Let me be the first to say it’s been a while since my last blog post, but to my defense I was on vacation in Mexico and thankfully did not have any random PPC thoughts during that time.

Anway, on Tuesday Google announced a new AdWords feature that could eventually become a “game changer” in the realm of PPC: AdWords Campaign Experiments. This blog post on the Inside AdWords blog provides some very basic information on what exactly Adwords Campaign Experiments is, but the best way to get a feel for how powerful this feature could possibly be for improving your campaigns can be found here. It is important to note that this feature is currently limited to only those advertisers who submit a form with their Client ID, not MCC.

An Overview of AdWords Campaign Experiments

Have you ever wondered what effect changing your Max CPC for a single keyword would have on clicks, click through rate, average position, and conversions while avoiding possible seasonality issues or other “shocks” that may affect performance from week to week and month to month?

Well ladies and gentleman, Google has taken a giant step towards helping you answer this burning question. Essentially, AdWords Campaign Experiments allows you test changes in keyword level bids, new keywords, and new placements during the same time period as your current settings. For a more in-depth explanation here are three examples:

Testing Keyword Level Bids

Prior to AdWords Campaign Experiments, an advertiser would be able to utilize a keyword with only one keyword level bid. In order to test the effect raising or lowering a keyword level bid had on conversions, clicks, and other statistics the advertiser had to change the bid and wait until enough data was collected to compare the two time periods. The important portion of this is to note the test was conducted over two different time periods. This means various factors such as seasonality, decrease in demand, or other factors could have a serious impact on your test.

AdWords Campaign Experiments creates a so-called “apples to apples” test. For each keyword they choose, an advertiser can use a control keyword level bid and experimental keyword level bid. As searches are conducted and the keyword is triggered, Google will determine which bid to utilize when displaying an ad. At the end of the time period in which you to choose to run the experiment, Google will have displayed ads utilizing each bid at the ratio you chose when setting up the experiment. That’s right; you control the ratio of your experiment whether it’s 50/50 or 70/30 and how long the experiment runs.

At the end of the experiment, Google will provide data suggesting whether the test was statistically significant for various statistics such as clicks, click through rate, and conversions. In essence, Google is providing you with crucial information concerning whether you should continue using the control bid or start using the experimental bid on an ongoing basis.

Testing New Keywords and Placements

Keyword research is an ongoing task for PPC managers because there are always untapped keywords that present tremendous opportunities for a client. When these keywords are added, PPC managers monitor performance to ensure they are performing well but sometimes the component of when these keywords were added gets lost in the shuffle. AdWords Campaign Experiments basically removes this time component from the equation and allows you to see the effect these additions have on overall performance over a set period of time.

For example, I’m currently only using the keyword “internet marketing company” in an ad group but over 1,000 keywords across all active campaigns. I want to see what adding the keywords “online marketing company”, “internet marketing agency”, and “online marketing firm” has on overall performance. Since I’m managing a ton of keywords, I could forget not only to check the performance of these additions but also when they were added. To avoid this, I can set up an AdWords campaign experiment for testing these specific keywords over two months. By setting up an experiment, I will know when these keywords were added, how long they’ve been running, and more importantly what impact they are having on overall performance.

Testing Ad Groups

With Ad Group Experiments, an advertiser is basically testing the effect changes have on ad group performance rather than campaign performance. If I wanted to see the effect adding a new keyword to a pre-existing ad group would have, this is the type of experiment I would conduct. Also, if I’m currently utilizing a content network campaign with only text ads and want to see the effect adding display ads would have on an ad group’s performance, I would conduct an ad group experiment. The only drawback I’m interpreting from the video tutorials (and which I may be wrong on) is that an advertiser will not receive that same data/information showing if the test was statistically significant as they would if they conducted a campaign experiment.

Lingering Questions

Some questions definitely popped into my brain while learning more about AdWords Campaign Experiments.

  • Can only 1 experiment be performed at a time? If I want to conduct 1 test for 30 days and another for 90 days, do I have to wait until one of these tests has already been conducted? Viewing the campaign settings within the tutorial videos suggests that only 1 experiment can be run at a time but I’m not sure on this one.
  • How is this data going to look in Google Analytics? Do I need to give each keyword its own unique URL in order to identify it? For example, if you currently look at the “Clicks” tab in Google Analytics each keyword has its own cost data. If you perform an experiment on a keyword using multiple bids, is Google Analytics now going to display multiple versions of this keyword within this tab or lump them all into one?
  • Is the same limit of 100 active ad groups in a campaign still going to be in place with this encouragement to create multiple ad groups for experimental purposes?
  • How is quality score determined for new experimental ad groups? If I want to test 2 ad groups that will use 3 out of 4 identical keywords, will the quality score of keywords in the control group be higher than the experimental group due to past history? If so, this affects ad rank and a host of other factors thus making the experiment somewhat pointless. This is one burning question I need Google to answer.

Final Thoughts

As said, AdWords Campaign Experiments has the potential to truly be a PPC “game changer”. Not only does it provide lots of additional testing in an AdWords account but it also opens the door to developing even more granular tests. In addition, Google is basically offering a free feature that plenty of 3rd party bid management tools (which you have to pay for) include in their platforms. Is Google working towards offering more features bid management tools do for free? I’m not sure about that slippery slope, but it does make an advertiser who has considered purchasing a bid management tool really look closer at whether it’s a good investment at this time.

In any case, AdWords Campaign Experiments is fresh off the presses resulting in even more feedback, ideas, and questions to likely arise once more people have this new feature enabled in at least one of their accounts. Look for future posts concerning setting up experiments once I have an account that uses AdWords Campaign Experiments.

Stay up to date on the latest online marketing news by subscribing to DragonSearch’s RSS feed, following us on Twitter, or becoming a fan on Facebook. Also, you can follow me on Twitter @Andy_G_PPC.

This entry was posted on Thursday, June 10, 2010 and is filed under Pay-Per-Click.

Related Posts

Your thoughts? Please comment below.