PPC Question

[DenverKelly]

After watching today's excellent WBF by Brian on Adwords, I still have a question. At one point do you believe a PPC keyword or adgroup has enough data to make a decision. Brian mentioned 30,000 impressions. I've heard 200 clicks.

Is there a concensus or set of rules anyone could recommend as a guideline. I do PPC but it's not a daily focus, and find myself vacillating, and frankly, probably wasting time and money by not having a more defined approach to PPC.

[john4math]

You can't really avoid doing the math to figure out the probability that the better ad is actually besting the older ad.  It's a function of both the impressions and clicks so it's not easy to eyeball.  I threw together a little spreadsheet that should do the math for you if you put in the click and impression data.  I checked it with some data and I think it's working. It'll calculate whether we're 90% sure, and 95% sure which is better.  Just put in your ad numbers and you should be good to go!  Usually, if the 90% variation matches up with my expectations, I go with it.  Otherwise, you might wait for 95% just to be sure.

My worksheet: http://www.mediafire.com/?pmu2mc6n5f6vw5s

A/B testing people deal with confidence intervals all the time, so they know their stuff.  Here's some reading about it if you're interested:

• http://support.optimizely.com/kb/advanced/what-does-chance-to-beat-baseline-actually-mean
• http://visualwebsiteoptimizer.com/split-testing-blog/what-you-really-need-to-know-about-mathematics-of-ab-split-testing/

Hope this helps!