Forecasting Paid Search Traffic

Despite the hype, being #1 in paid search results may not provide your company the desired bang for your marketing buck. Intelligently optimizing a paid search budget involves far more than setting random bid prices and guessing at returns.

A special report from the Search Engine Strategies conference, December 13-16, 2004, Chicago, IL.

The panel “Forecasting Paid Search Traffic” featured four experts demonstrating how marketers can make sense of the morass of paid search marketing metrics—and provided solid methods for forecasting paid search advertising spend, traffic and conversions.

Why forecast? According to Brian Cavoli from Media Contacts, forecasting allows you to control your marketing spend by predicting the business impact of your keyword choices and desired positions. That is, for every $X investment, a company can predict $Y in sales.

“Businesses need a higher level of strategy,” he said. “Forecasting helps businesses make smart business decisions.”

Positioning for maximum impact

It’s no surprise that rank matters for a PPC campaign. Whether rank translates into conversions, however, is another issue.

“A number one rank is not the best strategy for everyone. Companies need to balance volume and CPA [cost per acquision” based on goals,” said Young-Bean Song from Atlas DMT.

A 2004 study by the Atlas Institute indicated that traffic drops significantly by position. There was a 40 percent decrease in click potential in Google between a #1 and #2 position (“click potential” is defined as the product of relative impressions and relative click-through rate). In Overture, there was a 20 percent drop in click potential between a #1 and #2 spot. Overall, there was a 10 times difference in traffic between the first and the tenth rank.

This information helps companies decide whether it’s cost effective to stay in a lower position, or if bidding for a higher spot would net greater click potential. Although a company may feel like they are saving money by bidding on a #4 position, the approximate 60 percent decrease in click potential in both Google and Overture could be at the expense of possible customer acquisitions.

“The top spot will drive a ton more traffic,” says Song. “The question is, does that spot convert users?”

In many cases, it makes sense to pay the additional few cents to increase rank. However, this doesn’t mean that all businesses should automatically bid for the top position. In fact, some lower ad positions may actually gain higher conversions.

Another 2004 study by the Atlas Institute found that conversion rates for low volume keywords (which are typically highly specific search terms) may actually increase at lower ranks. That is, for some search terms, the #8 position may actually gain a better conversion rate than even the topmost position. This means that marketers have the ability to lower costs on some terms without sacrificing conversions.

“As users click on multiple listings, they are much more qualified users,” says Song.

Forecasting PPC key revenue with keywords

Fortunately, marketers can use keyword data to help them forecast both traffic levels and costs.

“Smart forecasting is about selecting the right terms,” says Cavoli.

James Colborn from Inceptor recommended starting with a sample of 100-300 keywords for accurate predictions.

“Ensure that there are branded, generic and specific terms in your keyword sample list,” says Colborn. “Once you research your search terms, you’ll get the idea of the traffic that you’d receive.”

Marketers should begin their research by reviewing maximum CPC in the engines to determine click volume and cost per click. Colborn suggests using Overture’s bid prices, Google’s AdWords estimator and results from PPC engines to determine keyword pricing. Once marketers gather keyword cost and traffic information, they can forecast data such as:

  • Projected number of searches (impressions)
  • Estimated CTR
  • Estimated clicks per month
  • Average CPC
  • Estimated budget required for the campaign

“You can determine traffic levels and costs from the keywords you research,” says Colborn.

Marketers can use this forecast to increase their spend efficiencies. For instance, keywords with lower CPC rates typically have less competition than more expensive terms. For additional forecasting, Cavoli suggested creating a matrix with keywords and a variety of conversion rates to show how small changes can lead to big results.

However, both Cavoli and Colborn warned that search projections are based on historical data and performance. Other factors, such as seasonality, sales performance, competition and landing page design may alter actual results. In this case, Cavoli recommended starting with a strategic approach, and refining it each week based on incoming data.

“Position as a plan, not as a process,” says Cavoli.

Testing, testing—understanding the true secret to forecasting

It’s true that forecasting provides solid data helping marketers predict what seemed to be previously unpredictable—how keywords, positions and impressions influence conversions. However, additional testing can help marketers measure the effects of ad positions and landing pages on bottom-line revenue, while still accounting for seasonality. Additionally, marketers can identify optimization steps before a chunk of the budget is spent.

Tony Wright from Zunch Communications recommended that marketers begin the testing process by knowing what goals they want to achieve from their PPC campaign. Is it CPA? Better volume? Increased sales? Understanding the desired result helps marketers determine how to reach their goal.

“By starting at the end with a specific goal in mind, you can get to the goal with a statistically valid method,” said Wright.

Marketers should start the testing process by understanding exactly what variables they are testing. Is it for ad copy effectiveness? Could the landing page need tweaking in order to gain higher conversions? Does ad placement really make a difference? Wright recommends portioning the test budget to account for various variables. When results come in, they need to be tracked and measured.

“You don’t know until you know,” said Wright. “Every time you make a change, you need to track it religiously.”

How much to spend on paid search testing? Wright suggested some test budget guidelines:

  • For a CPA-based budget—2 percent of desired sales
  • For a fixed budget—2 percent of the total budget
  • For a self-funding budget—2 percent of the original investment

It’s important to remember (and clearly explain to all involved parties) that testing PPC effectiveness is just that—a test. And only a test. The testing process will gain useful data that will make your company money in the long run, but, in the short term, won’t net tons of traffic and conversions.

“Don’t expect your test to garner great results. Remember, it’s a test.”

Is forecasting paid search for every company?

Some businesses, especially small-to-medium sized businesses, may feel that they don’t have time to efficiently forecast their paid search traffic. Rather than spending the time creating a list of candidate search terms and testing results, they pull bid prices out of the air and are pleased when they see conversions. This approach may be costing more in revenue than any time savings gained in a “set it and forget it” PPC campaign.

Additionally, forecasting and testing takes time. Developing a strategy does increase the time-to-market on a PPC campaign, and can be expensive to implement. And testing low-volume items, according to Wright, isn’t cost effective—it takes too long to get an accurate forecast sample.

At the same time, stepping back and creating a strategy before marketing implement a PPC campaign can eliminate many inefficiencies—and give marketers a clearer picture of expected total returns.

Heather Lloyd-Martin is the Director of Search Strategies for webSourced, Inc, and author of the ebook Successful Search Engine Copywriting.

Want to discuss or comment on this story? Join the Forecasting Strategies for PPC Budgets discussion in the Search Engine Watch forums.

Search Headlines

NOTE: Article links often change. In case of a bad link, use the publication’s search facility, which most have, and search for the headline.

Online’s top 25 search engines
The Guardian Feb 24 2005 12:52PM GMT
Ads Embedded in Online News Raise Questions
The New York Times Feb 24 2005 12:49PM GMT
How Google’s Blogspot Helps Spread Unwanted Software, Feb 24 2005 12:46PM GMT
AOL goes public with local search
ZDNet Feb 24 2005 12:37PM GMT
FindWhat Blames Low Growth on Quality Control
ClickZ Today Feb 24 2005 1:17AM GMT
The Four Horsemen of Vertical Search
ClickZ Today Feb 23 2005 10:37PM GMT
Yahoo stakes out new territory in Hollywood
Christian Science Monitor Feb 23 2005 10:34PM GMT
Who’s the coolest blogger of them all?
CNET Feb 23 2005 10:32PM GMT
Oneupweb Study Finds Longer Search Phrases Convert Better Than Shorter
Search Engine Watch Feb 23 2005 10:29PM GMT
New Search Engine Tracker Launches
iMedia Connection Feb 23 2005 7:06PM GMT
A JavaScript That Kills Google AutoLink Links
WebProNews Feb 23 2005 4:44PM GMT
A9 Search Looks to Patent Its Shtick
Internet News Feb 23 2005 4:36PM GMT
Old Microsoft ties to new Google feature
CNET Feb 23 2005 4:25PM GMT
Small Businesses Spend on Local Search Feb 23 2005 4:24PM GMT
Google empowers a quiet invasion
The Deseret Sun Feb 23 2005 4:23PM GMT

Related reading

Simple Share Buttons