A weekly Google Hangout dedicated to discussing content marketing, search marketing, SEO and more.
Topic: Semantic Search – Turning long-tail keywords into content: A discussion about how semantic search is changing the way we consider the use of keywords and how marketers use their keyword data to optimize for semantic search.
Ray Grieselhuber, Founder & CEO at GinzaMetrics
Erin O’Brien, COO at GinzaMetrics
Karen Scates, Manager Marketing & PR
FULL VIDEO TRANSCRIPT
Karen: Hey, everyone. Welcome to the July (June) 12th edition of FOUND Friday. I’m your host, Karen Scates. With me today are Ray Grieselhuber, our own founder and CEO. Hey, Ray.
Karen: And Erin O’Brien, our COO. I have a whole team here today. How’s it going, everyone?
Erin: Good. You skipped a month. It’s June.
Karen: It is June! Oh!
Erin: It’s fine. We forgive you.
Karen: Those “J” months, they all run together. Last week, you two started discussion about keyword best practices to answer some questions we’ve been getting. That discussion has led to even more questions. My inbox just blew up.
So I thought we’d take the opportunity this week to take a little deeper dive and to talk specifically about semantic search. It has become a hot topic again as marketers are trying to find ways to break through some of the noise to reach their target audiences.
Let’s get started. I’d like to get started by defining terms. Can we define keywords? What are they in today’s marketing world? Are there other kinds?
Ray: Yes. We talked about this a little bit last week. Keywords are basically what people are searching for. But that’s the all-encompassing definition. A more specific definition would be very specific terms. They tend to be things like nouns… Not necessarily nouns but it could be basically anything that is a very concrete thing that someone is looking for. It would be something like “pizza New York” or “coffee Los Angeles.”
Because of the way search engines worked for a long time, search has adapted their behavior to keyword-type searching. They would take what would be for them a natural question in their mind and distill it down to these very specific keyword searches.
Erin: Some of that is also pertaining to what we saw. Like Hummingbird’s release back in 2013. I want to say last year but it’s really been two years now. To what Ray said, it’s really a more specific description of something. People are looking for how to do something. “How do I do X?” or “What is the best way?” or “What is the best tennis racket?” All kinds of different things. I think originally the idea keywords really first started out, it was a very general thing and that’s really not something that people want to do.
Remember we were talking about last week. Very few people or very few brands are really going to have an advantage by trying to rank for the word “shoe” because most people wouldn’t just type in “shoes” into a search. What you’re really looking for is a more specific type of shoe. You’re looking to create content around those things – department stores, places like Zappos and Amazon – it may have some benefit to ranking for shoes in general. But even then, real content, real findability is going to be something a little bit more specific.
There’s also a search demand curve. I know that we had some questions around that, too. We’ll talk a little bit more about that with a long-tail and description of those.
Karen: Let’s start by just talking about semantic search. Can you wrap it up in a nutshell what is semantic search?
Ray: Semantic search is kind of a big topic. But if you are trying to get a definition, understanding the difference between traditional keyword-based search versus semantic search, keyword search if you remember is this idea of, when you think about something you want to look for, you have this natural language and you distill into those keywords.
An example would be, “I want some coffee. Where should I go get coffee? I’m here in L.A.” You basically distill that into “coffee Los Angeles.” Semantic search wouldn’t be quite that full initial sentence, but it would be something more like that. It would be something more of like a natural question: “Where can I find the best coffee in Los Angeles?” or “When can I find the closest movie theater that has this movie?”
These are examples of unpacking these distilled keywords into semantic searches. But that’s a very rudimentary example. A better example of the way semantic search works is – what Google has done particularly is they built this technology that is much better at understanding the overall context of what’s going on.
We just had a discussion before we started about duct tape. We were kind of making fun of people who think duct tape is spelled “duck” tape. We do the search on Google for “duck tape” and it just basically immediately gave us the search results for duct tape because it knows the context for what we were talking about. It’s an example that is good at showing context but the technology to solve that problem is actually not really semantic search, but it’s kind of what we’re talking about here.
To further talk about things, the semantic web is this technology that has been around for a long time. Basically, it’s a set of technology than concepts. Really what it means is it’s talking about building relationships from all of the text and content on the Internet, building relationships from the different entities that it finds in there. Entities are examples of things like people, places, things, and all sorts of things.
If we’re talking about a bunch of movies, a good example would be the listing of movies would be a set of entities. Each actor in these movies would be an entity and so forth, the directors will be their entities. Once you start graphing out these entities and parsing them out from all the text that’s on the web, you can start to build what is called an entity graph. It allows you to see on the backend all these relationships between different things. Once you have that data set built, you can use that to provide much more sophisticated results to questions that tend to look a lot more like natural language queries.
Karen: Erin, do you have something to add to that?
Erin: What’s going to keep it moving onto taking it to where we’re going is going to dive more into semantic.
Karen: Let’s talk now more about marketers and their focus. Where should they be focusing their efforts? Are there specific types of keywords that they should be looking at?
Erin: This thing that goes to this conversation about the search demand curve and there are a lot of different searches of their keywords fall and different things. We use similar search terms to what Moz uses because we like those guys and they got some really great content around the topic. So, the keywords that contain the largest search volume percentage should be around 18-20% of total search is the fat head of keywords.
Then you’ve got what’s called the chunky middle, which received 11-12% of keyword search volume with an average monthly search usually in the thousands or in the hundreds. Those keywords that are in more like fat head area is literally like search volume in the millions and hundreds of thousands opt-in.
Then you’ve got what we call long-tail keywords. This really comprises the remaining 70% of search volume for the most part. Those are keywords, topics, terms, and phrases that have somewhere in the hundreds of searches per month. What we’re talking about when we talk about long-tail keywords is things with usually lower monthly search volume, but also a lot of times highly specific.
There’s also an interesting thing around selecting keywords for marketers which is a lot of times the keywords that are included in this long-tail are, as Ray mentioned with duct tape, a misspelling or misuse of a common word or phrase. So people will oftentimes try to go ahead and rank for or include those topics and keywords in their targeting because they do know that it’s a commonly misspelled phrase.
I think that there’s a lot of conversation around this. One of the things, Ray, you and I have seen is folks that have a really small marketing or search department that then end up having 5,000 keyword lists, which is almost impossible to really maintain from a marketing standpoint because you can’t possibly really well manage and maintain content with that many keywords if you’re a really small team. So if you put someone who’s really going to be about prioritization in terms of picking some keywords across various areas and starting to understand how you can really leverage those and building and growing as you continue to evolve as a team, but need to be able to show traction somewhere. I would rather be able to show traction across 100 keywords well than across 100,000 keywords and be like, “Look at all the keywords we’re tracking but not freaking ranking for.”
Ray: Yeah. One of the things that we’ve talked about quite a bit has been this idea of moving away from a keyword-based mentality into more topics because the thing that happens inevitably with keyword-based content when you create a new content targeting specific keywords is – the pro to that approach is you can get very specific estimates around how much search volume you’re going to get. The con to that is because of the way semantic search is starting to work, there’s no guarantee that even if you’re perfectly well-optimized and you’re getting the right backlinks and everything else for that based on what Google’s semantic-based algorithms are deciding, you’re not necessarily going to get the search result for that, as well.
We’re in this middle phase but there are many cases now where it can be useful to target more general topics that maybe have an aggregated search volume across a broad number of keywords. There’s the relevant, the overall topic at hand and creating content for that. That is useful when you’re trying to brainstorm new content creation ideas because you can usually think about much more interesting topics, your ideas for content when you’re targeting topics versus keywords anyway.
Erin: What’s interesting with semantic search in this conversation is that because semantic search is really trying to tie together intention and context, it’s using things like location – if I just search for coffee, it’s going to assume that my likely intention is that I want to go get some coffee and that it’s going to say, “You’re in such and such place.” So it’s going to serve up places I can get coffee in my specific location as opposed to telling me necessarily the Wikipedia entry for the origin of coffee, how coffee is grown, all that kinds of things like how to grow it myself. It’s not necessarily going to go there because it knows the people’s typical intention is to go find a cup of coffee.
This idea of intention is really interesting from a marketing standpoint because as semantic search continues to evolve and we see algorithm changes – this is idea of topics – so as it starts to get smarter about the reactions to topics that are location-based, inferences around topics, it starts to learn and understand really when you type something in, do you want to buy something in this area? Are you trying to research something in this area? Are you trying to share an idea around this area? What are you really trying to accomplish?
This is what we talked about last week. There are all these ideas around keywords. One of the things that I suggested was taking keywords and trying to apply them to different areas of what people traditionally look at as a marketing funnel. I think semantic search is a really good example of that because it’s talking about intention and context together, which is what the marketing funnel is supposed to be doing. It’s talking about where are you in the funnel with your intention towards this brand or product or whatever it is you’re trying to do, and contextually how do we talk to you about it?
Karen: I’m just wondering if you think that the best keywords for advertising are different from the best keywords for organic search.
Ray: That is an interesting question that has a lot of really detailed answers depending on context of what you’re talking about, as most interesting questions do. There’s no way to give a blanket answer.
Erin: I say yes. They’re different.
Karen: There’s a difference in opinion. Okay, what do you think, Erin?
Erin: I just want to be definitive about – I figure yes. I think it’s different mostly because in one form, you’re telling a paid advertising opportunity specifically what you want to match and it will rarely match with something that has nothing to do with that. Versus if you create content, whether or not you’re targeting keywords or not, you may be found for things whether inadvertently or not around a specific topic. So you could actually just go out and create really, really great content and be doing SEO. Whether you’re trying to or not, but with advertising you specifically have to do this. You have to pick some keywords and it’s going to serve your ad to people who choose those keywords.
Karen: So do you have any advice for optimizing search content for long-tail keywords?
Ray: Yeah. The interesting thing about this whole conversation is that we’re in this transition phase between keyword-based search and semantic search. Keyword-based search is still happening all the time and targeting and optimizing for it is still very effective – and it probably will be for a number of years. But the semantic behavior of search engines is continuing to improve from a technology perspective all the time.
What marketers are stuck with doing is they’re still targeting content for specific keywords the way they always have been and that would be the case whether you’re talking about these big head keywords with long-tail keywords. So the core basic organic search optimization principles still apply here.
That being said, there are things that you can do to start boosting your support for semantic search. There’s this thing called semantic market, which is basically in addition to the metadata that you are already including in your web pages for like your title tag, your meta description tag, and everything else.
There are a number of different formats out there that you can use. The most popular one probably is called Schema.org. It’s pretty well documented. If you go to Schema.org, there will be a ton of information about marking up your content. One of the things you might think about doing – we’ve seen this work really well for e-commerce companies – is start creating content and including that Schema.org metadata directly in your content. You should do that whether you’re dealing with long-tail part of your content or content that you think is going to target the big head of keywords that you’re going after.
Erin: It’s interesting you talk about doing a mark up of content. I think that this tends to scare a lot of people when you say things like we’re moving towards semantic-based search and with keyword things, a lot of people spend a lot of time and money setting things up. As always, change tends to terrify a lot of people. They’re worried they’re going to lose traffic, that they’d waste a lot of money, which to be fair, it’s not how it would work if you have actually created really good content that is targeted towards what people are really looking for.
The thing that we were talking about, those creating content for long-tail keyword, specifically long-tail keywords are a lot more specific around specific queries and terms, which is why there’s lower traffic volume on them. So if you’re creating content around these really highly specific things, chances are you’re not going to do too bad with getting traffic there.
Ray, do you want to touch on why people should not be terrified about the move to semantic search if they’ve been creating good content all along?
Ray: Yeah. Just keep doing that, basically. Start doing some research. But it’s not like there’s some new algorithm that you haven’t been already dealing with for the last two years. It’s not like things have just changed because we’re doing this show today, as much as I would like that to be the case.
Just keep doing what you’re doing. But start learning about semantic search, start understanding. There are some good books out there. I think one of them is actually called “Semantic Search.” Understand how to move. Understanding the basics of technology is going to be good for you in your career anyway as a marketer because it’s not going away.
Karen: How do you think that long-tail search, semantic search is going to start affecting findability and keyword metrics?
Ray: Erin, maybe you can talk about this. I’m actually having a hard time thinking of an example where semantic search would somehow behave differently for long-tail keywords than it would for larger keywords. Other than the fact that these big head keywords are basically – they might be a little bit more predictable around search volume because you’re dealing with well-known established keywords. The advantage, of course, to going after long-tail keywords is that they tend to be less competitive. You have to make that up in volume. The only difference there is if you start dealing with topics instead of actual keywords, then the way the search engines are going to index your site and potentially rank you might be different. But I would say, as of today given the technology today, I don’t really expect that to be a huge difference.
Erin: I like to theorize. Let me predict what I think is going to happen. Ages and ages in the future or months from now I can look back and laugh. But the idea to me around semantic search and the semantic web and specifically the search angle of trying to tie intention and context together is – it’s like big head keywords are going to, I think, fracture a little bit because it’s going to take into account things like location and previous search intention and other contextual things around your actions.
If I were to search for shoes where I am right now and Ray was searching shoes where he is right now, ideally we would end up eventually with two separate different things because your location and my location are different and your previous actions and my previous actions are different. So, for me, things will actually start to see a little bit of a fracturing and so you’ll have less big head things that millions and millions of people all see the exact same results. You already see that changing.
I know that Google try to make a lot of changes when it started doing Circles and Google+ and all this other stuff, plus it would serve up to you what people are talking about specifically in your realm and I think that this actually segues into the effect that social and conversational context has around how Google is looking at intention and context around what everybody is talking about. I think that there’s what’s called right now “long-tail keywords.”
I know Ray mentioned having to make up for it in volume. I’m actually not sure you’ll have to make up for it in volume, not much because it goes back to the original supposition that I would rather have 100 really relevant people trafficking a piece of content that actually would probably convert than 100,000 people accidentally ending up on my page because I gave her a Google AdWords spot or some keyword ranking stuff and then people just bounce out. So I’m kind of okay with lower volume and higher conversion. It’s probably cheaper anyway.
Karen: Do you think that the importance of social media presence for brands is going to be become more important because of semantic search?
Erin: It’s hard because people do social so horribly. People do a lot of things really horribly. Social is just a very obvious one of them because it’s so public when you just throw out awful things all the time and don’t really have conversations.
One of the things that we talked about in a previous episode was if you’re going to do social, don’t do this “dive bombing” thing where you just come in, you make one comment, and then you peace back out. That’s actually not really going to help unless you already have a really huge established following. There are people who go out and participate in one conversation one time and then when you search for something, that conversation and that one thing actually ranks really high because they participated.
So I think a lot of this is going to be about establishing like domain rank, getting a really valuable set of content or domain authority. Like building up domain authority, I think that’s there’s social authority. Building up a following context around content and topics that you are relatively known for. Who used to do a really good job at this that they were bought but it was like Klout was talking about what topics do you really know about, what topics do you have a lot of authority on, and then assigning an authority rank about a topic to you. I think that that’s kind of similar to what Google could do in order to tie social more contextually into search results for people.
Ray: Yeah. I would agree with that. It’s a really good way of putting it.
Karen: How do you think semantic search makes ranking less technical and more content dependent then?
Ray: Again, it’s never going to be less technical versus more content. The technology side of things is always going to be really important because due to pretty much every marketer, there’s always going to be very strong technical side of things to this and making sure you’re indexed properly – we talked about semantic mark up earlier – all that stuff is going to matter probably even more going forward. In addition to that, you’re going to have to figure out the way to translate the actions and conversions that you’re trying to get your customers to engage with into this kind of more generalized topics.
One of the things that’s been very convenient about keyword-based searching has been – usually when you are searching for something – we talked about this purchase intent or commercial intent about keywords I think last week – the difference between saying recommended pair of shoes which turns into, as you move down the potential conversion funnel into things like best price on Nike running shoes where you’re already ready to buy a pair of Nike running shoes and you’re just looking for the best price.
Those are things that make it a little bit more muddy when we start talking about semantic search because it’s teasing out the commercial intent from this more entity-based, context-based search is probably a little bit harder to target from a content perspective. The search engines won’t have any trouble targeting it because they, in return, have a lot more information about where you are, what device you’re on, what you’re doing, what your searching habit has been. So they’re going to know a lot more about you than you will as a marketer.
But the challenge is going to be for you as a marketer to stay relevant in the midst of all that’s going on. So it’s kind of a new challenge. There is no definitive answer in a lot of this stuff but the thing I would just recommend is continue to pay attention. Stay tuned with us. We got a lot more about this stuff, too.
Erin: One of the interesting things about this competition between technical SEO versus content SEO is a lot of the stuff is legitimately the same thing and has really strong ties. So the way that I like to describe it is, especially if you’re a marketer that has communications and journalism background, if you remember having to write papers back in the day like I did and you actually physically had to write stuff down and go to a library and look up stuff, and cite your sources and do these things – that’s much the same as the more technical aspects of some SEO things, which is you want to make sure that you title pages appropriately.
If you’re going to write a report, title the report about something that has to do with the report. If you’re going to cite your sources, you have to make sure that if you click underlined resources, you reference them in the end and do all these different things. A lot of these is the same way that the Internet is working from a search perspective and so it’s looking at links to and from your content. Much the same as citing sources in a research paper, and then somebody might cite your research paper and their research paper.
I think that if you have ever had to put together a really big strong report and had to do any sort of referencing, had to put titles on pages, had to subsection things, do these kinds of things, you can look at a lot of what has to do with creating content on the web and building up those technical things as basically just the online version of Document Building 101.
Ray: Yeah. Yeah. Definitely.
Karen: That’s all the questions I have for today. Is there something either of you want to add in or some takeaways we can leave marketers with today?
Ray: No. I think that was a good discussion. It’s one of those discussions that can get really deep very quickly and I think that marketers are going to be looking for the biggest takeaways on some of the stuff.
From my perspective, the biggest takeaways today are: keep paying attention, keep researching, keep reading about this stuff, start thinking about semantic search, start thinking about some of the technical sides of things like mark up that we talked about, and just pay attention to things because they’re going to continue to evolve.
Erin: Totally. If you want a specific keyword takeaway from me, I would say that you want to get a good mix throughout the keyword spectrum of things to target. Again, take your team size into account when you’re looking at the actual volume of things that you’re really trying to track and create content around and manage from a marketing perspective.
Also this idea of topics, you really want to group keywords into themes or topic groups because this is going to help you really start to measure the efficacy of different topic groups like marketing themes or campaigns or product lines or features or whatever, however you want to group them so you can do things multiple different ways.
It’s one of the things we do at Ginza. I always try to stress the importance of, create keyword groups and create content groups. This is going to help you very much understand marketships and how your content is starting to perform. Not just how your content performs one on one with your audience but how your content performs with the search engines. I think that’s an easy tip that you could do today that’ll probably help you out in the future.
Karen: Great. Hopefully, this conversation will blow up my inbox again. Please feel free to contact me, firstname.lastname@example.org or Erin or Ray, all of us at GinzaMetrics.com. We’re here to answer your questions. And if you have any great burning questions you like us to address on future FOUND Fridays, we’re happy to hear that, too.
Thank you, guys.
Erin: Thanks, everyone.
Karen: See you later.