Posted by James Lee on 3rd December 2010

Experimenting on Your PPC Campaign

One of the frustrations of PPC management lies in the following set of facts:

  1. You must keep experimenting to improve performance
  2. You can’t perform true split experiments

In science class, we all learned that experiments should test just one variable at a time. Unfortunately, that’s a little hard to do in the world of PPC. Even if you change a bid price or add a keyword and measure the before against the after, you can’t be sure that something completely external to your campaign didn’t affect performance. (For example, say you sell penguin collectibles and want to compare your June search results against your July ones. If a popular new nature documentary on penguins comes out at the end of June, it’s really going to throw off your experiment.)

When it comes to ads, it can also be hard to determine what your audience is responding to.  Most ads contain more than one message in their three short lines, so unless you are meticulous about testing one slightly different ad variation against another, you won’t know if it was your call to action or the price in the title that got your prospect’s attention.

PPC experimentation is not an exact science, and some flailing around might be inevitable. However, here are some tools and tips we recommend for creating meaningful experiments that actually tell you something about your campaigns.

Tool: Google Adwords Campaign Experiments (Ace). This relatively new tool makes true split testing for bids, placements, new keywords and adgroups possible. Your control and your experiment groups run during the same period (you decide what ratio you want the two to show, from 90/10 to 10/90). At the end of the experiment, the tool will even tell you whether the differences are statistically valid. Read more about Ace on Search Engine Guide.

Tip: Your highest CPC keywords are the ones most worth experimenting on. This might seem a bit obvious, but it’s worth considering before you undertake a new experiment. Want to try to boost quality score? Increase exposure? Lower CPC? Don’t bother with your minor keywords. Sort your keywords by highest CPC. Those are the ones that are costing you the most, and the ones that are most worth experimenting on.

Tip: Google Analytics is your friend. If you never venture beyond your comfort zone, you may have missed a lot of ways to look at your Adwords account (under ‘traffic sources’ in the sidebar).  You’ll be able to identify problems (targeting, bounce rates, etc.) you had no idea were there and create experiments to see what’s going on and fix them. Spend some time getting to know Analytics, and you’ll be able to not only come up with experiments but also evaluate their results.

And of course,

Tool: ClickSweeper’s Ad Copy Evaluator. Yes, Google has its own ad copy testing. However, its primary criteria is clickthrough rate, which doesn’t necessarily translate to profitability for you. (Perhaps it’s no coincidence that giving preference to the most clicked on ad is directly profitable for Google.) Our PPC software uses both CTR and cost per conversion to determine the winning ad in each adgroup. Learn more about our ad copy management.

What have you found most useful when conducting PPC experiments?

For More Information

No comments yet!

Post your comments