I’m going to be honest with you. I think rap music sucks in other languages. I’ve been around the world and met a lot of talented emcees in various different countries and I’ve listened to people rap in Czech, Japanese, German and countless other languages that I can’t speak. I might smile and nod my head just to be polite but – I don’t like it. What makes rapping in English so awesome is the elegance of the English language and its beautiful ambiguity. However for Search Marketing that same ambiguity is a gift and a curse. I’ve invited a poet we all know and love to help explain how to conquer that ambiguity using A/B testing. He’s rhymed his way into saving the day before. No it’s not Johnnie Cochran.
Ladies and Gentlemen I’d like to reintroduce you to the one and only Cat in the Hat!
Like all inbound marketers the Cat in the Hat is no stranger to juggling many different talents simultaneously to get results. He’s also no stranger to navigating chaos to end up with spotless results. However he might be more fantastic than his inbound marketing counterparts because he can walk into anyone’s house wearing just a hat and a bowtie (that doesn’t actually go around his neck) and cause a ruckus without any repercussions.
Even so he’s taking time out of his busy schedule of home wrecking and cleaning to kick a few rhymes and help us argue semantics. Luckily he rhymes in English today.
Classical Keyword Classification
Typically in SEO we classify keywords into three groups: “informational,” “transactional,” and “navigational.” Briefly, informational queries are broad queries such as “plumbers” or “flowers” in which the user is just looking to find out more about the topic in a very general sense. Once the user is more informed and is ready to make a purchase the query becomes more specific and grows into something like “certified plumbers 19121” or “buy flowers Philadelphia .“ When this happens it’s called a transactional query and if your landing pages are great then your clients start thinking about naming their kids after you when these keywords get typed in. Lastly, navigational queries are for those strange people that go to Yahoo and search for “Google.” Weirdos.
This segmentation works on a broad scale but what happens if you are a store located in Reading, Pennsylvania and you sell both eyeglasses and drinking glasses? In both cases people may potentially search for “reading glasses” however this search is ambiguous and you don’t know what portion of those people is close to a prescription for medical marijuana and what portion is just trying to enjoy some lemonade. Stop it, I don’t smoke.
It would be a shame to put all that effort into building content, optimizing and link building to achieve the number one spot for “reading glasses” only to have an 80% bounce rate because you chose to highlight the product with the lesser demand.
So let’s find out how we optimize for the right term.
Know Your Audience, Know Your Keywords
Until our world starts to look more like something George Orwell and Philip K. Dick would have collaborated on there’s no way for us to know exactly what every user is thinking when they search or come to a site. However we can use social listening to figure out what words they are using to describe products, group users into personas and also determine their needs at each stage of search.
There are many awesome tools that can be used to tap into the existing conversation. You can also use mechanisms such as sponsored trending topics to artificially encourage people to talk about your brand or keyword. However in most cases it’s good enough to just pop your keyword into Social Mention and sift through the how people are talking about it.
Questions to Ask
- What words/phrases are used to describe the product?
- Would one of these words/phrases make a better target keyword for Organic Search?
- Why are these users searching? What needs are the trying to fulfill?
- What are the demographics of these users?
After cataloging this information you are ready to develop the different personas.
Personas are models of people that fit a specific ideal demographic. Depending how in-depth you want to go you can give each persona a name, location, ethnicity, the whole nine if you’d like. The point of a persona is to determine the target audience and use them as a reference point. Personas are typically employed in usability tests but as you may imagine their implications are far-reaching when applied to Search. There are plenty of tools that give you demographic data like Compete, ComScore, Quantcast and Google Ad Planner but in most cases social listening by itself is better because it gives you real conversation from real people. Adwords also allows for demographic targeting.
Keep in mind that developing personas is not an exact science and you may never represent all the groups that are visiting your site. Think of personas as great guesses that help define the parameters of tests against the performance of your landing pages.
Need States are classifications of goals that users are trying to fulfill when searching for a given keyword. Typically need states follows the AIDA Decision Making Process (Awareness, Interest, Decision, Action) although I have also seen it divided into Consider, Evaluate, Buy, Enjoy, Advocate (CEBEA) as well. Both systems are great but whichever model you follow it should be applied throughout your keyword list to improve messaging for the target demographics’ need state.
For example since “reading glasses” could mean that our friend Fisher Price is performing an informational search about eyewear the term could be classified as interest (AIDA) or consider (CEBEA). However “reading glasses” could also mean that Fisher is looking to buy eyeglasses or drinking glasses in Reading, Pennsylvania and therefore that search for this persona could also be Action (AIDA) or Buy (CEBEA). Maybe it’s the first. Maybe it’s the second.
Which is it?
Point out the Bounce!
Once you’ve identified keywords that fall into multiple need states you then have something worth A/B testing. In this case we have 3 things to test but we can easily accomplish them with an A/B test.
The questions that need to be answered are:
- Is Fisher Price searching for information on eyeglasses?
- Is Fisher Price looking to buy eyeglasses in Reading?
- Is Fisher Price looking to buy drinking glasses in Reading?
Luckily the Cat in the Hat always has a few friends that are ready for an A/B Test.
Thing 1 and Thing 2 will represent our landing pages. Notice how aside from the numbers on their shirts they are otherwise identical. Build your landing pages the same way. Make sure the pages are laid out identically aside from the specific info and pictures for the products in order to eliminate variables. In this case both pages would contain information about their respective glasses and both would have CTAs leading to product detail or purchase pages.
Once you’ve got your landing page in Unbounce set up an Adwords (or other PPC provider) campaign and allocate 50% of the traffic to each page and let ‘er rip. You want a big enough data set to make the test definitive so I’ll let you be the judge of that based on how much this keyword is costing you.
And the Winner is…
The keyword that performs the best in the A/B test wins. That is to say the landing page with the lower bounce rate, higher time on site and highest number of pageviews is the owner of that keyword for your Organic Search campaign.
Thing 1 “drinking glasses” has won rights to this keyword therefore in this fictional case the insight is that more people to purchase drinking glasses in Reading, PA than are looking for information about or to purchase reading eyeglasses. Once it’s definitive that drinking glasses owns the “reading glasses” keyword the next step is to optimize the landing page and the internal linking structure for Organic Search to reflect that and then do subsequent research to identify an additional keyword so your eyeglass stock doesn’t collect dust.
Here’s a quick reminder of the process:
In closing, it’s important to remember that the results of these inbound marketing processes are not constants. SEO and CRO actually create a feedback loop and keyword research must be updated to account for seasonality and other user behavior changes. In other words, for the best results … Always Be Testing!
Before I get out of here I want to give credit where it’s due. A brilliant man by the name of Tony Effik introduced me to the idea of personas and need states and how they help eliminate the ambiguity of keywords from the keyword research process. So I have to give him his props because otherwise I wouldn’t have known this to write it. Oh yeah and he’s my boss too.
Also I’d like to give a shout out to Dr. Seuss because “Green Eggs and Ham” was the first book I ever read and reading is fundamental.
Pleasure working with you!