Maximizing Your Search Budget

The relationship between organic and pay-per-click (PPC) programs in building SEM programs has been given short shrift in the industry. Most SEM practitioners subscribe to the notion that they don't need to account for organic listing position when making PPC buys-- either because there is no relationship between the two, or because the relationship is one of simple, positive reinforcement. This belief -- which is, on its face, rather unlikely -- is eagerly supported by many in the industry with a strong self-interested motive in perpetuating an attitude conducive to maximum spending.

In a recent study with a Fortune 1000 client, which, for confidentiality purposes we're branding "Company X," we put this cozy assumption to the test-- and the results were startling. The results showed that this company's organic traffic was being significantly cannibalized by its paid program. In fact, we found cases where half or even more of the paid traffic would likely have arrived from organic listings if the paid ads were not present. That means that the true cost of many of the keywords being purchased was far higher than the buyer realized.

This suggests that contrary to popular belief, understanding the complex interaction between paid and organic search engine results should be high on the list of needs for every serious SEM Marketer. Questions like these are vitally important:

• If I'm No. 1 with organic placement for a term, what effect will a PPC buy have? Will it drive lots of incremental volume, or will it just cannibalize free clicks?
• How does the situation change if I'm No. 3 or No. 5 or even, God forbid, below the first page fold?
• Is the relationship between paid and organic support the same for every engine?
• How does a strong brand impact results, especially if the search is brand-specific?

Based on our experience, the right answers to these questions can make a huge difference to companies with a significant brand and good or excellent organic presence. These issues are mostly felt, of course, by large companies with a strong brand, good internet presence, and the combination of aggressive PPC campaigns alongside a strong natural search presence. If you fall in that category, then knowing how to allocate your buy and how much money to spend may depend critically on your view of how PPC and Organic programs relate.

We had long been frustrated by the lack of interest in these issues-- and the seemingly unlikely assumptions behind them. When one of our clients expressed a similar frustration and a willingness to investigate, we decided to conduct a formal test. Our test client, Fortune 1000 Company X, was in an excellent position to conduct such a study: it had high market-share, a strong brand, a long-standing internet presence and an aggressive ongoing PPC campaign. It had a multiplicity of listings that shared PPC and natural exposure.

In the summer of 2005, the Fortune 1000 Company X PPC program was generating tens of thousands of visits a month. At the same time, Fortune 1000 Company X was well represented in natural search-- particularly for brand names.

To perform the study, Fortune 1000 Company X went dark for a 10-day period following a sustained buying effort, then measured the paid and organic traffic for 10 days prior to the "dark" period and compared this -- engine by engine and keyword by keyword -- to the organic traffic during the "dark" period.

There were no significant changes to the website during these two contiguous periods and the "dark" period proceeding. In addition, there were no changes in exogenous factors like TV campaigns, product announcements or major company news. In addition, overall traffic to the site on other sources was tracked to make sure it stayed constant.

Here is a highly level summary of traffic:

In the 10-day "Light" Period, Fortune 1000 Company X drew 28,000 visits-- of which just over 50 percent were from natural listings. This company tagged all PPC traffic with specific campaign values-- making it easy to identify the volume from each source. In the Dark period -- with other traffic very slightly down -- organic visits were up to nearly 20,000. Taken at the highest level, this means that their PPC program was providing significant incremental traffic (a 41 percent lift over organic traffic alone). However, it was also significantly cannibalizing organic traffic. Depending on engine, anywhere from 20 percent to as much as 50 percent (half!) of all PPC clicks would have been delivered organically.

Think of it this way-- if Fortune 1000 Company X is paying MSN $1 per click, it is actually paying $2 per incremental click because it would have gotten half of its PPC clicks for free through organic. This is a dramatic difference when it comes to evaluating PPC effectiveness and internal PPC optimization. This does not mean that buying cannibalized words is always a bad thing. If the initial cost was low enough, then the incremental cost may still be lower than comparative buys on other terms. The main point is that there is no way to accurately judge the true incremental cost of a word unless you've measured organic cannibalization!

Not only were there profound (and consistent) differences by engine, but there were significant differences by keyword. And the single most important variable in understanding the degree of cannibalization turned out to be -- no surprise -- organic position. The higher the organic position, the more likely cannibalization was to occur. It was difficult to study the impact of paid position, because Fortune 1000 Company X was almost always in the top two slots.

Here are some sample results showing cannibalization by organic position:

Because cannibalization varies by engine (with organic position held constant) and by position, the study suggests that a one-engine fits all strategy simply won't work. Not only is organic strength quite different by engine-- but the layout and habits of engines clearly influenced the degree to which PPC and organic listings interacted. In many cases, this appeared to have a substantial impact on organic cannibalization.

Bottom line, if your SEM provider isn't accounting for organic cannibalization and you have a significant natural presence, then chances are they are seriously mis-optimizing your campaign. Indeed, it seems likely that the worst possible scenario would be to use an automated PPC management tool that isn't taking account of organic cannibalization. Such a system would be particularly efficient at steering dollars into terms whose apparent (but not real) incremental cost is artificially lowered by organic cannibalization.

However, there is no reason to slight human buyers-- if you're judging and rewarding your SEM program and buyer by cost per click or conversion, without taking into account organic cannibalization, then there is no reason to expect that a human buyer won't do nearly as well mis-managing your campaign.

There are always plenty of people willing to give you the wholly self-interested "industry" point of view. But the best direct marketers know that there is no substitute for the relentless pursuit of measurement and incremental improvement. So while companies -- with different brands, different costs to buying brand words, different organic strengths and different product names (less prone to misspellings for instance) -- will ultimately perform differently, the Fortune 1000 Company X experience suggests that any company with a strong brand and significant organic placement ignores organic cannibalization at its own peril. How is your program impacted? Have you taken the time and trouble to find out? You'll never know unless -- like Fortune 1000 Company X -- you are willing to test, measure and live by the results.

About Gary Angel, President and Chief Technology Officer

Bringing over 20 years of experience in decision support, CRM, and software development, Angel co-founded SEMphonic and is president and chief technology officer. He's responsible for leading the development of web analytics and SEM decision making tools to market and support customers.

Before SEMphonic, Angel created and implemented multi-million dollar CRM systems whose customer-driven use of Web analytics increased profitability and customer loyalty for Fortune 500 companies including VISA, Bank of America, and American Express. His pioneering work in web behavioral profiling, neural-network web analysis, web-content success correlation, and point-to-point process analysis also has enabled major brands such as American Express, Charles Schwab, Intuit, CyberTrader, and AOL dramatically improve customer acquisition and satisfaction, shopping-cart conversion, and result in consecutive multi-million dollar bottom-line improvements.

Angel has published articles including: "Using Card Transaction Data, American Demographics," "Using Daytime Data for Direct Mail Targeting, Business Geographics" and "Worker's Compensation Providers Networks, Business Insurance."

He graduated, with honors, from Duke University and lives in San Francisco with his wife and two young girls.