How Google can determine trust

[goodroi]

Google is always trying to separate quality from spam. Sometimes they succeed and other times they completely screw up. Let's talk about the ways you can help convince Google to trust your website. Let's be clear - I am not claiming Google is currently using any of these. I am listing possible signals that I would use if I was a Google engineer. Many of these things may sound simple to you but I get many people asking me to review their under performing sites and often find these issues. Even if Google isn't using these signals these ideas smart for your business.

a) Use one of the original tlds like .com or appropriate country tld but more importantly avoid unusual tlds like .info, .zip, .work. The % of spam sites jumps exponentially when you compare .com with the cheaper & newer tlds. There are a bunch of reports from email filtering companies that complain about these domains, imho likely Google is seeing similar results with web spam & weird tlds.

b) Use https. Spam sites aren't likely to take this step and spending a few extra bucks says you are serious about your website.

c) Develop editorial backlinks aka backlinks that are embedded in relevant content with natural anchor text. If I was Google I would mostly ignore ROS links in the header and footer. Those types of links were hallmarks of paid links. High % of identical anchor text was also a easy way to spot paid links.

d) Make sure you have privacy policy, terms of service, contact pages and other administrative pages. Many spam sites don't take the time to do these pages. Even if it wasn't a possible trust signal you should do it because its good for business. Some governments require privacy policy, terms of service are good for legal protection and contact us is helpful for customers, journalists & potential business partners to reach you.

e) Intelligently monitor bounce rates. High bounce rates are not the kiss of death but they are a good way of finding bad user experiences on your site. If someone visits your online calculator I expect a high bounce rate or you need to make your calculator easier to use. If someone visits your sales video and leaves in 5 seconds, you probably want to rework that video or replace it with a better sales pitch. If I was a Google engineer I would monitor bounces back to the serps as a sign that search result didn't satisfy the user and should rank lower. As a business owner I keep an eye out for

f) Build a brand. If I was a Google I would want to see at least some searchers typing in your brand name into the search box. This is a good signal you can be trusted because no one (almost no one) would type in a spam site into the search box. What are you providing on your site that would make people specifically seek out your site and type in your domain or brand name on Google?

g) Get social. Spam sites don't take the time to build a coordinated network of social accounts. Popular brands do take the time to use different social platforms. If I was a Google engineer I would love to look for this signal. Even if Google isn't looking or can't crawl a social page you should still do it if you can connect with customers. Many webmasters used to rely on Google for their business, they diversified their traffic and now gain most of their profit making traffic from places like facebook, twitter, pinterest.

What would you add to the list?

[doc_z]

In addition to my previous post: read how Google Quality Raters determine the website reputation (section 2.3 [static.googleusercontent.com])

[aristotle]

Google has been wrestling with the problem of being able to identify trust and authority for yeasr. It's probably the biggest weakness in their current algorithm. Most of the measures mentioned in this thread can be just as easily done by spammers.

[keyplyr]

Age of domain
x Number of backlinks (high quality)
x Size of website (amount of content.)

[wheel]

Develop editorial backlinks aka backlinks that are embedded in relevant content with natural anchor text

Take this one cautiously. I'm sure it's a good idea to some extent, there's two counterpoints. First, every SEO uses this as a 'rule', so it can lead to footprints. e.g. paid blog posts with exactly three embedded text links, one to your site and two to authorities. How much of that is out there? Way too much; I will not do it personally. Second counterpoint, What's a strong reporter/journalist link look like? No links to you in the article, but a footer with a link saying 'blah blah blah is the author and his website is nnnn.com". Links like that, on the right sites, make you look like a NYT reporter, not a spammer.

[jrs79]

When I mentioned links earlier I was not advocating for link building. I was saying that in most niches there are certain links that are going to help identify your site as legit as they are associated with other legit sites in your niche such as trade organizations, regulatory boards, industry specific news sites, chambers of commerce, etc.

In my opinion these types of links are similar to what make people trust you in real life that you dont know like references. There are other things to garner trust from users like testomonials and specs, but I don't think google could use these well.

As others have mentioned there are lots of B2B and service sites that are neither informational (in that they don't depend on ad revenue or sponsors) or Ecommerce.

[Totalx]

What you put up there is already incorporated in a spammer's toolbox; nothing new, and nothing that would elicit trust. Google has already stopped spammers - for now - and most users on this forum are using techniques to game Google as well.

What is trust? Doing the right thing, it seems to me;without any notion of trying to rank a site, but putting up content for people to read. The algorithms, it seems, are there to keep people honest, and we all know that it is natural for people to game any system.

[edited by: Totalx at 4:09 pm (utc) on May 28, 2016]

[jrs79]

Totalx,

Point taken. However the types of organizations that I am referring to are not going to link to spammers. In general there are far too many requirements for a spammer to obtain membership or the detailed level of content that they would need to gain an editorial link. . It could be done, but it would not be easy. In fact if they did all of these things I don't think that they would be spammers at all.

[Totalx]

There is a possibility, that backlinking is not part of the new equation; that we think it is. Not too long ago. backlinking sites with high PR sites helped rank a site very well; hackers went as far as putting up their pages on trusted domains, and those ranked very well or served to link to their highly ranked money sites.

I think the answers may lie with the innocent writer who wants to put something up of value, and knows very little of "SEO" techniques.

[EditorialGuy]

But "guesses", yes, that's where you stand out.

We're all guessing, pal.

[JS_Harris]

My personal opinion, every webmaster has one:

Content accuracy: I'm just throwing this one out there because it's not often talked about. I don't mean spelling, I mean facts. It's no secret that google wants to become the source of information and not just the rank keeper of it and to that end they've created systems to gather information and fact check it.

I don't think webmasters realize that if they are going to state something as fact that it had better match what trusted sites are saying, have supporting information on site to explain the claim or thoroughly explain why other sites have it wrong and you have it right. Google is more likely to trust you if they can verify your content as accurate. Actually Google spends a good deal of resources on monitoring segments like the health industry because false information could hurt people.

How often do you see an article start out with something like "Everyone loves cats, but did you know yadda yadda". That opening line is a statement of fact, but does everyone really love cats? I think Google would find that no, some people dislike cats or prefer dogs or hate pets altogether. An innocent opening line like that might start the page off at a -1 in whatever system(s) monitor for accuracy. I doubt it would weigh heavily on the ranking overall but it's probably not going to help.

Another thing I see happening quite often is an article contradicts itself within the same page. This happens when someone is arguing the merit of both sides of an argument, for example, but you have to stay cognizant of how a machine might interpret your content if it is looking for accuracy, and thus deciding to trust you or not. Wikipedia goes to an extreme to try and support all statements written as fact, even requesting a source if one is not provided, and it's perhaps one of the most trusted user generated content sites online as a result.

It's a writing style change: You're not writing only to your target audience, expect every statement you make to be fact checked and being accurate likely makes you more trustworthy. It's food for thought. My last piece of advice is to be concise, which you can see I have problems with myself judging by the length of this reply!

[NickMNS]

@JS_Harris what you are suggesting is scary, at least to me. I think we all agree that we are just speculating here but I fear that what you are saying may well be true.

First off, when I do searches in my niche, regardless of the website that I find the information that is reported is almost always from the same source, sometimes re-fangled and refined but mostly simply repeated, the information is often wrong, not because the source is wrong but rather the interpretation is wrong. None the less these sites appear to do much better than mine.

I treat the same topic, but I have developed my own extensive data source, at times my site agrees with the "norm" and other times not at all. To the average user that lands on my site, when they see information that conflicts with the "norm", since I have not yet developed recognition or authority, this user probably believes the twelve other sites more than mine. Google probably holds the same opinion and use the approach to arrive at its opinion.

Now let pretend that I am right (I am right! I am always right, but that is another conversation), and the "norm" in a world without Google disagrees with me. In that world I only have to overcome those that disagree, by convincing them otherwise or disproving them in some manner. The key is we all have the same weight, if I am 1 and they are 10 it is 10:1 ratio. But in the Google-verse, the dynamics are different it is not simply 10:1, since Google is taking sides, that is siding with the "norm", I end up invisible, powerless, pushed to the bottom of the serps. User's never hear my arguments, so it's game-over or at least a steep up-hill battle.

I am still skeptical as to whether Google is doing this, but about a year ago they released a paper relating to the topic of being able to answer factual questions and the paper basically explained such an approach, where it searches the web for credible sites with answers to the question at hand and if most sites agree on one answer it assumes that it is right answer. This algo was said to power the knowledge Graph.

Moreover, A few weeks ago I had a conversation on Google+ with John Muller regarding the fact that my svg-graphs did not appear in image search. He suggested that maybe my graph did not appear in search because they did not rank or warrant being in image search. He said take a look at what other in my niche are doing, and do something similar. This was not the cause at all, since my graphs were picked up in image search as soon as I shared them on twitter. The real cause was because they were tagged as <object> and not <img>, whereas on twitter I posted png screen caps. Back to the point, John Muller, suggested if you want to rank do as others do.

This scares me, not because all the work I have done might simply be ignored. My site treats a trivial topic, so I am bothered, but not scared. But innovation stems from a one person's ability to disprove widely held beliefs, and Google has designed a monster robot that will quash any site that disagrees with the norm. To some extent this is probably necessary, because the other extreme is no better. That is, where anybody's point of view is given the same weight regardless as to whether the claim are substantiated or not.

The bottom line is that determining who to trust and who not trust, is challenging even for the smartest humans. Programming a machine to do it is probably close to impossible.

One final philosophical note, it is impossible to prove anything by confirmation, using Karl Popper's analogy of the black swan, you can find as many white swans as you like, but it only takes a single black swan to disprove the statement that all swan are white. Based on what JS_Harris describes, if someone finds the a black swan Google's algos will hide it deep in cyberspace so that others will never see it.

[Walt Hartwell]

We're all guessing, pal.

No, guessing was how people did SEO when you and I started.

Have you ever pointed tens of thousands of foreign language links to a site to see what would happen?
Have you run off-topic links through a similar topic site, which then links to an on-topic site which links to yours?

The list could run forever, but I'm sure you've done none.
You are enjoying the fruits of your labor which are totally dependent on the whims of another company you cannot control. What happens when your doorway page scheme becomes so offensive to Google they place a manual penalty and it all goes away?

Science and diversification. If you don't have it, the clock is ticking, PAL.

[aristotle]

JS_Harris -- I wish it were true that google can reliably determine which "information" is factually accurate. But unfortunately, enormous sums of money are being spent to flood the web with lies and misinformation. As someone once said, if you repeat a lie often enough, eventually some people will start to believe it.

Similarly, if the same lie is pasted onto thousands of web pages, google's algorithm might eventually start to "believe" it. With such huge amounts of money is being spent to paste lies all over the web, this is a real danger.

JS_Harris -- I've posted about this before, and don't have time to say anymore now. But since this is a thread about "trust", it should be part of the discussion, so I'm glad you brought it up. But I don't think you realize what's actually happening.

[wheel]

Google can tell garbage content. They can't tell good content from great content yet - if they could, I'd be #1 on everything (my content exceeds what others in my industry do). There is absolutely no indication that they are able to verify the validity or truth of content.

[EditorialGuy]

You were lucky in the sense you stumbled into a niche in the early days of the internet. That doesn't make you qualified to make algorithmic evaluations. But "guesses", yes, that's where you stand out.

Walt Hartwell, I have no idea what compelled you to ignore the forum rules by launching an unprompted personal attack on a fellow Webmaster World member. In any case, we are all guessing here. Like so many other discussions in the forum, this is a speculative thread. (Not that there's anything wrong with that, to borrow a line from Seinfeld.)

Also, while it's interesting to speculate on how Google might determine trust, the simplest way to earn the trust of Google (or any other search engine) is to have a Web site that's deserving of trust.

[iamlost]

Put aside trust for a moment and look at what Google can actually do by what it does, especially when it makes mistakes. The advent of the Knowledge and Answer boxes are great for this. Google doesn't actually 'know' anything, what they 'know' is what they've been 'told'.

It used to be that they simply took Wikipedia, for instance, as a 'trusted' source and reiterated their answers - even when, to knowledgeable humans, it was hysterically wrong. For quite a while I think they also sort of tabulated answers and regurgitated the most popular, which also occasioned errors. This led to serious errors sometimes, such as wrong numbers or addresses for hospitals because they didn't use the info from the hospital site itself because n-number of other sites said different.

With increased 'learning' and better chosen topic seed sites these sorts of errors have been minimised but still do pop up now and then. Which says that Google's 'trust sense' still has problems. To use an analogy: the New York Times is widely read as a trusted news source but, for various reasons, it has frequently got things wrong in minor to major to spectacular ways. Trusted, generally yes; always correct? Nada. So, how would you weight a new New York Times story on a trust scale?

Google works by popularity be it backlinks (some are more popular than others), or weather forecasts (no two weather sites ever seem to actually agree fully), or bucket sort analytics, or Roswell. All inputs are variously weighted but being on the popular side of a query answer is better in Google than actually being correct. Of course, as with most things crowd sourced answers are mostly sorta right...and that is good enough these days for recreational purposes.

Back to trust. And Google.
What I understand about machine learning and how that likely corresponds to trust in Google is that for any given subject Google has a manually chosen seed group that is ipso facto trusted. The closer another site is (by citation, linked and unlinked) to one or more of the seed group the more that site is trusted. The further away the greater the dampening factor and the lower the trust or whatever they actually call it. Yup, a variation on the old edu links are more valuable: a seed entity citation is what you need!

Fortunately for most sites in a niche 'trust' doesn't seem to be a dominant ranking factor. Or else Google needs better seeds.

[Walt Hartwell]

Three word term, Google shows a little over 100 million results. My foreign language links contain one of those words in English. Three weeks and my page is position 6. This isn't what some would call "churn and burn", but I expect it will lose position over time as I won't be supporting it. The value lies in the information gained which is exactly the opposite of speculation and guessing.

Following rules in their broadest interpretation isn't really my strongest trait.

[JS_Harris]

Google AI
+ access to all data sources
+ wanting to be everything for mobile users
= a limited lifespan for websites as we know them now I fear

Why would an AI smarter than all of the people on earth combined with access to all of the data sources on earth need websites when it can be the interface that provides all data? I think what concerns google in the boardroom is how to be trusted, otherwise they can't get their devices into our homes to monitor our everything.

The irony, here we are discussing how we can get google to trust us meanwhile they are probably having the same conversation in the boardroom because the need us to trust them enough to be let into our homes.

At least they don't lie about it - Simple, Beautiful things for connecting your home(to their data centers) [store.google.com...]

Aligning your web activities to that end might make yours one of the last online as websites go the way of the book. Still online but not really needed, virtually every part of Google is designed to that end now. I would think social sites would still be popular so that Google can monitor the chatter for emerging trends. If a meteor goes through your town people in your town tweet about it a lot before anyone else does... this gives Google knowledge of a meteor in your area and people's reaction to it(example). Google searches for a disease might signal an outbreak if they come from one area... etc. What I'm saying is - if you can build a website that feeds Google data it's lifespan is likely longer than displaying news that 1000 other sites do or knowhow that 1000 other sites have. A following means access to a group of people Google might not otherwise be able to access... you get the idea.

Google's parent company named itself 'Alphabet'. Perhaps we misunderstood that as meaning letters when in fact it means 'The Alpha Bet'? But now I'm just guessing.

[NickMNS]

@JS_Harris, you are missing one important point, AI algos need to be fed. They need a continuous stream of data, that data come from our websites. Without, websites Google is nothing.

This is one of the big differentiations between Google and Apple and Facebook. Apple can exist without a web (the web as we know it), it relies on the sale of apps, devices and media. Facebook is its own website, it has every interest in breaking the web. That is why it struggles so hard to try keep users away from the web and as much as possible and instead keeps them on Facebook controlled entities. Google has nothing, well maybe Google+. How will Google serve ads to users. If there are no sites to send users to, who is going to be searching.

As for the Alpha-bets, during the last shareholder presentation they were referring to the divisions as bets, eg: Google bets and other bets.

[Balle]

I am pretty sure the main trust measuring factor is still links - natural looking links pointing to your website, links that does not look spammy/paid or manipulative.

How to get such links I have no clue.

These days I just try to make an above average website and hope for goggle to update that Penguin beast.

[aristotle]

They need a continuous stream of data, that data come from our websites. Without, websites Google is nothing.

After Google copies a site's content into its Knowledge Vault, it doesn't need the site anymore.

[NickMNS]

That would be true if we lived in a static world where knowledge never changes, but we don't. My site is updated regularly, so Google needs me and others like me to continuously keep things up to date. Don't worry I am not being narcissistic, If my site disappears from search,it would be no big deal for Google as there are others providing similar information, the aggregate is stronger than any one site (and more reliable). But if Google, stops sending traffic, and paying customers to all the sites, then it would no longer be just one site, but all the sites that disappear. Then what? The best and most sophisticated AI or ML algos are useless unless you have quality data to feed them.

In fact if you have good data, you can get pretty far even with the most basic models. If your data sucks, no amount of sophistication will help.

[blend27]

@EditorialGuy

Also, while it's interesting to speculate on how Google might determine trust, the simplest way to earn the trust of Google (or any other search engine) is to have a Web site that's deserving of trust.

Yep, from NYTimes: [nytimes.com...]

Google is said to have followed the same playbook for years: introducing a free product into a competitive space, subsidizing that product with advertising revenue, and then closing off competition through discriminatory and exclusionary practices.

[EditorialGuy]

In fact if you have good data, you can get pretty far even with the most basic models. If your data sucks, no amount of sophistication will help.

Yes, and it's not just data. Much (most?) data is a commodiy. It's the "value add" (packaging, presentation, audience appeal, depth of coverage or analysis, etc.) that distinguishes site A from site B--or, that matter, from an answer box on Google, Bing, Yahoo!, DuckDuckGo, or another search engine.

[JS_Harris]

That would be true if we lived in a static world where knowledge never changes, but we don't. My site is updated regularly, so Google needs me and others like me to continuously keep things up to date.

They want ALL data sources, if they don't have them already, so why would they need to visit your site for updates when they were in the room with you as you learned about the subject? I don't think people quite grasp the level of surveillance(data gathering) being invited into our homes and offices. A popular brand of television had to recently append their terms of service with a warning not to speak about anything you deem sensitive within range of your television because it is always listening for voice commands and stores all data on the cloud.

Were talking about a company that was scanning for wifi signals and entering improperly secured homes to gather whatever information was there while driving by with their google maps vehicles. That takes planning and a specific mentality to do. Are they deserving of trust? One side of their company does amazing things, the other unfortunately works on projects and with people who don't respect our privacy.

#1 sign of website trust - interactions with other trusted entities(backlink sources, mentions by important people, a generally positive experience by most visitors etc. It's not about what webmasters do to their sites, it's about how useful and well received the site is by others on an ongoing basis. Nothing is static with google, everything is fluid.

[Walt Hartwell]

Yes, and it's not just data. Much (most?) data is a commodiy. It's the "value add" (packaging, presentation, audience appeal, depth of coverage or analysis, etc.) that distinguishes site A from site B--or, that matter, from an answer box on Google, Bing, Yahoo!, DuckDuckGo, or another search engine.

That's certainly an opinion. Could you please cite references validating that opinion? I'd like to research it more.

[martinibuster]

What part are you interested in researching? That data is a commodity? Or that "value add" distinguishes a site from it's competitors?

The "value add" analysis is an old way of promoting a site over competitors. You create a list of positive attributes (aka Mojo) associated with competitors and then create a list of deficits (aka Nojo). Then you cherry pick the Nojo to make them your Mojo and decide if there's enough there to disrupt them. That's my approach. It works.

[Walt Hartwell]

The "value add" analysis is an old way of promoting a site over competitors. You create a list of positive attributes (aka Mojo) associated with competitors and then create a list of deficits (aka Nojo). Then you cherry pick the Nojo to make them your Mojo and decide if there's enough there to disrupt them. That's my approach. It works.

Since it appears we are heading into a different direction and perhaps using wordgames to describe them.
People can certainly identify the positive aspects of competitors in the upper positions of the SERPs, and replicate those aspects for their own site(s). Identifying and eliminating negative aspects/building out the deficits of those same sites would be the next step, although it could be done concurrently.

No argument with that concept when it is presented with thought and description rather than opinion. I would, however, strongly disagree that data is a commodity. But that depends on the definition of data. The population of a given city is commodity data. But there are many other pieces of information that aren't readily available or understood and will most likely never be provided in knowledge graphs or free video services.

I'm guessing you didn't grasp my dislike of "this is my opinion" posts although I fully support hypothesis.

[Wilburforce]

But that depends on the definition of data.

Data (plural of datum): facts or statistics used for reference or analysis. (OED)

Yes, it is a plural, although I often feel I must be the only person who winces every time someone says "the data is...".

[seomaximizer]

It is a very useful list,

I think that everything related with signals, about aspects that real business care about, like legal and security are always positive for trust and we all should include them overall in new projects.