Search Engine Optimization

  • Google Hummingbird Update & Secure Searches

    Google Hummingbird Update

    It’s been a big month for Google with two big changes happening.  The first was the announcement of the Google Hummingbird update, or a big change to their search algorithm.  The second involves secure search: Google is now hiding ALL search query data in Google Analytics by forcing a secure search, whether you are logged into Google or not.  For most of us in the search marketing world, it’s business as usual, except now we’re going to have to get creative on getting keyword data, and continue to build on long-tail search queries in content.

    Part 1: Google Hummingbird Update

    This was probably the most extensive update since the “Caffeine” update in 2009, and also the most secretive.  Normally, Google gives signs that a new algorithm will be deployed, except this time they kept it under wraps quite well for over a month and “flipped the switch” last week.

    We’ve been saying for a long time that your website should have content that answers people’s questions, and long-tail keywords are a great solution for this.  For example, ask your sales team what the most typical questions are for inbound leads and build a content system around that.  Google’s new algorithm is built to handle long-tail “conversational” search queries, like “where is the nearest place to buy baby bedding” where traditionally Google might try to find a page for “baby bedding” like Amazon.  However, now Google might know your actual location (if you’ve shared that), or the meaning behind certain phrases like “place” to be a brick and mortar store and that you’re not looking for an online store.  Ideally, you would get a different set of search results from your home versus from work.

    Check out this article from SearchEngineWatch titled “The Resurgence of the Long-Tail Keywords in SEO“.

    Why did Google come out with this change?

    Google is always looking for ways to increase conversion rates, and typically, long-tail searches lead to conversions.  It is estimated that 20% of the 3 billion searches per day on Google are unique queries, to give you a sense of how big the long-tail really is.

    Will this affect my rankings?

    Probably not.  Most large publishers are not complaining about a loss of rankings yet, and this update was really targeted towards the more complex search queries.

    What should I do about this change?

    You can (and should) continue to add content to your website.  Make the content informative, and use Google Suggest and other tools to get ideas around long-tail options.  Here is a great post from Avanash Kaushik on the subject of how to monetize long-tail search.

    Part 2: Google Keyword Queries Hidden from Google Analytics

    secure search in google analytics

    Starting in October 2011, Google began encrypting searches for folks who were logged into Google (like GMail, for example), so none of the keyword query data was being passed into Google Analytics.  Back then, it may have accounted for 10-20% of the keyword report.  Last month it was as high as 80%, and now will be 100% going forward.

    Google says this is part of a larger “user privacy” initiative, but interestingly, the keyword data is already anonymous in Analytics, is accessible for paid advertisers in the AdWords platform, and can be seen in Google Webmaster Tools.  This begs the question: is this really about privacy, or is it a ploy to get more advertisers on-board?

    In fact, some of these tactics of “Unlocking Your ‘Not Provided’ Keywords in Google Analytics” are spelled out by KISS Metrics.

    What does this mean for SEO?

    It means that Google has closed the door on a critical component marketers use to determine the value of their SEO program.  One of the leading indicators we use for evaluating the performance of an SEO campaign is to compare the branded and non-branded searches.  Now, instead of relying on keywords and specific keyword combinations, we’re going to need to focus on the content that provides real value to customers and prospects.

    Here is another great post by Avanash Kaushik and some things you can do to perform secure search data analysis.

    September

    30

  • Internet Marketing Prediction Modeling

    Over the past couple of years we’ve been working with lots of customers in various industries from all over the world.  We eventually gathered enough data to be able to reasonably predict your success based on the products you purchase from us.  In short, businesses want to know how their Internet marketing spend is going to help grow their business.  At the end of the day, it’s not just about rankings, or just about traffic increases.  We see lots of Internet marketing companies pitch the importance around these “vanity metrics” – but did the increase in traffic result in any leads, and did any of those leads result in new business (i.e. paying customers)?  So, we had to create the Internet marketing prediction modeling tool.

    Those of us in marketing would all love to believe that every lead we send is qualified – but the reality is they are not.  Sure, there are keywords with more “commercial intent” than others, but this is a tough road with SEO.  Google is giving more and more preference to educational and self-help content in organic results, and pushing commercial content further down.  Why?  They want you to buy Adwords for that.  That’s why almost every Internet marketing campaign requires a mix of SEO to help generate educational, thought leadership content that can generate visitors we can “nurture”, and compliment that with a PPC campaign for those who are ready to “buy now”.

    So while our prediction tool incudes things like traffic increases and potential lead flow, we also have to account for things like:

    - Qualified lead % – what percentage of the total leads generated are really good leads?  We assume 5-10%.

    - Internal close rate – Is your sales team on the ball, or letting leads fall through the cracks?  Your customer acquisition cost depends on this.

    - What is the customer willing to pay for a lead, or even a new customer?  This requires knowledge about what the lifetime value of the customer is, and what makes sense from a lead cost standpoint.

    - What is the current conversion rate of the website?  Well, that’s sometimes a loaded question.  Most clients are not really tracking that, or are not tracking enough channels to get an accurate number.  We might assume somewhere in the 2-4% which would include calls and form submissions – maybe higher if you have other avenues such as newsletters, free trials or free downloads.

    How Internet Marketing Prediction Modeling Works

    We took actual analytics data from nearly 50 customers and tracked their KPI’s over a two-year period.  We normalized the data and came up with a linear regression model of how just about any client would react to any of our plans given a) current traffic; b) conversion rates; and c) market competition.

    Then, with our add-on products like call tracking, retargeting and a/b testing, we’re able to create a conservative assumption of how the conversion rate would be affected by these, which slowly improves the lead flow in the model over time.

    Finally, we built the model to work backwards from a client’s total budget.  Once we choose the recommended services, what’s left over will go towards the click expenditures, resulting in something that looks like this:

    Screen Shot 2013-05-30 at 10.42.29 PM

     

    From here we can easily do a sensitivity analysis and see how the lead flow and cost/lead are affected by changing the marketing mix.  Want less SEO and more PPC spend?  No problem, lets see how that changes things.

    Screen Shot 2013-05-30 at 10.44.20 PM

     

    We can also play with the model to get the client into the right cost per acquisition “zone”, and so we can indicate how much we are over or under their acquisition cost expectation.

    Testing Against the Model

    This model has been in play for only about 2 months now, and what we’re doing is using it to compare current customers who are maybe in their 5th or 6th month to see  how close we are to the prediction (we’re able to plot results out for traffic and leads for 12 months in advance), and so far we’re between 75 and 85% accurate.  As we add more data to the core analytics, I think we’ll be able to improve our accuracy, and even perhaps get enough data to model against certain industries, such as medical, law, technology, etc.

    Want to learn more, or get a demo?  Call us at (888) 427-2178 and let us help!

     

     

    May

    31

 
 
 
 

Have Questions? Give us a call (888) 427-2178

PearBook

facebook.com/pearanalytics
  • How to play the long game with your SEO strategy, with tips from the experts at SMX West 2014....
  • Scientific social media guidelines for the ideal lengths of tweets, Facebook posts, titles and everything in between....
  • If you have 30 mins, here is an interview you might enjoy....

PearTweets

@pearanalytics