Oct 29, 2011

How Big is Your Long Tail Keyword?



Choosing keywords to optimize for is a tricky business, made all the more tricky as keyphrases grow longer than a couple of words. As Google has said, up to 20% of search queries in any given day are completely unique. Should you try to optimize your tauntaun sleeping bags product page for "tauntaun sleeping bag," for "childrens' tauntaun sleeping bag," or for "childrens' star wars tauntaun sleeping bag from hoth"? How can you research whether or not to optimize for such a long tail query?

This week we're asking the question: How big is your long tail? No innuendo intended. This is a totally serious question for the search world, wink wink, nod nod, say no more.

Many of you are familiar with the fact that the world of search is really dominated by this concept of the long tail. Google talks about this incredible metric that 20% of any search that's performed every day is completely unique. Google has never seen that search before performed on their engine at all. No one in history has ever made that search. That happens on one out of every five queries every single day.


We are amazingly unique creatures, especially when we get in front of a search box. That's a great thing, but of course it means that doing keyword research can be tremendously tough. There are a lot of folks who ask the question: "I've heard of this tail concept, but I only do keyword targeting and keyword research really on the fat head, maybe the chunky middle." I'll talk about those in a sec. "I don't even know how to do keyword research on the long tail. I don't know how much of an opportunity it is."

This Whiteboard Friday is here to answer that last question: How big of an opportunity it is. Can we measure it? Can we look at the size? Can we understand? Because some industries are going to be very narrowly focused on a few head terms. That's what people search for. Those are the money terms. That's where people convert, that's where the value comes from. In other industries, the long tail is a huge, huge win, and you need to be able to understand that in order to do the right kinds of keyword targeting.

So let's begin. This is our classic long tail graph. We've got the quantity of visits that any particular keyword sends you on this axis, and then down here on this axis, the keywords themselves. This keyword sent a ton of visits. This keyword sends a ton of visits. This keyword sends a bunch of visits. Then there's this huge tail that comprises usually 70% of all of the quantity. If we were to take this area under the graph, do some calculus, figure out how big the whole opportunity is, oftentimes the tail is 60%, 70% of the full opportunity. It's because it extends for miles and miles and miles in that direction.

We kind of classify these into three chunks. So we have our fat head, our chunky middle, and our long tail. The fat head, in my view, tends to refer to the things that are very popular in your niche. I say in your niche, because depending on your niche, these may be very different in terms of quantity. Usually the categories we like to bring them into are something that sends more than 100 visitors each month. If there's a keyword that's sending us more than 100 visits a month, we put that in the fat head. That's sort of a big term for us. If there's something sending between 10 and 99 visits a month, that's our chunky middle. If it's sending fewer than 10, it's our long tail.

Some SEOs like to have very, very different orders of magnitude on these. Some people might say, "This is only things that send over 1000. This is stuff between 20 and 500. This is stuff that's only less than 5." Whatever you want to do is fine. You can classify your traffic that way. That's a good way to go. You should just be aware that this classification system exists. I think this is a very healthy way to be able to look at things.

It's harder in the long tail. I might have a subset of these that I'm monitoring as well just to get a sample, but I'm generally not paying attention one-on-one to them. The problem can be when as SEOs we naturally, since we're paying attention to these keywords we manage and rank track, we get obsessed with them. We stop thinking and worrying about the long tail and the opportunity we're missing here. Meanwhile, one of our competitors is going, "Yeah, yeah, yeah. Go ahead and let him win the number two, number one rankings for those keywords. I'm winning over here where there's no competition, and where there's generally higher conversions, and where there's tons more volume." We're kicking ourselves when we find that out.

Instead of losing out, let's figure out how much opportunity exists there. Before we answer that, we should know: Well, how much am I currently capturing of these? This is pretty easy. What you can do is you can create advanced segments inside Google Analytics, or you can create segmentation inside Google Analytics for each of these buckets. You decide how big these buckets are. You can say, "I only want keywords that sent me fewer than 10 visits." That works great. Those segments can then be classified and you can say, "All right. This sent me 13,510 visits last month. That is up from the month before, so I'm sort of doing better in my long tail." Or long tail demand's getting better, whichever. From that, measure the quantity of keywords and visits in each bucket.

You can also measure the quality. Measure quality with one of two things. If you have goals set up, hopefully you have goals and conversion rates set up in your Analytics, that's a great way to look.

The other way, if you don't, if you're just trying to say, "How much is this traffic worth to me from a branding perspective, from a usefulness perspective, from a reaching new audience perspective?" The metric I really like for that is browse rate. The reason I like it so much is because browse rate says on average how many pages did a visitor visit in a single session when they came via this keyword.

Browse rate is great way to say usually, when you have a higher browse rate, that means more engagement. It means someone's surfing around your site more. They're spending more time on the site. They are more likely to convert or come back and convert. Browse rate is a good sort of substitute metric. It's not great, not perfect, not nearly as good as goals and conversions, but if you don't have that, browse rate is a great way to judge qualitatively: How's this traffic performing?

You take that and you sort of go, "All right. This is how well I'm doing. This is my trend over time. Am I improving? Am I not improving?" For some people they want to know, "Yeah, but what's the opportunity? Am I really missing out here, or am I doing a good job?" Just measuring your own progress won't tell you that. You need broader industry statistics. There's a number of ways to do that.

The most obvious one, of course, is to go to Google AdWords and try and figure out what the fat head and chunky middle distribution looks like. But because there's no real keyword research available for the long tail, Google through their AdWords tool or their AdWords API, or Bing through AdCenter, are not generally showing you keywords that send fewer than 10 searches, that have a very small search quantity, but there are tons of those keywords. That's a really challenging thing to search. Of course, no search engine's going to be able to tell you what those 20% of queries that they've never seen before every day are. So that's frustrating too. Solution to this . . . it's not that ingenious, you probably know how to do it, you can probably guess, but let me walk you through it anyway. I think it's super exciting.

This is asking how much opportunity do you have. Oftentimes it's a lot. One of the best ways to figure this out - this won't answer it perfectly, but it does a nice decent job - is to say, "I'm going to go to AdWords or AdCenter and I'm going to create a paid search campaign." This is one of the times when paid search and organic search overlap very, very well. It's just because the research is so handy.

For my major fat head terms, I'm going to create campaigns around those and target groups in both the exact and broad matches for those keywords: Exact match, of course, the example would be like "chess tournaments." "Chess tournaments" is an exact match. I only want that precise phrase. Google do not show my ad and don't tell me about that. I want to know only the clicks that I get for "chess tournaments" exactly.

Then I'm going to create another group that contains phrase based matching. Show me phrases that contain "chess tournaments" but not this exact phrase. Meaning things like, if somebody searched for "playing chess in a tournament in Miami." That would show up in . . . well, yeah, it contains the phrase, but it's not that exact phrase. Or I'm sorry, "Miami chess tournaments" would be in there. "Minneapolis chess tournaments" would be in there. "Pro chess tournaments" would be in there. It's not the exact phrase "chess tournaments", but it contains that phrase.

Then you have that final bucket of contains the words but is not matching a high-volume phrase and does not necessarily need to be the exact phrase. This could be "tournament style chess games" or "tournament video game chess". All this type of stuff can add up to a bunch, and what will happen when you buy these keywords in AdWords is that they will show you something called impression count.

The impression count can actually be drilled into and you can see all of the terms that send those. From that impression count, you can then take them and segment them into these buckets. You can say, "Oh, okay. We had 500 keywords that had 10 or fewer impressions, so we're going to put those 500 keywords in here. Then we had another 110 keywords that fit into our chunky middle stats. We had another 42 that fit into the head."

This kind of distribution is incredibly valuable because it gives you a sense for a phrase or a bunch of phrases, if you're doing this work consistently, how big is the opportunity in the tail? It really does vary. Some things are hyper-geographic, so geographic modifiers get in there. Some things are very tuned to customization and specialization and weird sorts of searches. This happens a lot in the programming world. You'll see "mySQL calls" have a bunch of volume and then you might say "PHP mySQL calls", and then there's a ton of long tail weird stuff. Being able to see that versus something that's much more narrow is really, really handy and where the volume is constrained or confined to the fat head, chunky middle.

Now that you've got this process, if you've got some budget for AdWords, you can start testing, grouping these things into buckets. You can measure your buckets over time and how you're performing and you can see: Am I capturing the long tail? Or am I losing out on an opportunity to capture the long tail and maybe I need to be spending a little less time and attention with the fat head and chunky middle?

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.