According to popular reports, 25-30% of the clicks search engines generate are produced by paid results. That's a number that seems reasonable on the surface given that paid search is a multi-billion dollar a year business.
But a study titled 'Investigating customer click through behaviour with integrated sponsored and nonsponsored results' in the International Journal of Internet Marketing and Advertising throws this figure into question, and also produces some food for thought when it comes to other common beliefs about paid search ads.
According to research conducted by Bernard J. Jansen, an assistant professor at Pennsylvania State University, and Amanda Spink, a professor at the Queensland University of Technology, search ads account for 'only' ~15% of all clicks on search engines.
This is based on an analysis of millions of search queries that were recorded by DogPile, a popular meta search engine.
Excluding queries that produced no clicks (which represents over 35% of total queries), sponsored results accounted for approximately 15% of clicks while organic results accounted for 84.2% of clicks.
What's most interesting about this is that, unlike Google and Yahoo, for instance, DogPile mixes organic and paid results (although paid links are identified as such). Jansen and Spink point out:
...it appears, perhaps counter intuitively, that integrating sponsored and nonsponsored links in the same listing does not raise overall sponsored link click through. However, it could also be that the popular press reports of click through rates are inflated. This seems highly likely.
Jansen and Spink went even further; they used an automated process to classify search queries into 3 categories: informational, navigational and transactional. They then analyzed the clickthrough breakdown for each. Navigational queries, which are defined as queries whose purpose is to "locate a particular website", produced the highest percentage of clicks on paid results (19%) while transactional queries, which are defined as queries related to products and services, produced the lowest percentage (14%).
While they acknowledge that DogPile's users may not be representative of Google or Yahoo's users, this is interesting research and provides quantitative data in an area that really hasn't had much.
They conclude by suggesting that the segregation of organic results and paid results may not be necessary, or beneficial:
There may be substantial benefits in an integrated SERP. First, given the reported negative bias users appear to have for sponsored links (Jansen and Resnick, 2006) and that research has shown sponsored links are relevant for user queries (Jansen, 2007), the prominent highlighting of presenting separate listings may be denying users relevant results.
Studies have shown that web searchers do not understand how web search engines rank results. Web search engines do not generally disclose how nonsponsored links are listed and ranked. Why do so with sponsored links? Certainly, exploring this line of research is an area worth pursuing. To improve, we search engines must leverage an increased knowledge of user behaviour, especially efforts for understanding the underlying intent of searchers and how this intent relates to sponsored links.
It's an interesting and counter-intuitive suggestion. One that, incidentally, just might be worth exploring if follow-on research validates Jansen and Spink's work.
In the meantime, I think this study demonstrates why organic SEO and PPC aren't an either-or proposition when it comes to success. While I know lots of companies are doing very well with PPC, it's clear that marketers who aren't investing in obtaining good organic SERPs are potentially leaving lots of clicks - and lots of money - on the table.