The Challenge of SEO Impression Data
While RKG’s expansion into offering full-fledged SEO services brought on board a team of experts in the field, it has been a learning experience to those of us whose background was on the paid side of search marketing. In the last few weeks, one of the biggest stories to touch on both PPC and SEO was Google’s move to begin encrypting searches by default for signed-in users. As a result of the change, a meaningful percentage of search queries were no longer passed in the referring URLs of organic clicks. There was a palpable level of indignation that Google would continue to provide query information for all paid clicks, as it was seen as hypocritical in some circles and a sign of Google being disingenuous about its concerns for user privacy.
To me, the tumult around the switch to SSL obscured the larger issue that data for organic search traffic has never been as robust as the data we have for paid search. No doubt this change won’t make it any easier to apply data-driven methods to SEO, but it has always been more of an uphill climb.
Clicks, but not Impressions
Analytics tools like Omniture, Google Analytics or RKG’s own Attribution Platform can do a good job of tracking incoming traffic and its subsequent performance across a variety of metrics, but what if we want to know more about what is happening before users get to our site? Having impression data for our organic listings allows us to determine our click-through rate, and that can help tell us whether an increase in traffic is the result of our listing appearing more frequently, being clicked more frequently, or both.
One idea that we’ve written about a few times in the last couple years is that Google’s paid listings are looking more and more like the organic listings and that ads are taking up more real estate on the SERP. A natural question to ask is: what sort of impact are these changes having on the breakdown of our organic and paid traffic? If organic traffic has declined relative to paid, is it really fair to blame our SEO efforts or give credit to our PPC program? It’s informative to see how click share may have shifted from one channel to another, but we’d want to eliminate as many variables as possible to see if Google’s changes have boosted PPC’s click-through rates over time at the expense of organic CTR. To do that, we are going to need historical impressions for both.
Organic Impression Sources
Since April of 2010, Google’s Webmaster Tools has provided impression counts for search queries where your organic listing appeared, but the data is imprecise, incomplete and not easily accessible. On top of all that, it only goes back for 30 days. Webmaster Tools impression counts are rounded, meaning that they are not exact and they are only provided for top queries and/or queries that generated a click. Where impressions or clicks are below 10, a specific figure is not provided:
Webmaster Tools impression counts are really only useful for getting a snapshot view of top traffic queries. Pulling query data is not supported via API, so we must extract it from the web UI, which has limited functionality. One of those limits is the 30 day time window, which makes viewing historical CTR trends impossible unless you have been manually pulling reports and databasing them yourself.
Another potential source for organic impression data is Google’s Adwords Keyword Tool, but it comes with its own set of limitations. First and foremost, the Local Monthly Searches provided by the tool aren’t meant to reflect the impressions our listings received, but rather the total number of searches for a particular query. Even if we have a listing in an apparently stable high position, we can’t be sure that it appeared for all searches due to SERP personalization and other factors. Although historical data is available in the Keyword Tool for the past year, it is only available at the monthly level.
If we’re running paid search ads, we’ll also have impression counts for the keywords we have in AdWords. Again though, even if we have an impression share of 100% for the paid keyword, it is no guarantee that Google showed the organic listing consistently over the same period or even at all.
An Example: PPC v SEO Traffic Share
Getting back to the question posed above, I did attempt to gauge the extent to which Google’s SERP and ad format changes may be shifting clicks to the paid listings. Since robust historical impression data is not easily come by for organic traffic, I opted to compare PPC and SEO click traffic for brand terms where we expect to be showing in position 1 nearly all of the time. This effectively takes impression and position variability out of the picture and the traffic share by channel should be a good reflection of click-through rate changes.
Here’s an example of how the trending has worked out for one client’s top brand term over the past 17 months:
In this relatively short period, we have seen the paid ad go from generating clicks at about half the rate of the organic listing to a rate 40% higher. Examples from other sites showed smaller changes, but all showed a shift to PPC over time.
Year-to-date, PPC’s share of traffic for top brand terms across a sample of clients has increased a median 20%. Whether we can extrapolate these results to the larger picture is an open question and one we intend to pursue. But, the paucity of organic impression data, along with the extra headache of Google’s encryption policy change will make it difficult.
One’s view of whether Google should feel a responsibility to provide more robust data on organic traffic or whether its even in their interest to do so likely depends on how you feel about the role of SEO. On one end of the spectrum, SEO is seen as gaming the system to elevate commercial content, with Google left to play whack-a-mole using its algorithms to keep its results from being overrun by spam. A more charitable view of SEO is that it is largely about making content readable and understandable to Google’s smart, but imperfect crawlers, allowing Google to make better decisions for its SERP using a more complete view of the web.
If the truth lies closer to the latter, then SEO can be considered beneficial to both site owners and the search engines. Google, often via engineer Matt Cutts, frequently offers SEOs and site owners advice on optimizing site content and its clear they see a benefit to themselves by doing so. The bad actors out there are free to use this advice and the data we have now, just as they would be free to use a more complete and accurate data set, if Google were willing to provide it. If it’s a matter of cost, dare I suggest Google offer more comprehensive organic traffic data for a small fee? Or maybe Google just doesn’t want us to get too clear of a view into some of the larger shifts they have set in motion.