Google ZOMBIE Traffic Observations

[samwest]

Mods Note: Split out of the October 2015 Monthly Observation thread into a separate Zombie Traffic Thread
---------------------------------
Sunday morning, usually banging away, but nothing but slow moving Zombie traffic, one or two at a time and sitting on the same pages for several minutes. Switch is currently OFF.
Still in decline. No sign of any seasonal upswing, which is way overdue.

[edited by: aakk9999 at 7:08 pm (utc) on Oct 19, 2015]

[aristotle]

glakes wrote:
In the past a certain percentage of visitors were always definitely mismatched, but there were always matched visitors that kept sales going. Now we see multiple days in a row where Google is incapable or unwilling to send well targeted traffic. In the end zombies and mismatched traffic are the same for ecommerce websites because none of them buy. We could add bots to the list too because the net effect is the same. However, I never had problems in the past with mismatched traffic depleting my Adwords account so quickly. And mismatched traffic in the past would normally have some page views, time on site whereas zombies don't. During major updates mismatched traffic would fluctuate for a week or two, unlike the zombies we have seen for the last two months. These zombies appear here to stay and have devalued Google as a stable paid source for sales, which is why my Adwords campaigns remain off.

Thanks glakes -- That's a good explanation.

Maybe the fewer page views recently is due to even worse mis-matching than before. Obviously there must be some gradation from slightly mis-matched to badly mis-matched.

[samwest]

To those wondering about mismatch vs. Zombie.
Mismatch is equivalent to rare INDIVIDUAL hits, and yes we all get a small percentage of them. Think of a crowd walking down a busy New York City street...most (thousands) are sure of where they are going, but a very few are pushing shopping carts full of aluminum cans and even fewer are walking along muttering "I feel good, I feel great, I feel wonderful" over and over. . That's normal "natural" traffic..

Zombie is en mass traffic that is walking along blindly muttering "I feel good, I feel great, I feel wonderful" or with shopping carts of aluminum cans. It's a total shutdown of conversions, nobody seems to know or care where they are headed.
Zombie refers more to an entire period or time than to individuals.

That's pretty much how I perceive it.

[Some_Bloke]

I posted earlier in this thread about my ip address being identified at the wrong location in GA. (showing as a town over 200 miles away). I've been delving into this a bit more and would be interested in some opinions.

Using this tool [iplocation.net ] my location is shown as being in 4 different towns in the UK - from the Isle of Wight off the south coast to as far north as Lancashire and Yorkshire (I guess around 300 miles away) - and none of them are actually the correct place.

If this is happening on a larger scale and google is using the same source database(s) to target its traffic, could this appear as zombie? It could also conceivably affect adwords spend by wrongly placing the ads in the first place then sending the wrong searchers to them as well (thus amplifying the effect)

Edit: Should point out that GA used to identify location correctly - I only noticed the change recently.

[RedBar]

Month / Traffic / Sales

APR 100 100
MAY 101 102
JUN 91 85
JUL 89 82
AUG 92 66
SEP 80 78
OCT 86 51

Ok, I did ask a few days ago about Zombies and AdSense, heisje's metrics above almost mirror my AdSense earnings, realistically June was another 100 earnings' month, otherwise my AdSense earnings have seen an identical fall and I have no idea why since my traffic is up ... So you ecommerce guys are being sales throttled plus I know many AdSense publishers are too ... This is not coincidence.

[Simon_H]

I believe the definition of zombie traffic on this thread was supposed to be (1) sudden on/off changes in conversion rate often between one day and the next that are too significant to be statistical noise AND (2) no change in traffic volume or even traffic increases when conversion rates suddenly drop AND (3) suddenly got worse in or around September as this is a branch of that original thread AND (4) seen across all Google channels but no others.

My understanding is that this phenomenon is NOT supposed to include (1) a general decline in sales over time, or (2) a general decline in traffic over time, or (3) an excuse to point out that Google is evil.

[heisje]

No need to complicate matters by groping for definitions : CONVERSIONS (being a function of traffic and sales) rapidly diminishing within a short time period are the reflection of zombie + mismatched traffic. Ex. if traffic falls by 5% and sales by 45% then conversion tanks. Defining zombie traffic only when overall traffic is stable is wrong. Because who on earth has 100% stable traffic! Rapid conversion deterioration reflects zombie + mismatched traffic.

Strict definitions are beyond the point : whodunit is the point. I suspect most here have a good sense about whodunit.

.

.

[reseller]

I'm fine with this thread being locked.

Me too. It seems that we have approached the end of our journey as far as "Google ZOMBIE Traffic Observations" is concerned.

Maybe we need a new thread; Google ZOMBIE Traffic Analysis And Solutions.

[samwest]

Solutions would be nice, but I'm not sure that is at all possible. The only solution I can imagine is to stop relying on free organic traffic as much and start doing some real world marketing. In the mean time, my online hook will stay in the water and I'll take whatever I can get while I work other offline interests.

[heisje]

Most people here seem to know better than throwing in the towel as a solution. Complainers, yes, but year in, year out, they have devised ways and means to overcome. They will be here next year, and the year after, and so on -- cursing the darkness (as they should) but meantime lighting a light too.

Merry Christmas, everyone!

.

[goodoldweb]

These zombies appear here to stay and have devalued Google as a stable paid source for sales, which is why my Adwords campaigns remain off.

Same here. I'm done paying for fake traffic.

[ecommerceprofit]

Here's my update. I did not turn ads off but I made my bids super low...just burning money while my clicks are through the roof. Will turn off at end of week if I do not see improvement. Yesterday and today were dismal. I expect a few up days and back down again both in organic and ads. This phenomenon started about the same time that everyone else reported...for several years never had this problem. Since this thread is still open I figure I would add more information but know there are not answers.

[ecommerceprofit]

Also cut out "search partners" leaving only pure Google search...as someone mentioned last week under an article written about this - perhaps Google relaxed some filters letting more outside click bots in...

[Shaddows]

In the past a certain percentage of visitors were always definitely mismatched, but there were always matched visitors that kept sales going. Now we see multiple days in a row where Google is incapable or unwilling to send well targeted traffic. In the end zombies and mismatched traffic are the same for ecommerce websites because none of them buy.[...] During major updates mismatched traffic would fluctuate for a week or two

I believe the definition of zombie traffic on this thread was supposed to be (1) sudden on/off changes in conversion rate often between one day and the next that are too significant to be statistical noise AND (2) no change in traffic volume or even traffic increases when conversion rates suddenly drop... AND (4) seen across all Google channels but no others.

My understanding is that this phenomenon is NOT... an excuse to point out that Google is evil.

These are pretty much how I see the Zombie phenomenon. However, I would add that it is not necessarily site-wide. We have products siloed into groups, which may not be highly related to each other. We rarely get site-wide zombies- they are confined to a silo.

It seems interest has picked up recently, but I don't agree it has only happened since September. Indeed, I have not seen any increase on my site recently.

I cannot see how this is an intentional play by Google. It makes more sense to be an accidental combination of factors.

I personally believe that sites are given a traffic quota (AKA shaping). This means your traffic is artificially boosted if low, as well as lopped off if high. If your normal matched traffic is diverted elsewhere for whatever reason, you get unqualified traffic instead. Could RankBrain be more than a smokescreen- what if it has fundamentally changed user-intent matching and the "old" quotas haven't caught up yet? What if Google is going the Big Data equivalent of a double-blind randomised controlled trial, or even lots simultaneously. Both to train RankBrain, and potentially to re-calibrate the existing quota system?

I reject the deliberate self-sabotage hypothesis. If it were correct, the following assertions would have to be true:

Google had a large bank of non-matched traffic they could choose to deploy to previously successful sites

Google is happy for a percentage of its USERS to be let down by its results (mis-matching works two ways)

The "missing" traffic has is also currently unhappy with whatever results they are being shown.

Just had another thought. The OTHER type of trials (not randomised) is a cohort study. What if, for a significant Personalisation cohort, Google is testing sending your traffic to new destinations that it thinks will be a better fit. But no one told the Traffic Shaping Algo.

[ecommerceprofit]

Shaddows...best analysis yet

[Simon_H]

@Shaddows Thank you so much for that; it's great to get back to some sensible debate on this. I also believe Google uses traffic shaping with allocated quotas. I'd previously proposed quality quotas as well as volume quotas, where Google will switch in and out high vs low confidence traffic based on a site's quality allocation, using long tail and ambiguous serps to control this.

Your final paragraph is really interesting, suggesting this phenomenon is caused by a combination of several things Google are working on that are conflicting. That makes a lot of sense. The question is why it's only hitting some sites or , as you say, only some parts of sites. May I ask if your site has an quality issues, e.g. Penguin/Panda/Phantom?

[Shaddows]

No, no issues that we know of. Sales and revenue acquired through Google channels have steadily increased for years.

To be honest, quite frequently sales lost to Zombies in one product subset are made up from increased sales (usually through increased volume, occasionally better quality) in another subset (by subset, I mean within the same silo). Usually the Average Sale Price is different for the subsets, so revenue is impacted. I call this a referral shift.

If the disruption slight, it normally sticks. If it is significant, it usually reverts. If it is significant and does NOT revert, usually (more that 50% of the time) a big enough update to be reported by the SEO news sites rolls through. Significantly, the referral shift pre-dates the update chatter. This would seem to be an exciting predictor of updates, except that the shift usually reverts (and no update happens), so a lot of false positives.

Sometime after the referral shift, the situation either suddenly reverts, or the zombies fade over time (or wander elsewhere).

We have good data on products as we have been around for 15 years, and our products breakdown easily into silos, and within the silos into product groups. Due to this delineation, different products have different seasonality, different niche competitors, different collateral content types and different outcomes through updates.

ETA - apologies for the parenthetically over-indulged post

[nomis5]

I personally believe that sites are given a traffic quota (AKA shaping). This means your traffic is artificially boosted if low, as well as lopped off if high.

I'm not e-commerce so forgive me for butting in. But that statement above is something I have believed for a long time applies to non-ecommerce sites. I would go further, G tries to even out highs and lows in useful traffic (as far as it is able to) to many types of sites which have an average traffic volume above a specific level.

Clearly they can't do much if your traffic volumes are very low because there is little room for leveling, but even daily traffic levels in the low thousands offer that possibility.

I've never seen anything wrong with a certain amount of "levelling" of good / bad traffic from my perspective. In fact, I hate to admit it, I have trusted in G's overall perspective of my market better than my own very narrow view. I still do.

[Simon_H]

@nomis5 @Shaddows Really interesting views on traffic shaping. I think (hope) it's more scientific than levelling low and high traffic, and the algo is in fact intentionally allocating varying traffic qualities to sites based on certain criteria. I also trust Google overall with the algo, but I do agree with @Shaddows that this smells of traffic shaping conflicting with something else. It's worth considering that any traffic quota enforcement is *very* susceptible to issues/negative SEO, because any bad/malicious/undetected bot/wrongly-assigned traffic will eat into the quota and cause havoc with conversions. So if Google is A/B testing RankBrain or messing about with Panda 4.2, then that may explain what we're seeing.

@Shaddows Do you use paid as well? If so, are you seeing similar patterns in traffic from search ads or shopping? I'm completely confused by how this is being seen on both organic and paid, but we're definitely seeing this.

[samwest]

Today I'm watching very robust traffic, very natural "human" pattern. Hardly any one page and gone visits, mostly Google organic and direct. Site is converting just like in the old days, on the hour or better. Let's see how long it holds as I'm already at my "perceived" quota for the day and it's only early afternoon. If only everyday could be like this....like it was.

My question is: why did it suddenly turn back on and why will it as suddenly turn off?
Is everyone jacked up on Halloween candy today? The world may never know.

[wgchris]

Got myself a fresh Google conversion. Here's to hoping!

[isellstuff]

Just had another thought. The OTHER type of trials (not randomised) is a cohort study. What if, for a significant Personalisation cohort, Google is testing sending your traffic to new destinations that it thinks will be a better fit. But no one told the Traffic Shaping Algo.

Let's explore this a little more... So, what if traffic shaping is actually a response to click through rates in the SERPS. So, we get fewer impressions in the Google SERPS when CTR's fall. Now, what happens when a particular entry or set of entries dominates CTR for some reason. Then, the Google algorithm, at least for a short period of time, might consider our entry in the SERPS to be less relevant and cut back on the impressions.

Now, assume that some clever webmasters have figured this out and are running bots on the SERPS to manipulate the traffic shaping... They could then artificially inflate their "relevance" and negatively impact everyone elses. Also, assume they are targeting specific keywords that are the most profitable.

Lastly, assume that this negative affect has a decay rate measured in days. So at some point, it goes away and there is a spike in quality traffic.

Once again, I'm just throwing ideas at the wall... I personally like the idea of Google not taking into account the affect of the shopping widget grabbing most of the clicks and thus allowing the natural SERPS to spin through entries looking for the highest CTR (and thus the highest relevance). I think that in many cases, the answer is "all of the above". These aren't simple problems...

[samwest]

I really think this has a lot to do with good referral back links returning to favorable position temporarily. When my conversions return, I start seeing a lot more of the old good back links returning temporarily today was a record of pinterest hits and a lot of old back links...but tomorrow they'll be gone. It's the new everflux, but the week will balance in the end. funny stuff.

BTW, the surge of pinterest is no surprise, seem every search I do, they are on page one for many, many unrelated or very thinly related queries. I search "How to..." and pinterest comes up every time, but with no details on how to do anything other than look at pics. Just seems like they are G's flavor of the month.

[Nutterum]

Lately I have seen plenty of these temporary referral links myself. Our niche is very "elite" when it comes to backlinks and even the big boys have small but extremely high quality backlink profiles. Yet, since November I've seen increase in both spam referrals and temporary backlinks coming from auto generated content from even stranger websites. First I thought it was negative SEO attack, but it wasn't as these links appear for a day or two, then vanish as the landing page pointing to my propery disappears. Zombie traffic, now phantom links, it really is Halloween.

[samwest]

Total OFF day. Yesterday shut down at 2pm, then a couple of abandons (rare) followed by the old reliable 7am conversion (which abandoned), then nothing since. predictable after the Monday morning run.

[goodoldweb]

Still no sales here. Nothing but dead beats.

Today's web according to Google...

[masterjoe]

samwest where are you located? I always seem to have a sale at 7am too, however I live in Australia.

It's been dead the last 2 days, total apocalypse.

I've been working on another website that targets the same keywords and it's starting to pick up traffic. The difference is this time it won't be penalized by any algo in the future.

[samwest]

@mj - I'm in the Upper Midwest, USA. It's been a real rarity to get a conversion after between 3pm. That user to be my busiest time.

[masterjoe]

Yesterday was a bust too. Zero sales. You guys are saying that you don't think this is intentional by Google. but quite honestly I don't see how that can be the case. They are a multi-billion dollar company and the slightest change in their algo could cost millions or make them millions, I think they know exactly what they are doing, we can only guess as to the reason.

[Shaddows]

They are a multi-billion dollar company and the slightest change in their algo could cost millions or make them millions

Every decision them make will change their outcomes by fractions of a percent. They make tiny changes (or even quite big changes) looking at aggregated data.

I do the same on my site. I make changes, outcomes are affected, each change is not that big relative to my business.

Now, for an individual product I sell, it might go from being a decent mid-level seller, to a relative pariah. As long as my revenue goes up overall, I don't care. More specifically, I did not plan anything for that product, I do not even notice that product, either before or after.

Then a bunch of products get together and start chattering about how I am deliberately reducing their sales, because, like, I must be.

[doc_z]

You guys are saying that you don't think this is intentional by Google. but quite honestly I don't see how that can be the case. They are a multi-billion dollar company

Even in the past, when the ranking algorithm was more simple, there were bugs (example 1 [webmasterworld.com] example 2 [webmasterworld.com]). This is quite normal for such complex algorithm.

And even in the past we had the discussion about intentional action or side effect, e.g. the sandbox [webmasterworld.com]. For me it sounds not like an intentional action.