9 Tips for Organizing Your PPC Campaigns (2 of 2 continued)

6. Organize ad groups around keywords

Campaigns are for organizing your ads conceptually. By the time you get down to the ad group level, you need to start specifying ad groups in terms of specific keywords. Not concepts --keywords. So under the “Action” campaign for your movie rental site you might have the following ad groups:
• Schwarzenegger
• Stallone
• Bruce Willis
• Sci-Fi
• Monster
• Etc.
Obviously this can turn into a huge number of ad groups very quickly. That’s the idea. It’s much better to have 50 ad groups with 20 keywords per ad group than to have two ad groups with 500 keywords each. Why? First, using this method you can write specific ads for each ad group that actually contain the keywords
under that ad group. For example, the “Schwarzenegger” ad group could contain many keywords, such as:
• Schwarzenegger flicks
• Movies starring Schwarzenegger
• Movies starring Arnold
• Arnold Schwarzenegger
• Best of Schwarzenegger
• Schwarzenegger action movies
• Etc.
(Are you starting to get the idea of how your keywords can balloon?)

The ads under this ad group will all contain the word “Schwarzenegger” and therefore there will always be a match between the
keyword searched and the ad displayed. Not only does this create the critical keyword/ad relevance factor that Google likes, but you’re likely to see a better click thru rate because the visitors see the relevance too. Both of these factors will improve your quality score and result in a lower cost per click (CPC).

Another reason to use plenty of ad groups is that you will find it easier to determine which ads are working well for you by looking at it from the ad group level rather than zooming clear down to the keyword level and peering at a massive list of keywords. It’s a way of breaking the data down into manageable chunks just as we did at the campaign level. Again, you’ll want to triage at the ad group level before diving into the keywords themselves.

7. Use the “Sculpture method” to get your campaigns under control

Let’s say you’ve set up 20 campaigns with 20 ad groups per campaign. You have 20 keywords and two ads per ad group. That’s 8,000 keywords and 400 ads to keep track of. Now it’s time to start analyzing their performance and making adjustments. But how?

Chances are, in the beginning you will use the “sculpture”
approach to managing your campaigns. This means that you will notice situations that look like they could be improved and make little adjustments in an ad hoc fashion, chipping off some CPC here, building up some keywords there. For example, you may notice that you are showing up near position one for a given ad
group and, looking closer, you see some keywords that might get you into a decent position by bidding significantly lower. This is a perfectly fine way to gradually mold and sculpt your campaigns into the basic shape you want.

As you do this, you must always be aware of the date range you are looking at. The metrics for one date range may be completely irrelevant or even misleading to the question you are currently trying to answer. On one hand, you want to make changes based on the most recent data. If you make a change to a keyword bid based on the last months’ of data, you may not be factoring in that the CTR for that keyword has changed significantly in the last week. So all things being equal, you want to make changes based
on the most recent data. On the other hand, you don’t always have enough data recently to reach a valid conclusion. You don’t want to make decisions based on the behavior of just a few individuals, so you may need to expand your date range at times. Of course if expanding the date range means you muddy your results because you’ve made big changes during that time—well, now we’re back to the first case.

There’s more to optimization than just making bid changes. Optimization also includes modifying ad copy so it flows better with the keywords that generate traffic and modifying landing pages to meet the expectations set by ad messaging. Pausing or deleting poorly performing keywords is another effective way to improve and optimize campaign performance.

Obviously, keeping track of many ad hoc changes is tricky. An indispensible tool when using the sculpture method is the “My Change History” located in the Tools section. If you are ever in doubt about why some metric has suddenly changed, you’ll often be reminded of what caused the change right there.

8. Conduct controlled experiments

The “Sculpture” method is best used in the beginning as a quick and dirty way to get your campaigns under control. Once you’ve achieved a certain level of stability, however, you’ll want to switch to more controlled experiments in order to determine ways to optimize your campaigns. The sculpture method is based on gut
instinct and assumptions—for instance, you may assume that being in position four for a lower cost is more optimal than being in position one for a higher cost. Controlled experiments allow you to turn assumptions into facts. If you think being in position four is more cost-effective, find a keyword that has been consistently showing up in position four for a while and test it.

There are basically two types of tests: a “Time Split”, where you compare the performance of elements across consecutive time spans, and an “A/B Split”, where you compare the performance of elements against each other simultaneously.

Generally the A/B Split is preferable because with the Time Split you introduce the time factor. For example, if you compare an ad’s performance in October against its performance in November, changes could have as much to do with the approaching holiday season as with changes to the ad itself. However, sometimes the Time Split is your only option. In the above example you would use a Time Split because Google does not provide a way to simultaneously apply two different bids to a given keyword.

To execute a Time Split on bid adjustments, mark the date, and bump your bid to push the ad into position one. Let it run that way long enough to collect a significant amount of data. Then check to see if your cost per conversion has changed significantly. Simple.

A/B Splits are also straightforward. Simply set up two distinct versions of the ad you are testing and let them run long enough to collect significant data. One thing you will want to do when testing ads is go into the campaign settings and ensure that “Ad serving” is set to “Rotate” rather than “Optimize”. This will cause both of your ads to be displayed a roughly equal number of times. The “Optimize” setting, (which is set by default), will cause the better performing ad to show more often. But you aren’t leaving it to Google to automatically optimize for you—by conducting your own tests you’ll be able to expedite the process of finding the most effective messages.

While A/B Split tests on ads and Time Splits on bids discussed above are among the most important types of tests, there are actually several others you should consider:

Landing Page – comparing conversion metrics with different versions of the landing page

Day parting – testing performance between the same ads during different parts of the day or days of the week

Geotargeting – comparing performance from different location either within the your country or worldwide

Match type – testing the performance of exact, phrase, or broad match keywords

Regardless of what type of test you are conducting, three key points about controlled experiments must be emphasized:

1. You must get significant data. If your control condition gets two conversions and the test condition gets three conversions, you are not ready to conclude this experiment. Those numbers just aren’t big enough to be statistically reliable.
2. There should be no other major changes other than the factor you are testing. For example, if you are testing the effect of a bid change, you certainly should not change the ad text associated with that keyword or the design of the landing page. Those are pretty obvious. But some changes are not as obvious and not even subject to your control. For example, a competitor who suddenly enters the scene with a very compelling offer could throw off your results. In order to avoid the possibility of random factors throwing off your results it’s best to conduct experiments as quickly as possible. That means conducting experiments on your highest volume keywords.
3. When measuring test results, use rates or percentages rather than absolute numbers. For example, while comparing the performances of two ads, conversion rates are a better metric than the number of conversions. This ensures data is not skewed by the level of impressions an ad receives.

Once campaigns attain a level of stability, it’s best to set-up a systematic process of reviewing performance. Set benchmarks for performance and use the process to effectively monitor, regulate, modify and optimize keywords and ads against this benchmark.
Analytics can help expand your keywords effectively and sync your campaign with seasonalities and industry trends at this stage.

9. Use a Tool Like LyrisHQ PPC Management

While the AdWords interface is quite capable, it is far from perfect. To expedite and extend the management of your campaigns in a powerful way, you can use a tool such as LyrisHQ PPC Management.
Lyris HQ provides an enhanced interface to the same data you manage through the AdWords site. It incorporates visual data representation techniques that make it much easier to identify abnormal data that might require your attention. It also includes a “Problem Campaign” report that automatically identifies those campaigns, ad groups, keywords, or ads that aren’t performing well and it puts them in a handy list.
Lyris HQ uses advanced Web 2.0 technology for a friendlier environment with a lot more dragging and dropping and a lot less clicking and reloading. This is a huge time saver.
Integrated keyword research tools go well beyond those provided by Google, allowing you to more easily build the most complete set of keywords.

The Lyris HQ PPC Management tool also emphasizes the tracking of campaigns for web analytics. Rather than a single tracking parameter, the system will automatically append a set of tracking parameters that allow you to easily segment visitors from any campaign, ad group, keyword or ad. When combined with the organizational structure described above, Lyris HQ PPC Management gives you the ability to perform granular analysis on your campaigns based on visitor behavior after the click.