing changes at WW.

A few weeks ago I decided to add a 'trending articles' report for visitors to see. I thought it would be interesting for my visitors to see the popularity rise on articles as different holidays and news events occurred.

Instead, what I found was that the pages on my site seemed to be cycled on Google, irrelevant of topic. I haven't been recording this data long enough to see a full cycle, where the same article gets promoted more in G over two different periods of time.

My guess is that because G says they can measure end-user-experience, they cycle through many of their listings, promoting them higher in the rankings long enough to see if they are still giving a good end-user experience. Then the normal ranking will be determined by those results.

This is the only way I can see G balancing new possible listings for search terms while maintaining old listings.

This doesn't speak directly to throttling, but it does speak to yo-yo'ing. Perhaps this is part of Everflux.

[philburton]

@dataguy

I think the YoYoing and Throttling are part of the same strategy from Google. The correlation (or anti-correlation) of my traffic from Google is amazing if you could only see it, there's no doubt whatsoever that Google is controlling traffic flow in an unnatural way.

As I mentioned above, I will have of period of say 3 days where I get 1 hit a day from a keyphrase - I will then go 3-4 weeks without seeing a hit from a keyphrase again - thats what I refer to as "cycling" - it's obvious G are giving my page a decent-ish listing for that keyphrase for a few days, before turning it 'off' again.

Obviously most of you guys on this site are dealing with much higher figures than I am - Im still trying to "launch" my site and get it some decent traffic - I'd be happy with 100 visits a day, but it seems like for the time being at least, I'm stuck with a traffic ceiling of 100 per week!

[philburton]

Oh and Ive been reading the "zombie traffic" thread - I see this too, I sometimes have days where I only have zombie traffic - ie users who hit one page and thats it.

I dont know if this has been mentioned yet, but I suspect "zombie" traffic might be traffic from google images. Which would explain why they hit one page, and thats it.

[ohno]

/\ Or the preview I wonder? Do you also see lots of foreign IP's on these days?

[philburton]

Yeh, it could possibly be the previews, but as you say, these are foreign IPs. (Usually North America)

But I seem to remember that the previews are represented differently on the keyword chart in Analytics: ie, when you get a hit from a preview, you get a hit registered against that keyword, but Analytics doesnt actually register it as a page hit - have you ever noticed when you list your traffic from each keyword for any given day, when you've got a keyword but the pages per view metric = 0? I think thats a preview.

Whereas, a hit from an image search actually loads your page doesn't it? When you look at your keyword report in GA, I think these are the ones where you have 100% bounce rate for the keyword - people are loading your website through the Google Image viewer.

Thats my theory anyways.

[BaseballGuy]

Jumping into this thread without reading the above posts...

I can confirm that Google throttles traffic. I'm getting throttled right now as well.

<<edit>>

Guess I need to backup my claim:


Ranking for one keyword which usually drives "x" amount of traffic (over the course of the past 3 years) has been throttled down to 75% of what it normally was.

Another keyword that used to drive a lot less traffic (when ranking #1) is only pushing out 30-40 visitors per day.

Throttling does exist. I firmly believe it to be a component of over optimization.

[dstiles]

I asked this question elsewhere once before and no one answered.

For any given "throttled" or "yo-yo" period does the relevant page/site appear in SERPS for the relevant keyword and expected area(s) and expected position? I haven't been able to deduce this from any discussion so far (or I missed it!). Most discussion seems to be centred on site stats.

If it is in SERPS then how can it be google?

If it isn't, it's probably google.

[tedster]

Yes dstiles - that's the issue exactly. At the level of some of this reported traffic, statistical significance tends to be lacking. The problem with "showing up in the SERPs" is that, as we often discuss, there are so many variations of "the SERPs" these days.

[TheMadScientist]

And to expand on / add to what tedster is saying about variations of the SERPs...
Who's SERPs? Yours or mine or the neighbors?

The SERPs of someone who hasn't cleared their cookies in a year, or the SERPs of someone who just bought a new computer?

[Reno]

Throttling does exist. I firmly believe it to be a component of over optimization.

Of all the stupid penalties that Google wholeheartedly embraces, none IMO are dumber than so-called "over optimization". If it's a true penalty, then the antidote would seem to be "do a crappier job constructing your pages". Personally, I find that repulsive.

...................

[philburton]

@Reno

If throttling is a result of over-optimisation then I'd be astounded.

Could it be more to do with "type" of site. Mines an ecommerce site, and there are plenty of people out there selling the same product as me - so there's lots of other sites vying for a slice of the traffic

[ohno]

/\ Not so sure on that. Our sites sell products that many also sell, however our one site sells a product we manufactuer, it's unique. That site has more signs of throttling than the other sites. Yesterday was the best day in a while, foreign traffic dropped off & sales back up. I wonder what today will bring?!

[almighty monkey]

The only way throttling can possibly happen would be for Google to remove or alter your ranking for search terms for a time period. That's the only possible way they can influence clicks in the SERPS.

Have any of you ever seen them do this?

I haven't.

For the OP: You don't track SERPs, which means you haven't got the data you need to determine what is happening. If I were in your shoes, that would be the first thing I'd rectify. Are places fluctuating when your traffic is down? Remember, since Caffeine Google is processing and reacting data much quicker. Its entirely possible for them to find a competitors backlink, rank them over you, then to find two of yours and rank you above them, all within an hour. Did someone tweet about a competitor and give them a temporary boost over you? Did they update their site 2 days ago and you only updated yours last month? Tracking your SERPS is the easiest way to flag up these possibilities.


One other theory that occurred to me after typing that out; sometimes they 'test' technologies for a time period, such as Places, Social feeds etc. If they are, for example, sticking a twitter feed at the top of the SERPS for some of your keyphrases, or a bloody great map, that'll push you further down the page for as long as they're conducting that test. Hence a drop in sales.

This is why you don't build an entire business around SEO if your smart, folks.

[philburton]

@almighty monkey

You're right - the only way they could do it would be to change where you appear in the SERPS.

Have I ever seen this no? But to be honest, I've never paid much attention to where you appear in search results because if you're logged in to your google account, it changes the results to suit what G thinks you want. Even if you're not logged in, I'm sure they base the results on your IP/Location/Browsing History.

When I have paid particular attention to positions in search results, whenever I have asked someone else to test for me - most of the time the results are different - so I have no trouble in believing that G alter or 'cycle' the positions of "lesser" sites in it's index

[ohno]

/\ Exactly. Tracking SERP's these days is both pointless & impossible IMO. It vary's so quickly, I've seen results change in a BIG in way within minutes.

[dstiles]

Tedster, TMS: As far as I'm concerned SERPS means no cookies and no javascript - no detectable history at all. It's the only way anyone can now determine how google THINKS it rates you. I accept that may still be locationally incorrect.

We have a fixed IP (which google can use to tag us - on three computers, two people!) and google thinks we live 200 miles away from where we do (a long way in the UK) because they go by our ISP's location. :)

Ok, guys, you win. SERPS cannot be determined across two computers let alone two counties.

[ohno]

Looking at logs this week I'm convived there is a problem with geo targetting. Our sites are set to UK, they used to go dead around midnight, as I guess you would expect. However, since late last week we have traffic from all over the wold constantly throught the night/early hours. Converstions this week have been good, I can see small changes in the SERPS that I SEE but that is about it.