Google Analytics is a free, very robust web analytics tool that I’d recommend for implementations both large and small.  In my opinion, though, one of GA’s greatest strengths also serves as one of its greatest weaknesses: it’s so easy to use.  A slick, AJAX-y interface can have the dangerous effect of sending our brains into autopilot, which in turn leads to lazy interpretation and analysis.  In order to truly understand the metrics, we have to see past the pretty graphs and ask ourselves questions like, how was this data collected? and what equation defines this metric? Today let’s talk about a metric that is often insufficiently understood: referring source.

Referring source is one of the most useful reports in Google Analytics.  This single report gives us insight into both quantity and quality–how many visits and what types of behavior a traffic source drives.  Here’s an example of the referring source report from Searchability’s Google Analytics account:

At first glance, this report seems dummy-proof.  But here’s where understanding of this report typically breaks down.  See if you can answer this question with an unwavering degree of confidence: does referring source persist across multiple sessions, or is it reset upon each visit?

This is a question that deserves investigation, as the answer carries weighty implications.  Consider a scenario in which a person visits your site via a banner ad, leaves without making a purchase, and then reaches your site via a search engine two days later, this time making a purchase.  To which referring source will Google Analytics attribute the purchase, banner advertising or search?

As it turns out, referring source persists across sessions.  Below is a statement from Google’s documentation on Google Analytics and cookies:

When visitors reach your site via a search engine result, a direct link, or an ad that links to your page, Google Analytics stores the type of referral information in a cookie. The parameters in the cookie value string are parsed and sent in the GIF Request (in the utmcc variable). The expiration date for the cookie is set as 6 months into the future. This cookie gets updated with each subsequent page view to your site; thus it is used to determine visitor navigation within your site.

To reiterate from the citation, the cookie containing referral information persists for six months.  So, unless your visitors delete their cookies, their initial referring source will persist across any subsequent visit within a six month period.

How will this affect how you do web analytics?  If holistic referral attribution analysis is critical to your business, you will need to consider a method of capturing a visitor’s most recent referring source.  One possible solution is capturing the HTTP_REFERER in a session variable.

I first began questioning the nature of referral persistence when I was managing online media and analytics for a small website.  We had stopped running a particular banner advertisement, and yet we were still seeing several visits a day from that banner.  The explanation, of course, is that these were returning visitors who had initially been referred by the banner.

If you found this post helpful, link to me or share it on your favorite social bookmarking site using the links below!

OK, so it’s not really shocking at all.

Despite the comparably greater amount of press activity surrounding the spinach E. coli poisioning in 2006 (see bottom row of chart for news reference volume), America was much more bent out of shape over the peanut butter outbreak a few months later.  We weren’t any less upset over this year’s incident.  After all, who on earth would want to eat a jelly sandwich?

If you take pride in the number of times a day you avoid reaching for the mouse on account of your superhuman shortcut skills, get excited: now there are keyboard shortcut commands for Google search!

Scroll all the way to the bottom of a search results page and you’ll see a link labeled “Try Google Experimental.”  Google gives you the opportunity to participate in a number of interface BETA tests, including two keyboard shortcut experiments.

Experiment 1: Keyboard shortcuts.  Join this experiment and turn your keyboard into a Google search navigation control panel.

• J Selects the next result.
• K Selects the previous result.
• O Opens the selected result.
• ENTER Opens the selected result.
• / Puts the cursor in the search box.
• ESC Removes the cursor from the search box.

Experiment 2: Accessible view.  This experiment gives you an extended set of keyboard controls, including the ability to magnify the selected result.

• j or DOWN Selects the next item.
• k or UP Selects the previous item.
• l or RIGHT Moves to the next category (results, sponsored links, refinements).
• h or LEFT Moves to the previous category (results, sponsored links, refinements).
• ENTER Opens the selected result.
• / Puts the cursor in the search box.
• n Moves to the next result, and fetches more results if necessary.
• p Moves to the previous result, reloading earlier results if necessary.
• = Magnifies current item
• - Shrinks current item
• A Switches to Accessible Search Results
• W Switches to regular Web Search Results

Try these for yourself at http://www.google.com/experimental.

Offering advanced targeting capabilities at a relatively low cost, Facebook’s ad platform can be a highly efficient means of reaching your target audience.  When setting up a Facebook ad campaign, you’ll have to choose between two pricing models: CPM (cost-per-thousand-impressions) and CPC (cost-per-click).  In theory, your choice of pricing models would simply depend on the goal of your campaign.  However, depending on the competitive nature of your niche, your decision may not be that simple.  Ultimately, you need to determine for yourself which model makes the most sense for your business–and that will require testing.

Facebook’s original intent in offering two pricing options was to allow advertisers to essentially pay for “goal completions.”  If your goal was to drive click-throughs to your website, you’d bid on clicks.  And if your objective was general exposure, you’d bid on impressions (read Facebook’s official statement).

Let’s say you were running a brand awareness campaign, the goal of which being general exposure, or ad impressions.  While the average advertiser would choose CPM without a second thought, you are not the average advertiser.  Your marketing savvy leads you to perform a test in which you run two versions of your ad, identical in every way except for the pricing model.

Mathematically speaking, what circumstances could prove the CPC model more successful for your brand awareness campaign?  Ultimately, if the following expression holds true:

I*CPC*CTR < (I/1000)*CPM, where

• I = impressions
• CPC = minimum CPC bid required for Facebook to run your ad
• CPM = minimum CPM bid required for Facebook to run your ad
• CTR = target audience’s propensity to click on your ad (fancy way of saying click-thru rate)

The expression could be satisfied if CPC and CTR are low or if CPM is high.  This has proved true in campaigns that I’ve managed.  There were relatively few competitors, and so CPC was low.  Furthermore, the CTR on my ad was relatively low.  The result was millions of inexpensive impressions.

You’ll only know if you test, and the Facebook ad platform is excellent testing ground, being both inexpensive and easy to use.  So go run an experiment.  It’s fun, and it could save you precious advertising dollars.

CTR can be a deceptive little metric in your paid search analytics.  Too often I’ve seen analysts base their entire optimization strategy on this single acronym–unfortunately, to the detriment of the campaign.  Let’s talk about what CTR is, how it is commonly misused in optimization, and how it should be interpreted in analysis.

What is CTR?

Standing for “click-thru ratio,” CTR represents the percentage of ad viewers who clicked on your ad.  It is calculated as follows:

CTR = (# of viewers who clicked on your ad) / (total # of ad impressions served up)

So each ad in your PPC campaign has its own CTR.  You can calculate a campaign-wide CTR by aggregating total clicks and total impressions.

How is CTR often misused?

CTR is often wrongly-used as the primary decision-making metric in optimization (persisting, removing, or changing ads, ad groups, budget controls, etc.).  CTR is such a catchy, convenient metric, but most of the time, it should not be the basis of your optimization decisions.  Here’s why.

PPC campaigns can have a variety of objectives (and it’s critical that you understand what your objective is).  More often than not, PPC has a direct response objective (buy a widget, call a phone number, request a quote, sign up for a newsletter).  Driving qualified traffic, then, means driving traffic that has a high propensity to complete the desired action.  So in deciding which ad copy is the most effective, or which ad group drives the highest quality traffic, you’d look at the conversion outcomes.

While that may seem obvious, it is not what many marketers do.  Instead, many look to CTR to give them answers.  Let’s consider an oversimplified example to illustrate why this can be problematic.  You sell red tennis shoes.  You run an ad with a headline, “Cheap Tennis Shoes!”  The ad has an extraordinary CTR–almost 10%!  Your ad with the headline “Cheap Red Tennis Shoes!” has a much lower CTR.  But which ad do you think drove the highest quality traffic, and consequently, produced the better return on investment?

If you’re new to PPC campaign management, beware, as text ad networks (particularly, Google) will often set your campaign to optimize to click-thru ratio by default.  This benefits no one but the ad distributor, as they are compensated based on clicks.

How should CTR be used in analysis?

While CTR generally shouldn’t be your ultimate decision-maker, it is by no means unimportant.  CTR can help you make the most out of your ads in the following ways.

• Targeting strategy.  If your click-thrus are extremely low, you might want to consider tightening up your targeting strategy.  Perhaps you’re casting the net too wide.
• Ad copy.  If you’ve found a highly-profitable niche target via one of your ad groups, analyze the CTR for each ad in the ad group.  If some of your ads have lower CTRs than others, run a test, and pause these ads.  You might see your traffic AND total conversions go up.
• Ad position.  There are many variables contribute to your ad’s position (including your bid and your landing page quality score), and CTR is one of them.  It makes sense, as the ad distributor has an incentive to reward high CTRs (higher CTR=more clicks=more ad revenue).  If you’re unsatisfied with your ad positions, consider your CTR.  Rewriting your ad may prevent you from having to raise your bid.

Best of luck in your pay-per-click campaign optimizing!

Interestingly, AdAge just posted an article today (Who Had the Better Media Strategy? CareerBuilder vs. Monster) evaluating the marketing tactics of Monster.com and CareerBuilder.com.  The author asserts that Career Builder’s efforts have been slightly more successful than Monster’s.

One fact the author states is that Monster’s 2008 advertising budget was larger than Career Builder’s–2.5x larger, to be exact.  I would expect, then, that Monster would be receiving a higher volume of website traffic.  Interestingly, traffic estimates from Compete.com suggest that Monster is the leader in visitor share.  Refer to my last post to see the graph: http://www.searchabilityinc.com/2009/03/11/recessionomics-part-i-online-job-searching/

If Compete’s estimates are in fact accurate, I find the following two things interesting:

• Estimated traffic to Monster.com in 2008 did exceed that of CareerBuilder.com, but not by 250%.  In other words, Monster’s ad dollars were less efficient in driving website visitors.
• Oddly, estimated web traffic for Monster.com and CareerBuilder.com in February has an uncanny correspondence to 2008 ad spend.  Monster.com: $100K in ad spend, 100K visitors; CareerBuilder.com: $40K in ad spend, 40K visitors.

A surging U.S. unemployment rate is forcing millions of Americans into a new full-time job: job searching.  How are these job seekers using the internet as a tool in their quest for employment?

1. Search Trends for “Jobs”

The internet has been a key resource in job searching for quite some time now, with search volume for “jobs” being steadily high from 2004 to 2007 (see below chart).  You’ll notice in 2008 a gradual climb, consistent with the belief that the recession began taking its toll in early in 2008.  Then at the end of the year, when recognition of the nation’s poor economic condition became mainstream, searches for “jobs” nearly doubled.

2. Job Search Websites

There’s no shortage of job search engines, from niche industry sites to wide-reaching portals (Monster, Career Builder, etc.).  According to Compete.com, Monster.com has been the most successful in reaching job seekers in what were arguably the hardest-hitting months of the recession.  Not only is Monster the unchallenged leader in market share, they were also the only well-known job site that saw an increase in traffic in the month of February.  They must be following the old marketing rule of thumb: advertise during a recession!

More Recessionomics posts to come…stay tuned!

UPDATE: see a continuation of this post, written in response to AdAge’s “Who Had the Better Media Strategy? CareerBuilder vs. Monster”.

A colleague recently asked me what my thoughts were on optimizing AJAX content for SEO. Of course, this isn’t a new issue: just run a search for “ajax and seo,” and you’ll see a myriad of articles spanning the past two years addressing this very concern.

For those who aren’t familiar with AJAX, the acronym refers to “Asynchronous JavaScript & XML,” a coding technique in which a website retrieves information from the server without a page reload. Some of the most widely-used web applications (Gmail, Facebook, etc.) make use of AJAX, and the result is a smooth, seamless user experience. Many other websites utilize AJAX to compartmentalize large bodies of information on a single page.

AJAX may be slick, but it can present SEO challenges. Fewer page loads means fewer indexable URLs. And if a page can’t be indexed, it’s essentially invisible in search.

So how can you ensure that your website doesn’t fall into search oblivion on account of AJAX? For starters, here are three tips!

1. First ask yourself if your AJAX content needs to be indexed. In the case of password-protected web applications, the intended target audience was never search engines in the first place.
2. If your content must be indexed, consider using client-side-only JavaScript as an alternative to AJAX. Just make sure it’s implemented well, as JavaScript can create its own set of SEO problems.
3. If you must use AJAX, make your content accessible via a unique URL. Then make sure a search engine crawler can find this URL, either through direct submission, or through crawlable links on your Site Map page.

Now go build a beautiful, slick, SEO-friendly website!

A little PR can go along way in getting your name searched online.  Of course, this rule holds true for positive and negative news alike.  So what feeds our urge to search more, good press or bad press?  Let’s look at three anecdotes from recent history and see what we observe about human nature.

Alex Rodriguez: World Record vs. Steroid Scandal

A-Rod had an eventful past three years.  In August 2007, he became the youngest man in baseball history to hit 500 homeruns.  About a year later, he filed for divorce.  And then just recently, he admitted to steroid use.

Interestingly, his world record (Marker B) only generated about half as much search intrigue as his divorce (Marker E) or his steroid news (Marker F).  I find it most interesting that searchers seemed more interested in the divorce than the steroid news.  Just shows how America loves a good love scandal.

Miley Cyrus: Sold-out Concerts vs. Vanity Fair Embarrassment

Miley Cyrus hit the tween-age scene as Hannah Montana in late 2006.  As her TV show grew in popularity, so did her music, which generated nearly $30M in tour revenue.  This teenage star came crashing down in mid-2008 when questionable photos from a Vanity Fair photoshoot leaked to the public.

Miley (and Hannah) were consistently popular in searches around the time she was selling out arenas.  But look at that spike after the Vanity Fair fiasco (Marker C)!  Unfortunately for Miley, the search buzz around her name in that small time period pretty much equaled all of the searches for her name to date at that point.  She’s learning her PR lessons young!

Michael Phelps: Gold Medals vs. Guilty for Drug Use

Michael Phelps became an American icon in the 2004 Olympics when he emerged as a gold medal winner.  By the time the 2008 Olympics were over, he’d annihilated seven world records and brought home fourteen gold medals in his short career.  Then in 2009, Michael appeared in the British tabloids smoking marijuana, and it all went downhill from there, sponsorships and all.

It’s hardly reasonable to compare search volume in 2004 and 2008, but it is interesting to see the amount of intrigue Michael’s first gold medals created online.  His unhuman performance in the 2008 Olympics (Marker D) clearly gave the world Phelps fever, which is more than evident in the graph above.  And interestingly, unlike our other two anecdotes, the impact of Michael’s bad press (Marker F) didn’t quite amount to the buzz around his athleticism.  Lucky Mike.

So I guess our lessons are as follows (of course, they’re anecdotal, so take them with a grain of salt):

• People are far more interested in learning about your dirt than your achievements.
• If you win 12 gold medals and break 7 world records, not all the rules of nature will apply to you when you screw up (but you might still lose a big sponsorship!).

Are there any other celebrities whose search buzz you think I should profile?  Post a comment or shoot me a message.

Two days ago I wrote a post about the future of search, and how it will dictate the future of SEO.  In particular, I mentioned the possibility of search moving more toward a social media aggregation model, an end to which Twitter is already working.

Social media-based search would be an interesting shift in our search paradigm, a shift that would create interesting new challenges for SEO.  But just last night, Marissa Mayer, VP at Google mentioned the possibility of an even more radical paradigm shift: search engines that return answers, not just results.  Listen to Marissa Mayer’s commentary.

Now this could make SEO a very interesting job.  Just imagine: an SEO’s role would no longer simply be to drive traffic to a website, but to convince a search engine to include a website’s content as part of a greater solution returned to a user.  It’s hard to imagine exactly what Google means by delivering an “answer,” but one example Marissa gave was assembling a chart or video to answer a searcher’s question.

One implication I could envision for SEO is a shift in targeting strategy, making SEO less like direct marketing and more like advertising.  Given the nature of search today–a search engine returns results deemed explicitly relevant to a particular search query–SEO strategy generally begins with keyword targeting, or identifying specific query spaces where a website can intersect a user.  Generally, these query spaces must be highly relevant to a website’s offering.  But in “Search 3.0,” a search engine would try to “figure out” the entire picture of what a user is really looking for.  So a site could “rank,” in a sense, for terms other than highly-relevant query spaces.  For instance, “used volkswagen” might not just turn up information on where to buy a used VW, but where to find aftermarket outfitting, too.

How do you imagine the evolution of search into an answers rather than results might mold SEO?  Share your thoughts!