|
Getting Setup To Measure SEM 
By: Gary Angel 2007-07-09 The first step in any web analytics measurement project is to make sure you understand what needs to get measured and what is getting measured. This is as true with Search Engine Marketing as it was with Internal Search.
If youre new Search Engine Marketing, heres a quick run-down of the basics. Search Engine Marketing is all about generating traffic from Search Engines like Google. Search traffic comes in two basic forms: paid and organic. Paid Search Traffic comes from ads placed by a company with the Search Engine in response to particular Search Terms (also called Key Words) entered by a user. Typically, these paid ads are short, text-based listing that go at the very top of the page and to the far right. Paid Search Marketing is usually called Pay-Per-Click (PPC). Its called Pay-Per-Click because an advertiser doesnt pay to have the ad displayed " they pay when a visitor clicks through to the site from the Search Engine. The other half of Search Engine Marketing is generating organic traffic. Organic traffic comes when users click on the listings a Search Engine has displayed in response to a Search Term because the web site appears to be "relevant" to the Term. Organic listings are also called "natural" search. When a visitor comes from an organic listing, the site getting the traffic doesnt pay anything. In one sense, this traffic just happens " it isnt classic marketing at all. However, with so many web sites potentially relevant for most any Search Term, there has sprouted up a discipline (called Search Engine Optimization or SEO) whose function is to get a site listed as prominently as possible in the natural listings for commonly used and relevant Search Terms. In both PPC and SEO, the higher the position of the listing the more traffic you are likely to get. Getting high position typically costs more " either in hard dollars (PPC) or work (SEO). But not every Search Term is equal. Some terms are queried tens-of-thousands of times each day. Others once or twice a month. Newcomers to Search are often shocked at how light the query volumes are for many Search Terms. This means that having a high position for a single frequently searched Term can generate way more traffic than having high position for hundreds of less common terms. So what about measurement? When a visitor clicks on a Search listing (paid or organic), the visitor arrives at the site with a fully-qualified referring site. That referring URL will tell your web measurement solution what site sourced the traffic. It will also contain the search term used by the visitor. Most web analytic tools will be able to parse this information right out of the box. They will know that Google is a Search Engine and that Semphonic is not. So they can separate out Search-based traffic. In addition, theyll understand what query parameters each Search Engine uses (like q=) to designate the query. Since the Search Term is so important for understanding what drove the traffic, nearly every web analytics solution will automatically provide a report showing traffic by Search Engine by Search Term. So far so good. But there are some things that dont come "out-of-box." Heres the most important to understand: your web analytics solution cant distinguish between an organic and paid link when its passed as the referring URL. This, of course, it vital to any real-world measurement. Since a site has no control over the Referring URL for an organic listing, pretty much every web measurement solution has the same fix for this problem: add a code onto your Paid Campaigns. This code goes into the target URL attached to an Ad " and it can contain pretty much anything you want. Step #1, therefore, of doing Search Measurement is insuring that the target URLs in your PPC campaign (these will be set by the PPC Buyer " often an Agency) contain a unique identifier that will allow your web analytics package to recognize Paid Ads, and then making sure that the WA package is configured to recognize these. How about recognizing organic traffic? That happens naturally if you setup paid campaigns. Your web analytics package will assign any Search Engine traffic that isnt attached to a Campaign as "organic." This means that you MUST always assign Paid Traffic to a Campaign or else you will miscount BOTH your organic and paid traffic. Simply identifying Paid traffic isnt enough. Most PPC programs these days are astonishingly large. Buyers have embraced the concept of a "long-tail" with a vengeance. It's not unusual for programs to have ten thousand, a hundred-thousand or even more Search Terms. Inevitably, the vast majority of these terms drive little or no traffic. From the web analytics perspective, you may never see many of them as actual clicks. But this enormous "tail" raises a problem. With so many words, the "tail" may be a big part of a program even when most of the words drive very little volume. That "tail" is hard to measure. Some Web Analytics tools crop the tail. So you dont even see any of these keywords. Others provide access to the full tail " but no good way to do much with it. On their own, none of these words are likely to drive sufficient volume to be statistically analyzable. And if you find yourself reporting on words with less than 50 click throughs, you should BEWARE. That information is almost worthless. There is a way to group pieces of the tail analytically " but to do so the web measurement team needs to be proactive. Buyers already group these words " they never actually manage thousands of Search Terms individually. Instead, theyve grouped them into what are called Ad Groups. These Ad Groups generally have a single bid, a specific set of creatives (ad text) attached, and (in many cases) a specific landing page. Unfortunately, many programs arent setup to pass the Ad Group through the URL to the web measurement solution. They should be. And if you are starting to measure PPC programs, this should be one of your first priorities. Dont assume, by the way, that the Ad Groups will always be clean. Theres a decent chance that youll find they make little or no sense from a measurement perspective " grouping totally unrelated words together. It shouldnt be that way " but it often is. Badly structured Ad Groups are a tip-off that your program isnt well managed. I always recommend reviewing the Search Terms in each Ad Group when you implement measurement. So Step #2 is passing the Ad Group code in the Target URL and using this code to generate Search Campaigns. This makes it possible to generate Campaigns and Segments in your web analytics tool based on visitors reaching the site via a whole group of Search Terms around a single concept. For some campaigns, even passing the Ad Group isnt enough. PPC buyers typically buy at the Search Term level. The Search Term is a single word or phrase. However, the Search Term can be set to EXACT or BROAD match. With EXACT match, the search user must type that specific Search Term or the advertisement wont be served. With BROAD match, any search the user types that includes the purchased words will generally qualify for the advertisement. So if you buy "analytics" on broad match, your advertisement will be served in response to the user entering either "web analytics" or "stock-market analytics." Why is this important? Because in Reports from the Search Engines, you typically see information grouped by the Search Terms you purchased. But in the Web Analytics tools, you see Search Terms reported by what the user actually entered. If you are buying Exact Match, these will be the same. But if you are buying Broad Match, they will be quite different. This can make comparing (or matching) numbers between your Search-based reporting and your web analytics tools almost impossible. Step #3 then, for sites that buy "Broad Match," is to break out these terms into separate Groups and pass the actual Search Term purchased in the Target URL. This makes it much easier for the analyst to map "Search Term Used" to "Search Term Purchased." Another important variation on Search is Content Networks. When you think of PPC programs, chances are youre thinking of classic Search (user entered a Search Term and got back a set of listings). But most of the Search Engines also have Ad Networks " meaning they serve your Ads on content sites with publishers all around the web. In theory, they match your Ads based on the Search Terms to the content the Ad is displayed in. Once again, the advertiser pays when a user clicks through. Buying on Content Networks is a huge topic. Many of our clients choose not to do it because of Brand, Quality Control and Fraud issues. These are ALL very real concerns. From a measurement perspective, however, whats important to understand is that Content Network buying is always at the Ad Group level. There is no one Search Term to match up to. Nor was there any Keyword typed by the user. So this traffic shows up very differently in your web analytics tool. You should also know that most times the site that generated the click through is buried in the referring URL passed to your site. Thats important, because it can be extracted and reported on. And thats not information that your buyers can get via any of their basic PPC reporting tools. So Step #4, if your company is buying Content Network traffic, is to make sure that this traffic is isolated in separate Ad Groups and that youve setup your measurement solution to parse the URL and extract the true referring site (the publisher) and store it in a variable. This has been another of my LONG posts, so heres a quick summary of the basic steps to getting Search Engine Marketing programs setup in a web analytics tool. * Step #1: Insure that every target URL in your PPC Program contains a campaign identifier.In the next Series post, well get started with some basic Search Analysis. -Click here to read Part I- Comments Tag: SEM Have a bookmark! - 
About the Author: Gary Angel is the author of the "SEMAngel blog - Web Analytics and Search Engine Marketing practices and perspectives from a 10-year experienced guru. | |||||||||||||||||||||||||||||||
| SearchNewz
is an iEntry, Inc. ® publication
©
1998-2008
All Rights Reserved. Privacy Policy - Legal - Sitemap - Contact Us - RSS Feeds - Newsletter Signup |
