Google can't figure out my site - Is it the AI?

[MrSavage]

---

System: The following 2 messages were cut out of thread at: https://www.webmasterworld.com/google/4819521.htm [webmasterworld.com] by goodroi - 6:44 pm on Sep 25, 2016 (utc -5)

---

I don't want to swing way off topic here. However, Google saying that they won't need to comment on future refreshes? Perhaps it's as simple so the staff don't have to field those questions anymore. Up, down, left, right, lots, none, black, white, 1's, 0's, whatever. They are hardly concerned about whatever results are showing. The fact you can search something a get #1 or #2 results that are articles completely void of one or more of your typed in search terms tell me they don't give a S behind closed doors what is going on. Want traffic to your business? Buy ads. Free traffic? It's being treated as such. No need to comment on future refreshed only tells me it's another nail. Do tell. What's left now? No need to tell anyone what is going on or when it happened. You can't be accountable for favoritism in your search when it's the AI. Not us, it's the AI doing it. Our own properties? Nope, it's the AI and it knows best. Doesn't matter that the Google build the AI. It's on its own and we trust it 100%, even though it can't find anything new that isn't popular today.

[martinibuster]

[seobythesea.com...]

[RedBar]

So, why did you move your post MB?

[MrSavage]

I will comment from the perspective of a user of Google search, which I most certainly am. Perhaps some of us aren't, but when I'm on the computer (which I usually am) I'm searching with Google constantly during my work.

Google does not understand unless there are millions of conversations about said subject. Haven't we all started to notice that? It's like walking into the car dealership about the new model and all the staff walk you over to a different model. You try telling the salesman he's wrong, but he's oblivious to your concerns. He's a robot afterall. In essence, he has S for brains unless told otherwise.

So discovery? Stunted. Will that change? Certainly. It just takes millions of people (like robots) to talk about something then the AI can figure it out. Google can always say search and AI will progress. Sure it will. When something popularizes. It would appear to me as a fundamental flaw. A few cutting edge websites can't get through to this AI dummy. Afterall, what is a few articles vs. 10 billion articles talking about Widget X.

You cannot get through to Google. Auto suggest isn't the friend of some of us. It knows what everyone know. It doesn't know anything that isn't popular. How about that? How do you SEO your way through that S for brains?

Search quality team? Let's see. Today I searched for "how to fix widget ZYX" but I got all the results I got were about widget CBA. No instances of ZYX!? If that isn't F up then advise me. How can that not be a kick to the nuts to anyone involved with search quality? I come to the conclusion that search quality team is non existent.

It just doesn't matter. That's my conclusion. Free traffic gets about as much consideration as, I don't know. How do you write content for a subject when Google will never get traffic to your site? When Google learns about the subject, the corporate sites will all be there ready to get all the "free" traffic. This is where it's at.

I'm not sure the solution if there is one. I'm curious to know if the AI at some point will not require such a high threshold before they know what I'm talking about. If there is no discovery aspect to Google, I'm not sure there is anything to pursue in that regard.

Note to competition: You can market yourselves as actually finding pages with the words you've typed into the box. You type, we listen!

[Walt Hartwell]

Can't this get put in the Google rant thread? I'm sorry, that would be the Google update thread.

[MrSavage]

Which begs the question, where do off topic, irrelevant comments belong?

[Robert Charlton]

The permalink of the Bill Slawski SEO by the Sea page that martinibuster intended to link to was, I believe, this one....

Are You Using Google’s New Automated Assistant?
09/10/2016 Bill Slawski
[seobythesea.com...]

I'll let martinibuster expand further on his thoughts.

[robzilla]

The permalink of the Bill Slawski SEO by the Sea page that martinibuster intended to link to was, I believe, this one....

Or was it this one? Machine Learning Inside Google [seobythesea.com]

[martinibuster]

Or was it this one?

I meant the entire site, chronologically. Anyone who's upset because the sites in the SERPs don't have the keywords on the page hasn't been listening for the past six years and should start over.

Good luck.
;)

Roger

[edited by: Robert_Charlton at 12:35 am (utc) on Sep 27, 2016]
[edit reason] Edited at poster's request. [/edit]

[MrSavage]

Go into a store, ask for a product and have the staff walk you to a different section of the store and act like it was what you wanted. If that means I restart, then I question that advice from the sanity perspective. Results that don't include what you're looking for? If you can somehow justify that, then I really have no use for your views on anything. Period. You know what usually happens if a store won't get you to the aisle you're asking about for that particular product? You go into a different store. You would also start asking if the store gets a kick back for taking to a different section of the store. Maybe there is an incentive to show you these products instead of the product you were asking for.

Alternative search? In this world there is Bing and for me I would rather eat glass than use their search. I would rather work around with Google's shortcomings than deal with Bing. However when it comes to my sites or content, that's a different matter. If store staff keep taking people to Aisle 4 when you're in Aisle 5, there is little I could do except for hoping for alternate stores, as it were.

[martinibuster]

Go into a store, ask for a product

No. That's a flawed analogy. Walking into a store that exists to sell something and asking for a product is completely different from asking Google for a product. Google does not sell products. Just as walking into a bar and asking for whiskey will get you a drink poured and walking into a liquor store will get you a bottle on the counter. Two different contexts.

You are complaining that the sites in the SERPs do not always contain the keywords used in a search query. Slow down please and think about it. It is a fact that there are always multiple words to express the same meaning. In fact, there are multiple ways to express the same answer without using the same words the questioner asked.

Furthermore, if you can take a moment to consider how search engines work in 2016, I mean the actual science behind them, you will understand that search engines conduct tests to understand most of the common search queries and know that for many search queries there are multiple answers. There are informational answers, there are transactional answers, there are answers that are educational, there are answers that are better served with a geographically tied answer and there are answers that are topical and are best served with the latest news. There are even answers that are best served with videos.

So your point about keywords missing from sites in the SERPs has been explained: Matching keywords in a search query to keywords on a web page is painfully inadequate and does not work.

That is the modern search engine. It goes way beyond the simple search engine of 2001 that matched keywords to keywords. I truly hope you appreciate the time I spent trying to help you understand how modern search engines are way beyond matching search terms in the search box with keywords on a web page.

Good luck to you!
;)

[MrSavage]

Whoa. If I ask you what the weather forecast is and you tell me the time. Is that a good enough example of Google search intelligence? I ask you where the nearest Starbucks is and you tell me where the nearest McDonalds is. I ask you about the new Honda Accord 2016 model and you tell me about the 2015 Honda Accord. I ask for the white paint and you hand me the blue paint. I ask for a screwdriver and you pass me a pair of scissors. I ask for tomato soup and you get me clam chowder.

Have I made my point?

If there is something that everyone isn't talking about it doesn't exist in Google SERPS. IT DOES NOT MEAN THAT IT DOESN"T EXIST. It means that Google search doesn't understand what that "thing" is. The fall back is that you must have meant something else. There are hundreds, or even a handful of articles with your keywords in them, but Google is too much of a dunce (AI right?) to "get it". They must steer you where everyone else is looking and thus, disregarding what you are searching for. It's not like we are talking typo.

Searching for a 2017 item information and Google grabs everything 2016? Is there some type of justification for that? It's nothing more than a dunce of an AI system that is unable to comprehend anything that hasn't been popularize (searched for thousands or millions of time previously).

This is not about words meaning the same thing. Are you kidding me? Not sure if that is your site linked in this thread but it goes to show. Regardless of schooling or credentials, it certainly doesn't equal quality reading.

I speak from an end user perspective. I use Google all day every day for my work. I question whether other people do. I'm not trying to SEO anything. As we all know and Google says, don't write for SEO, write for your users. I will not write about a 2016 item when it's a 2017 item. I'm not going to call my widget something that exists already just so that I can catch from free traffic. I will live with the current stupidity of Google search. No choice.

[NickMNS]

@MrSavage I think the point you are trying to make is that Google works fine for common or popular searches, but seems to fall apart for less common search topics.

I think this has to do with two things, both closely related. Google returns results based on a ranking, from the one that is most likely to meet you search criteria to the least likely. ...[small side bar, when a search term is ambiguous Google often returns a variety or results that address multiple facets of the ambiguity. Eg: a search for "Python" might return results for snakes and for programming language]... Back to my point, assuming there is no ambiguity, when a unpopular search term is searched, Google may find many results, but most are poor matches. But since they return them based on rank and not a specific matching score (I assume that a some point there is cut-off for matching) the results shown all seem like poor matches. In my experience it is relatively rare that a search returns fewer than ten links.

My second point is that with Machine Learning, AI and statistical inference, these all work great when dealing with things that are popular, in the middle of the bell curve. But the further you stray from the mean the more the results can become wonky. But if you ask me I think Google does a great job dealings with tails of the distribution. It's not perfect by any means, but I think we underestimate how difficult it is to deal with these edge cases. When you search for 2017 item in 2016, you kind of have to expect that there wont be that much relevant content show you, so they show you the next most likely result.

There is one other point, user intent. When you search for 2017 item and it is 2016 and there isn't a lot of content on that 2017 item. Google's AI might assume that your true intent was to to enter 2016 item (because there is a lot of content for that). So again it gives you a bit of both, or even may ask you "did you mean". In most cases you didn't mean 2017. So most users benefit from this feature.

[MrSavage]

@nick, thank you for the rational comment. You have explained better, a lot of what I'm saying. Regarding the 2016 vs 2017 and "did you mean", I can say that Google rarely pops that up. Bing? They do that often. The way Google is, it should be a default by asking "did you mean". It's crazy to think you could search a 2017 item and then get top results with zero instances of 2017, but instead it's 2106. I don't care who you are, if you can somehow rationalize that a situation like that is fine, normal, or working as expected, I would call you an FI. Sorry.

One addition thought I had was this. And I know it's a stretch and I'm sure there is nothing to it. But just like that person in a store who would ignore the information that you were asking about and keeps taking you onto a different item, the thought of a kick back would possibly come to mind. Company X pays a bit extra so staff can subtly suggest their product over another persons. Bias, or whatever you want to call it.

I can only say this. So if you have search terms that are not common place? If it's new or rare I can say for certain your results will be much much smaller. One page? Two pages? Only a few results? Why is that an issue to a searcher? It should not be. However if I take you instead to the popular, somewhat related search term(s), guess what you get? Loads of results. And? Take a big guess what you are more likely to get. Ads.

I'm not saying it's intentional, but if I could accidentally, or "blame the AI" for taking you to Aisle 6 instead of Aisle 7? And if Aisle 6 happens to have more ads? I say I'm not complaining at all. Especially if customers keep coming back regardless of my insistence that you are not looking for XYZ, but instead you must mean ABC. Being stupid can pay off. It's a stretch and I'm not suggesting it's intentional. However, what's wrong with giving me what I actually want in the search results?

[glakes]

Rank Brain was originally used for queries that Google's main algo could not easily interpret, which left it a small percentage of search results to digest. Google opened it up to broader queries, which expanded its workload. Keep in mind Rank Brain is said to be taken offline to learn. The web moves much faster then that. Even seasonal topics, including products, might experience a heavy processing lag from Rank Brain considering how much it must be taught and the time it may need to be fed this data while offline. It seems like a monumental task to me, which may explain some of the weird results in the SERPS.

[MrSavage]

If you build a machine to produce apple juice but you are getting orange juice instead, then people involved in the process should assume that something is not working as planned. For anyone involved in the operation to suggest that it's okay to produce orange juice when the production is supposed to produce apple juice would be called a fool and fired on the spot. However, search appears to attract different viewpoints. It's a mentality where getting orange juice makes sense, on some level. There is some justification. The issue for me though, is I'm trying to get people to find my website about apple juice, but the search engine thinks everyone wants orange juice instead. On some level, this is how goofy Google search is. I'm not expecting a lot of people to see Google in the same light as I do because we may be dealing with different niches of the web. If I am covering apple juice and Google interprets "apple juice" searches to mean "orange juice", I don't give a S who you think you are, you cannot tweak your content to accommodate such stupidity. The only real question is whether Google AI can in fact comprehend something that few people know about, but still want information about. The way it is now, in a lot of instances, if there isn't enough search results for what you wanted, Google will take you to the part of the web that is chalked full of websites regardless if those results don't have your keywords anywhere on the page. This is called....madness.

[NickMNS]

@MrSavage your analogy doesn't work. Building a device to accomplish a single and very specific task is not the same thing as building a device with a very broad and largely ambiguous scope. If you asked someone to build a machine to produce a beverage based on a users request, and the user requested juice from a round fruit that grows on trees, and the machine served a glass of orange juice. I would imagine most people would think that that was a pretty good outcome.

If Google is having trouble interpreting the topic of your site, then one possible solution is to feed the bot. That is provide more text content, written for humans that explains the topic and provides references and links to related topics.

I used to have this issue with my site in the past. My site displays statistics about widgets. I had presented the stats as graphs built entirely with CSS, I had only a few table a relatively little text. A user would come to the site and get a good idea of the topic, but there was no way for a bot to understand the page. So I would get a lot of searches for widget + some random number. The number coincided to a statistic on the page but clearly the searcher intention was much different. I eventually reworked the site, built all the graphs in SVG, added a table with all the figures of the graphs and produce a text summary of each graph. Now the traffic better matches the content as do the ads.

[goodroi]

We need to be careful to avoid the general mindset of SPAM = sites positioned above mine. We may not fully understand Google's AI at all times but that doesn't mean it isn't working correctly. Just because Google's AI might be contributing to less traffic or conversions to our website doesn't mean we are right and Google is wrong. It is more likely to signify that the old fashion way of doing business might not be the most productive method for your business today.

Be careful not to get angry when the the old ways of manipulating Google search rankings become less effective and instead focus your energy on learning the constantly evolving best practices to optimize your website for the future. Even if Google is wrong and we are right, we still need to play in Google's sandbox since they have the majority of the search market.

[Robert Charlton]

Mod's note: I've changed the title of this thread, from...

"Google's AI Influence on Serps",
to...
"Google can't figure out my site - Is it the AI?",

I think the new title is more descriptive of what is being discussed.

[MrSavage]

@nick, my only point is that input does NOT equal output. If I search X and get no X, but Y instead in the results and Y is not another word for X, then that is broken in my books. Apple juice and getting orange juice? It's just to point out that apple juice cannot be confused with orange juice. Yeah, sure it's juice. Is that how low your standards are for Google search? In my books, 2017 does not equal 2016.

I am talking about getting top results with ZERO instances of the keyword. I'm not talking about words that mean the same thing. If you want to believe that is my point, then go ahead as an apologist.

If I write about X and Google wants to provide Y results, then there is jack S I can do about that. If X is cutting edge information and Google ignores it until it's mainstream, then that is not something that I can SEO or content my way out of. I need to acquire hundred or thousands of links, making it relevant and then I can watch every big brand site take those top search spots. This is how it works as of right now, whether you choose to believe or disbelieve. I would again point out that everyone has their own area of expertise and being aware of shortcomings is part of humility...