A Key Paid Search Finding *Every* SEO Should Study
The very natures of Paid Search and Organic Search may seem different, but they are actually substantially similar. Most folks specialize in one or the other, but comparatively few excel in both. This has resulted in a severe lack of cross-pollination between the two fields, other than in the area of keyword research.
Where overlap exists, we can leverage learnings in one field to understand the other, provided we’re mindful of their differences. One of the landmark papers on Paid Search was a Google paper called Predicting Bounce Rates in Sponsored Search Advertisements; here we’ll ruminate on the similarities and differences between the two fields and see how we can apply the key finding of that paper to the field of Organic Search.
Differences Between Organic Search and Sponsored Search
We’ll skip the obvious, being that you’re paying for a click versus getting traffic “for free” (i.e. via “hard work”):
1. Location of results on the Search Engine Results Page (SERP) differs.
In the case of Sponsored Search, some of the results appear above the organic results (almost appearing to be organic, in a way, except differently shaded), and some appear on the right part of the screen. Eyetracking studies have shown that people tend to scan the upper left corner of the page, the left side of the page, and across the top. Google [eyetracking study SERP] for many examples.
2. The landing pages typically have different goals.
For sponsored search, you really don’t have to worry so much about ranking, keyword density, and things along those lines (other than your bidding, and making sure your creative is relevant to the keyword). In Sponsored Search, the overarching goal is to convert prospects. A conversion is typically either having someone add something to a cart, or give up contact information (i.e. a lead).
It’s widely known that, for landing pages, in order to facilitate conversion, the best practices are to make it *very obvious* what the visitor should do, and to *remove all obstacles* to them doing so. This explains why the best landing pages typically have: very little text, a large orange button with no other choices presented, and, at most, one picture (usually of a happy person, or of the product – to evoke happy feelings and to encourage the person to click).
In Organic Search however, the immediate goal cannot necessarily be conversion. I know this sounds heretical, but if you make a page with very little text on it, optimized to get people to click a button…guess what, no one will ever see it, so it won’t matter that it converts well!
Factors like keyword density and copy length, regardless of what you’ve been told, *do* matter in organic search, in order to be found at all. Look at the Organic Search results for [debt consolidation] – you won’t see any sparse pages there – on the contrary, they’re all quite wordy and keyword-rich. So landing page design, particularly length and keyword density, is probably the most major area where these two fields differ.
3. Sizes of the elements differ.
Search on an easy keyword like [debt consolidation]and then look at the Title, Description, and Meta-Keywords in the organic results, and compare them to the Title, Creative, and Desitnation URL in the sponsored results on the right. The sponsored results are about 1/3 the width of the result for the organic results. In fact, (no coincidence), a 2-3 word keywords can often just about take up an entire line of a creative with little room left over.
4. Ranking algorithms differ.
Obviously, in Sponsored Search, there are *very few* variables involved in ranking, most importantly the results of the auction (i.e. how much you bid), and the relevance of the keyword to the creative. Organic Search utilizes many variables including on-page variables such as keyword density as well as off-page variables such as PageRank (i.e. links) and increasingly, social signals (shares, likes, and so on).
Commonalities Between Organic Search and Sponsored Search
Despite all of those differences, the bottom line is, both types of search are very similar:
1.) Both are in response to a keyword query by the user.
2.) Both are presented in columns, with the user scanning down the columns to select a result.
3.) Both present titles, some sort of description, and a URL.
4.) Both display highlight keywords from the query in bold when a result contains them.
What is the difference between a search user evaluating one type of result versus the other? Very little as it turns out. In fact, how many people have you met who don’t even understand the difference? I often run into people who do not understand that the first few results in pink at the top of the page on Google are Ads (even though they are clearly marked with the tiniest of fonts ;-). If searchers don’t know the difference and *think* of them all as the same, then aren’t they in some sense, the same (even just in terms of how searchers will evaluate them?)
So, in terms of people looking the page and evaluating results, we can probably draw some valuable learnings from people’s behavior, whether they are from studies on Sponsored Search or Organic Search.
Quality Score – A Quick Explanation for SEOs
Before we continue, we need to run over this quickly. If you’re into Search Engine Optimization, you may not be very aware of how Quality Score works. The AdWords auction is in a class of “second-price” auctions where your best move is to actually disclose, via your bid, how much a click is really worth to you. This is because Google will not actually make you pay that amount; they only make you pay that amount based on what the next bidder down bid – so you will pay what it is worth to him.
However, Google still makes a “Quality Score” adjustment to what you pay – essentially rewarding relevant ads and penalizing less relevant ones. For instance, if you sell lawnmowers, set up some ads for them, and bid on the term “Britney Spears”, the fact that your advertisement has little to do with her will cause you to pay more. Someone selling Britney Spears calendars will pay, relatively, less.
Quality Score is a whole discussion in itself (there are actually several different ones – details here). Interestingly, Quality Score represents an interesting lever by which Google may actually be able to manage its earnings by allowing or disallowing less relevant ads across the board (purely speculation on my part).
Google’s Landmark Paper on Quality Score
In 2009, some Googlers published a landmark paper “Predicting Bounce Rates in Sponsored Search Advertisements” that appears to disclose how Google calculates Quality Score. The findings of this paper apply just as well to Organic Search as they do to Sponsored Search.
A couple of key points are:
1.) Quality Score is largely just Click-Through-Rate.
This is implied throughout the paper, but never explicitly stated. An interesting supporting fact is that the team used human raters to calibrate the experiment, categorizing advertisements into “excellent/good/ok/bad” which hearkens back to when Quality Score was not a 1-10 value as in the last few years, but instead a three value variable (“Excellent/OK/Poor”).
This point is also confirmed by the pie chart shown in the video by Google’s Chief Economist, Hal Varian, explaining Quality Score as well as numerous official AdWords articles and FAQs.
2.) When Click-through-Rate is unavailable, Google must use a proxy.
When you have deployed a new keyword-creative combination in your AdWords campaigns, until there are enough impressions and clicks to reliably determine the average Click-through-Rate, there is a need to estimate what that rate is using some other more-available measurement.
3.) Bounce Rate is highly predictive of Click-through-Rate.
This raises the question, how can someone predict a bounce rate if there’s not enough click-through-rate data available? Surely a click must occur before a bounce can even happen!
4.) The variable that most correlates to Bounce Rate is “relatedness” of the keyword to the creative.
Surprisingly, the relevance of the landing page itself barely seems to matter at all. The interesting thing about this is, to evaluate those, one not need even leave the Search Engine Result Page and travel to the landing page! Hard to believe but true – read the paper if you don’t believe me.
5.) Google measures “relatedness” of the keyword to the creative using “a proprietary process similar to Latent Semantic Analysis”.
If Organic Search and Sponsored Search are similar enough, then perhaps Google uses Click-through-Rate to determine “quality” of Organic results; and if so, then Google would need to establish a baseline, and until it could establish one, it might utilize some proxy measurements to determine where a page should rank. I wrote a bit on this previously in “Does Google Use Click-through-Rate as an Organic Ranking Factor? Answer: Maybe“
One interesting takeaway from the paper is, the landing page matters little in Sponsored Search (other than, of course, Google is probably checking to make sure it’s not a 404 page, probably checking the speed it loads and so forth).
This is great news for Organic Search; one of the most major areas where Organic Search and Sponsored Search differ, the landing page itself, doesn’t really matter for prediction of bounce rates. This implies that the major finding of the study probably *does* apply to organic search as well.
What was the major finding?
The best predictor of bounce rate (and therefore click-through-rate) was how “related” a result was to the keyword.
How can result elements be as “related” as possible to the keyword?
This is pretty obvious. If you’re searching on “cayenne pepper”, the word “hot” is pretty related. But the term “cayenne pepper” is, by definition the most related phrase of all that you could possibly use (i.e. it’s 100% related to itself!) So this argues for including the keyword in the title, in the meta-description, and in the URL (the Organic Search equivalent of Sponsored Search’s title, creative, and destination URL.)
Where should they be located?
Well, if you look at the eye-tracking studies, it’s pretty clear that the more to the left, the more likely someone is going to see it when they’re scanning down the page. It could also be that having white space to the left of it is what’s making it stand out; either argument supports placing it at the beginning of each element. So your keyword should be all the way to the left in the Title, in the Meta-Description, and in the URL. Look at a SERP yourself – you tend to scan down the results, looking at the first few words of each line.
How can you best have a keyword as far as possible to the left in a URL?
Obviously, by having the keyword either in the domain or in the subdomain. This explains the high values in the domain marketplace, and some SEO practitioners’ fetish for using keywords in subdomains. Aaron Wall actually has an interesting posting about subdomains and Panda worth a look – I don’t recommend you spend a ton of time creating thousands of subdomains, but if you do and it works – don’t tell anyone about your success – or everyone will then do it and Google will demote its worth 😉
I recently did a test of 6 different meta-description formulas on a website with multiple thousands of pages. These six variations differed in many respects but the primary difference was where the keyword was located in the description (at the beginning, after one word, after two words, and so on).
Then I took all the resulting Click-through-Rates and positions after a few months and used the information from my earlier posting “Estimating Organic Search Opportunity” to estimate what Click-through-Rates one would expect at each position, and compared that to what was actually experienced.
As with anything, wide variations occurred, but I then examined the top 20 highest CTR meta-descriptions, and the bottom 20 lowest CTR meta-descriptions, and guess what? The variation that appeared *most often* in the top 20 happened to be the one that had the keyword all the way to the left in the Meta-Description. The variation that appeared *least often* in the bottom 20 was – guess what – the same one.
Although not rigorous, this was ample enough proof to convince me that the age-old best practice of putting the keyword all the way to the left in a Meta-Description is indeed the correct one. Logically, this ought to be the case for titles and URLs as well.
It ultimately doesn’t matter which variable Google is using to help determine organic ranking (Bounce Rate or Click-through-Rate), because they are correlated to each other, and are both driven by relevance of the keyword to what is presented.
So, in short – if you want to have a low bounce rate in Organic Search and a high Click-through-Rate (and in both cases, happy/satisfied users and therefore, higher rankings and traffic), then rather than having some random title, meta-description, and URL, you should ensure that the keyword the page is about is located in all of them, and as far to the left as possible.
1.) Searchers will know what they are getting into…
…and are more likely to pick your page because it’s what they want.
2.) They are also less likely to be unhappy when they get there and bounce back.
3.) This takes advantage of the tendency of searchers, when they search on “blue widget”, to scan the results for the same term “blue widget” and will tend to favor those that include it.
4.) The keyword (or its elements even if they are separated) will appear in the SERP result in bold…
…garnering more attention by searchers and increasing the likelihood of a click even more, simply by standing out.
So by making sure you have the keyword in the title, meta-description, and URL, and in all cases as far to the left as possible, it’s win-win-win-win for you – and your potential visitors!