When recently evaluating a new business opportunity, we encountered a rather frustrating problem with Google’s Traffic Estimator. At Rise, we utilize the Traffic Estimator to help estimate total media spend for upcoming paid search campaigns and to analyze if opportunities will be profitable. We began noticing variances between the costs the Traffic Estimator was projecting and the actual costs once the campaign launched. It was concerning to us that the Traffic Estimator may not be the most accurate tool.

We reached out to our Google team to determine if the cause of this was possibly related to the impact of quality score as we hypothesized, and to better understand how quality score factors into Google’s estimated cost-per-click price. Unfortunately, Google did not have this information.  Given this, we had no choice but to investigate ourselves and attempt to draw some conclusions.

We chose three Rise Interactive clients as our “lab rats,” two national clients in different industries and one regional, hence geo-targeted, client. To qualify, each tested keyword needed a minimum of 50 clicks during the month of July 2009. The position, clicks and historic CPC were actual numbers that we reported last month. We then included the average CPC per Adwords, the variance (real CPC—Adwords predicted CPC) and the quality score for each match type. The historic CPC includes an average of all match types for that keyword.

*A negative CPC Variance indicates that our Historic CPC came in lower than what the Adwords Traffic Estimator said the Average CPC would be. A postive CPC Variance indicates that our Historic CPC came in higher than what the Adwords Traffic Estimator said the Average CPC would be.

As we had suspected, the theory that a high quality score would lead to a lower historic CPC than the Adwords estimated CPC held true for the most part for the national clients.  Our test proved that those clients were indeed paying less than what the Google Traffic Estimator was telling us. However, we were concerned with the extreme variances between the two—one averaged 50% lower than Google’s estimate and the other averaged 20% lower.

The results of the same test applied to the regional client produced unexpected results. Our test showed that we were paying more than what the Google Traffic Estimator was predicting despite often having a quality score of 10 (the highest score possible). Note that we were sure to request geo-targeted estimates to keep the results as accurate as possible.

After sharing our findings with Google, we were told that the Google Traffic Estimator pulls historical data to try and predict what will happen in the future, assuming everything stays constant. In the case of our regional client, because they are geo-targeted there is less data for the tool to pull from, therefore resulting in less accurate results. Google passed along a link to help shed some additional light on why the Traffic Estimator’s results may differ from real-time costs.

Because of the concerns customers have expressed to Google in regards to the inaccuracy of the Traffic Estimator, they have recently launched a new tool, the Bid Simulator. Instead of looking at accumulated historical data to predict the future, the Bid Simulator looks at the auctions each keyword has triggered over only the last seven days and re-runs the auctions with different bids to determine what would have happened during this time. Though it does not predict future outcomes, it does help you make more informed decisions and determine what you should be bidding on your keywords.  While this is all well and good for current initiatives that you are looking to optimize, it does no good for new projects you do not have more than seven days of data on, and are therefore without a frame of reference.

Time will tell if the Bid Simulator is truly a more reliable tool for estimating CPC amounts. For now, we’re wondering if any of you have experienced similar challenges when planning and analyzing possible paid search campaigns. If so, we would love to hear any of your findings and/or solutions.

More From RiseInteractive

• Bing vs. Google – To Stay or To Leave the SERP
• Google and Bing Go Social – Twitter and Facebook Updates to be Included in Search Results
• Google Analytics: Advanced Segments and Why You Should Use Them

RiseInteractive Recommends

• My Thoughts on the Newest Google Algorithm Update [w/ Results]  (StayOnSearch)
• Improve Your Website Traffic by Cleaning Up Your Google Index  (StayOnSearch)
• How Many Ways Can You Get Traffic From Google?  (Anthony Kirlew)

Ask RiseInteractive To Recommend Your Posts

This entry was posted on Monday, August 24th, 2009 at 3:55 pm and is filed under Paid Search (PPC), Rise Interactive. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.