Part 4 in a Series on Web Analytics and Search Engine Marketing Programs
Traffic and conversion numbers at the channel (Search/Organic/Paid) and engine (Google, Yahoo, MSN, etc.) level are the backdrop for almost all Search Analysis. But there are several deeper layers of Search data in web analytics and these deeper layers are where most of the optimization analysis actually takes place.
If you’ve setup your Search program to pass Ad Group as a campaign code (strongly recommended – see Part I of this series), then you’ll be able to break down PPC campaigns into much more detailed "concept" groupings. The words in an Ad Group should all revolve around a single concept – and they share both Landing Pages and Creatives. Because Search Term is often too granular for analysis, Ad Group is probably the single most common level of analysis in good Search Analytics.
Set up in this way, you’ll automatically get your basic tool reports about campaigns. Typically, these include at least "responses," "repeat responses," "conversion by type," and, quite often, information about the average "time to convert." For eCommerce sites, this will also typically include DETAILED information about actual purchases sourced by this campaign.
There is already some pretty useful information here. Repeat responses helps you understand how frequently visitors are coming in on PPC campaigns. Time to convert provides good insight into the length of the sales-cycle on your site. And, of course, having detailed eCommerce data is invaluable for really good optimization.
With many tools, you can also create roll-ups to different campaign levels. With SiteCatalyst, for example, you use SAINT to attach additional campaign attributes to a specific campaign. With these roll-up levels, you’ll get the same kind of statistics (and analytic opportunities) at every level. That means you can evaluate data like repeat responses for an Ad Group, a Campaign, an Engine and for Search in general. That’s good information. Information that’s impossible to get from your Search Engine reports.
For organic search programs, there is nothing quite like Ad Group. That makes it extraordinarily difficult to impose much order at the Search Term level. In some cases, though, you can use Entry Page as a form of grouping SEO. Since SEO search terms are strongly related to pages and pages to concepts, you can often analyze SEO terms by Entry Page or Entry Area.
Entry Page can also be very useful for analyzing Paid Search. You’d think you already have this information if you’re coding by Ad Group. But web analytic tools are often limited in the cuts they can give you by any one piece of information. So there are many cases where using the Entry Page will give you additional reports (such as path, next page and link analysis) beyond what you’ll get from campaigns.
When a Landing Page is specific to a PPC campaign, you’re set to go. You can use it without additional segmentation and be confident that you are only seeing Search visitor behavior.
For SEO and for cases where PPC visitors are landing on multi-use pages, an additional level of segmentation is required. The easiest segmentation is generally at the visit level; with the criteria being that the visitor was sourced by Organic Search or PPC (or a particular PPC campaign if your campaigns share Landing Pages).
Applying this segment and then looking at the Entry Page reports will tell you exactly what happened when search visitors landed on the page. Note that it is much more efficient to apply a single segment (PPC source) and then look at Entry Page behavior than it is to build a segment for each Entry Page and then look at PPC sourcing!
If you use Broad Match and you’ve setup your PPC campaigns to pass the "purchased term," this becomes another level of analysis. Sometimes, this may just be an Ad Group. However, some sites prefer to pass this as additional attribute on the URL. When done in this fashion, it will have to captured as either a campaign attribute or custom variable (depending on the tool).
At the lowest level, most web analytic tools provide you with at least some reporting about the actual keywords entered. This data will typically be associated with each of the higher levels already described. So you can look at all search traffic by keyword, engine traffic by keyword, Ad Group (campaign) traffic by keyword and Broad Match Term by ACTUAL keyword.
In some cases, data about search terms is broken out into two categories: clicks and referring clicks. What’s the distinction? People tend to think of Search Engine visitors as always "arriving" at the site. Most times that’s true. But it isn’t actually definitional. Suppose a visitor searches on Google, clicks on your Ad and Lands on your site. Then hits the back button and goes to Google. Now if that user clicks on your link again, searches again and clicks on another of your links, or clicks on your organic link instead of your paid link, then that visitor will land on your site again. But here’s the really interesting part. That second land DOES NOT start a session. It’s counted as a click, but not as a REFERRING click.
From a web analytics perspective, a session is the same machine landing on a web site within 30 minutes of a previous page request. What the visitor did in between those two requests is both unknown and immaterial. Situations like this are why you can have multiple EXIT LINKs and multiple Search Clicks for a single SESSION. Confused yet?
Incidentally, if you’re a publisher and you carry Google (or other Content Network Ads) on your site, one of the more interesting web analytic projects is to figure out how many of the visitors who click on those Ads come right back to your site. It can be quite a large number – good news for publishers but usually bad news for advertisers.
The upshot of all this is that Clicks (as opposed to Referring Clicks) don’t tie to other variables like Entries or Visits. But it’s often very interesting to see Clicks vs. Referring Clicks. We commonly observe 15-20% differences between these two numbers – a phenomenon we call PPC Self-Cannibalization. It’s important to keep this phenomenon in mind when you evaluate your PPC reach and cost per visitor.
Search Term level analysis is usually confined to the relatively small set of terms that generate lots of volume. For sites with long-tails in SEO or PPC, the majority of these terms simply won’t have enough volume to analyze. This is also an area where web analytic tools sometimes "clip" your data. That means you won’t even see all the words in the tail. If you think your data is getting severely clipped, it’s worth talking to your vendor and complaining. In many cases, these limits are adjustable on request.
So if you are set up correctly, you should be able to go from Total Search Traffic, to Organic and Paid Traffic, to traffic by Engine, to traffic by Ad Group and Broad Match Term and, finally, to traffic by Keyword. And you should be able to see Keyword, Ad Group and Engine cuts at every higher level of granularity. Finally, you should be able to use Entry Page to track some groupings of both organic and paid search terms and to get reports like next pages, path and link analysis for just the population of Search entries.
These reports form the bulk of what you’ll be looking at in most Search Analytic projects. And for all of these reports, you should be able to get traffic numbers out of the box and eCommerce and Conversion numbers if you’re site supports those.
For many sites, eCommerce and Conversion numbers aren’t really applicable. So in the next post, I’ll start to discuss optimization metrics, measures of engagement and conversion proxies. These are behavioral measures that can help both PPC and SEO optimizers better understand the quality of the traffic they drive and how it can be optimized.
Other Posts in this Series: Introduction, Searchnomics Issues, SEM Data vs. Web Analytics Data, and High-Level Search Engine Reporting.
[Don’t forget to register early for X Change – Like "Top Gun" for Web Analysts! The Early Registration Discount ends July 31st! Use Coupon Code XCForum at http://www.semphonic.com/conf to reserve a spot!]
Comments