Hacker News new | past | comments | ask | show | jobs | submit login
Email Reputation Causes Penalties in Google Search Results (lockergnome.com)
77 points by joahua on May 31, 2011 | hide | past | web | favorite | 46 comments



I left a "this isn't true" comment on the original thread but I think the comment will need moderator approval.


Matt, feel free to answer this in either place but you singled out a sentence in the post and claimed it wasn't true.

Then you followed that statement up by saying that the title isn't true.

While I most definitely appreciate you weighing in on the issue, I feel you would better serve readers by explaining whether the initial statement, the title, or both are inaccurate.

Are you saying email reputation wasn't the reason his site was penalized, that he didn't have a tip from a Google employee, or that email reputation doesn't factor into the search rankings?


Matt you left a "this isn't true" without actually saying what about my article isn't true. In fact, you singled out one sentence in the article as being untrue, which would suggest you consider the rest of the article to be true.


> you singled out one sentence in the article as being untrue, which would suggest you consider the rest of the article to be true.

This is not even remotely valid reasoning.

He weighed in on one part of your article because he's in a special position to know whether or not it's true. Choosing not to weigh in on the rest of the article says nothing about his opinion of the rest of it.


Are you at liberty to speculate why those sites may have seen a big dip in Google referrals, or would that be giving away too much of your spam fighting wizardry?


We can't go into detail about specific signals because we don't want them to be gamed, but we did do a blog post with advice for publishers about this update: http://googlewebmastercentral.blogspot.com/2011/05/more-guid...


Huh. Most of those guidelines make sense (but they are so redundant they probably would not pass Panda's tests for readable-content vs keyword-stuffing), but this one stuck me as completely inappropriate, biased, and entering the realm of content manipulation: "Does the article describe both sides of a story?"


A system of rules which are only partially known, leaves the the advantage in the hands of those with access at the top of the barrel. You really think your ranks are so tight that information doesn't leak? And there for create more harm than good?

If the article is untrue, put it in writing. "Matt Cutts says Google doesn't use email as an indicator in search rankings" would kill the story.

You haven't said that, so it leaves me to believe that the article is true. And it is only the method of discovery you dislike.


Okay, here you go: Google doesn't use email as an indicator in search rankings.


This patent for "Ranking blog documents" seems to disagree: http://bit.ly/msgrsv

"[0044] References to the blog document by other sources may be a positive indication of the quality of the blog document. For example, content of emails or chat transcripts can contain URLs of blog documents. Email or chat discussions that include references to the blog document is a positive indicator of the quality of the blog document."

Andriy Bihun, the first name on the patent, is still a Google software engineer according to LinkedIn.


That's a good find, but the fact that a Google employee has a patent on using email links to rank blogs doesn't necessarily mean that Google's using it.


Realizing not everything Google puts in Patents ends up in search, I think I still don't believe you.

http://www.xyhd.tv/2011/05/industry-news/search-engine-optim...


actually, he has put in writing that the title of the story isn't true. I still think there are reasonable questions that haven't been answered but I'm not sure we'll actually get answers to those.

To be honest, I'm surprised we got this kind of response from Google at all. They generally don't weigh in on SEO theory.


Matt once told me that if I did a bad thing on my site there wasn't a penalty, I just didn't get as many points as I could. Semantics. When he says "there is no correlation between Email and search" I will maybe believe him. Maybe.

Maybe you are too young to remember the Matt Cutts "use no follow on internal links" debacle where he said that was a good idea.

I read the response from Google as Matt's annoyed a minion said something they shouldn't have, and Google's caught with their pants down on what is a pretty scary thought. "Google Reads Your Email And Does Scary S___ With It"


Matt - If the search team or gmail team are using email engagement as an indicator of a site's reputation or usefulness... or even to grade the quality of email from a site, it's really problematic. This can so easily be exploited to hurt someone else's site. You just create thousands of fake gmail accounts, sign up for your competitor's newsletter with all of them, and then never open any of their emails. Presumably there are already people doing this... and making sure those "fake" account appear as real as possible so Gmail bots can't tell the difference.


To the best of my knowledge, we don't use email reputation in any way in our search rankings.


Of course you can't because they are so accurate. 70% increase in traffic one month followed by a 80% decrease on the same site. You people have lost touch and keep experimenting live with people's businesses and jobs. Wake up


You people have lost touch and keep experimenting live with people's businesses and jobs.

Not sure who's out of touch here. Is Google really experimenting with people's businesses and jobs, or are they experimenting with their own service on which people have, perhaps foolishly, bet their businesses and jobs?

If their changes are pissing off one segment of their users it may be because they are trying to please another, more profitable, segment of their users.


Jamesbritt - [i]Not sure who's out of touch here. Is Google really experimenting with people's businesses and jobs, or are they experimenting with their own service on which people have, perhaps foolishly, bet their businesses and jobs?

If their changes are pissing off one segment of their users it may be because they are trying to please another, more profitable, segment of their users.[/i]

Fine, all Google has to say is "we will use arbitrary ways of ranking and will support more those that $upport us with adwords" not spew nonsense.

P.S. With 70% market share Google is a monopoly and totally different rules apply.


ps:

  an *italicized* word, or *more than one*.
results in: an italicized word, or more than one.


Thank you very much Groxx.


Hey Matt,

It's been said before, but I just wanted to say thanks again for coming into HN and sharing some perspective. It's very cool and insightful and appreciated.


Posted this on the original article, but had a feeling you might not return, so here goes:

Surely that points a crucial flaw in the system, how can you manually penalize a site and not put any tests in place that would let you know when the reasons for the penalization have been removed.

If Jake hadn't filed a reconsideration request he would have continued to be "trusted less" by Google thus presumably receiving less traffic.

Google should create a simple algorithm that would alert the webspam team who I presume take the manual action on websites that the reason for penalty has been removed and they should take another manual look at it.

As I see it the only reason Jake was informed of such penalization is because he filed a recon request, what about the thousands of other websites who inadvertently had bad links or another reason for penalization on the site and didn't sport the traffic dip and didn't file a request.

Maybe I'm wrong, Matt, do Google have these stops/tests/algorithms in place to stop incorrect penalization?


Sorry Matt, just approved it then. Thanks for weighing in on this thread, it's really appreciated.


No worries, happy to clarify.


"There’s really only one way to solve an email reputation problem. You have to make an effort to clean up your list. Here again I was thankful for the awesome reporting built into Aweber. I was able to quickly track down which subscribers were receiving my emails via @gmail.com addresses but not opening the email."

Does this reporting use some method other than tracked image loading (which GMail disables by default)?


Many email services use an implied open which is based on clicking on a link in the email, because you can't click on something you never opened.

Those people are actually more active than someone simply reading an email and if you have an email list full of subscribers who never click on anything, you have a very disengaged list.


Depends heavily on the email in question. If your email exists primarily to point people off to links elsewhere, then sure, you can measure "success" by how many people click those links; if you actually put useful content in your email, people might read your mail for its content and get value out of it whether or not they click the links.


I wouldn't be at all surprised to learn this isn't _technically_ true (as Matt Cutts says), but that something like it is indeed a signal in the search algorithm which in at least some cases produces the behaviour described.

I'd be very cautious advising clients to have links to their websites sent out in large, obviously spammy email campaigns.

Matt's quite likely being 100% honest when he says the domain sending the mail doesn't get penalised in search.

I wonder if he'd be quite as quick to explain whether the domains in links inside spam received by gmail are fed back into the search algorithm (or, if gmail doesn't feed that directly, whether the search team use honeypot gmail accounts to find it out themselves...)


I suspect that if the email rep theory were anywhere close to true, Google would just keep their mouths shut about it rather than tapdance around the truth.

But it might be an unintended side-effect of their obviously complex algorithm, which they are probably investigating right now.


Two prudent things to do when you see someone claiming to have insight into search algorithms -

Ask where they are getting their information. "A source at Google" doesn't cut it.

Ask yourself how much sense it makes. What is the likelihood that this could be a signal that improves search quality without causing significant collateral damage? Pretty low in this case.

I can think of plenty of genuinely high quality sites that do a horrible job with email. They probably send bad user signals through Gmail - but why should that impact web search? It shouldn't, and it doesn't.


>> "without causing significant collateral damage?"

Google worries about collateral damage only for top sites and that's so Google is not embarrassed.


I know for a fact that Gmail has implemented some (overly aggressive, and really far-reaching) "spam" triggers over the last few months. Many of them have been scaled way back over the past few weeks but the crux of the new Gmail "spam" filters centered around what Jake is explaining here: Non-engagement.

This is troubling in it's own right because of the possibility of exploitation for nefarious purposes to hurt someone else's site. I did not know about the possibility of search results being affected by the same overly-aggressive spam filters. This may not be the case, but if this is true, it is troubling.

The notion that email engagement is a good indicator of a site's sender reputation or a site's search reputation is problematic. This can be gamed so easily. I could set up 5,000 fake email addresses at gmail, subscribe to my competitor's email newsletter, and then never ever open any email for the next 6 months. Presumably, this would get them blacklisted in Gmail and (possibly) blacklisted and dropped from organic search results as well.

Cue the black market for non-engaging gmail accounts in 3...2...1...


You assume there aren't easier ways to game Google to de-rank people. Buy popunders on porn sites to your competition.

You can de-rank them, get them kicked out of adsense, and more in a week for under $500. Seen reputation management services do this to get bad reviews to fall in rankings lots of times.


The original blog poster left more details in the comments, which let me get to the bottom of this. If you want to know what really happened, here's my comment with the explanation: http://goo.gl/6A8f9


Isn't email tracking usually done using an image beacon and aren't images turned off by default in GMail? It sounds to me like he may have accidentally purged a lot of active readers by putting too much faith in flawed reporting tools.

Sure, you should never send email to anyone who marked you as spam, but deleting people just because they don't have images turned on is probably overkill.


It makes a lot of sense to improve the performance of email campaigns, but to link poor email reception to Google Search penalties is highly speculative and misguided.

The site in question was most likely affected by the Panda update which was released (in several iterations) around the same time and looking at the articles on the site I can see why this could've happened.


Hi Matt

I am the biggest fan of yours on the earth , What about Google Group emails and spamming ?


FWIW, mail service Campaign Monitor have picked up on this as possibly true, while stopping short on actually advising people to make list changes as a result:

http://www.campaignmonitor.com/blog/post/3486/do-email-newsl...


That's not what the blog entry at campaign monitor says at all.

> Personally, I haven't seen or heard of any evidence to back up this claim, beyond the anecdotal - as yet, there is no official information from Google linking email reputation to search results.

All the campaign monitor blog post says is that if it's true it has interesting implications.


Hence 'possibly' - they thought it relevant enough to share, even in its unconfirmed state. Sorry if I made that more ambiguous than it should've been...

CM seem to be pretty good at taking a long term view of how to do email marketing given a lot of their own marketing targets white label types (small agencies, etc.)... Often the same people managing websites. That said, there must be a slight conflict in their desire to keep list sizes up (check out http://www.campaignmonitor.com/pricing/ for why) in the short term.

Why wouldn't Google use data they have to improve quality? I'm not convinced they're doing it as the post reports, and there are probably privacy ramifications (like, oh my gosh, Google actually read my email I host with them for more than targeted ads?), but it doesn't take a lot of mental gymnastics to arrive at what the post suggests as yet-another-metric for page ranking.

In terms of its relative value, though, to what extent could Google Apps/Gmail users be deemed "typical" searchers if this were occurring?


Ros from Campaign Monitor here - I've just updated the post based on Matt's contribution above.

It would have been very unwise to recommend that any sender (not just our customers) change how they manage their lists based on an anecdotal report, thus we really did stress that there was no evidence at this stage to back up the claims in the original Lockergnome post.


Ros, if you don't recommend your customers keep their lists clean of inactive users, regardless of SEO impact, you are doing them a huge disservice.


By all means our customers are welcome to clean, or segment their email subscriber lists as they see fit. However, there are some sound reasons why we don't recommend this - http://help.campaignmonitor.com/topic.aspx?t=54

(To clarify for others, 'inactive' in this context means 'does not open campaigns' - unlike bounces, which do get automatically removed from lists)

That said, list management isn't the topic at hand - it's relationship between reputation and Google rank.


Of course it hurts their bottom line, so any sliver of a doubt would be reason not to be prescriptive about it.


Or....

Their point is totally valid. There is no way reliable way to know who DOESN'T open your emails. You can get a list of who definitely opened your email AND said "OK to loading images and/or clicked on a link in the email. But if they don't load images and/or don't click on a link, you have no idea if they opened your email or not. Maybe they're a privacy nut, so they're read your email then type in your website's URL into your browser instead of clicking the link in your email (because they know you or someone is tracking who clicks on what).

Assuming that it is true that Google is NOT linking the two together, I see no reason to delete subscribers that APPEAR inactive.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: