SiteProPPC » A/B testing

Ad Optimization….and A/B Testing Part 2.

admin — Fri, 20 Nov 2009 17:42:10 +0000

This is a follow up to our initial article Ad Optimization….and A/B testing Part 1. In that article I said that if you are serious about testing ads, you should not use the default setting that has Google optimize your ads for you.

In this article I would like to show you a real life example that will help illustrate that Google Ad optimization is not a necessarily a good idea.

The figure below shows the ad statistics for one of our clients. (Our thanks to The College of Golf for letting us use this information in our example.)

Figure 1.

Click Thru Rate.

If you look at the two ads that we are going to use for comparison, you will see that one has a CTR (click thru rate) of .65% and the other a CTR of .36%. If those were the only two ads, the Google Optimization feature would automatically start sending all or substantially all of the traffic to the ad that has the .65% CTR.

That’s a good thing right? After all you would get more clicks! Well it’s good for Google, but not necessarily good for your campaign.

Figure 2.

CTR or Conversions?

For a business owner, Figure 2 highlights the more important numbers. Notice the poorer performing ad in terms of CTR did much better in terms of conversions. The ad with the .36% CTR had two conversions at a cost of $13.81/conversion, where as the better performing ad with the .65% CTR had only one conversion at a cost of $52.40.

Frankly, this is an extreme example….but these are the kind of numbers that make a PPC manager lose sleep. And these are real numbers, not something made up to make a point.

If these numbers continued to hold up it would make for a tough decision on whether to keep both ads or not, but I would most likely keep both. Naturally, you will keep the ad with the higher conversion ratio, but retaining the ad with the relatively high CTR will improve your quality score. (If you need a refresher on why CTR is important for quality score, check out this article.

Google Ad Optimization.

So, like I said, the Google Ad optimization algorithm would have picked the first ad and started sending nearly all the traffic to that ad…and it would have made that decision fairly quickly. Maybe in as few as 10 clicks each.

So if you are serious about testing ads, you will want to turn off the optimization option on AdWords and set the ads to rotate evenly. This puts YOU in control and lets you decide when to retire your ads.

Figure 3.

Need more proof? One last surprise!

The primary point that I hope to make in this article is that you need to test ads, but you need to do it with enough information. Most PPC Managers feel that the Google Optimization Algorithm makes decisions much too quickly. And actually many folks that run campaigns will make decisions on ads with too little data.

Here is an eye opener–

Take a look at figure 4 below.

Figure 4.

Identical.

The ads that we have been comparing are identical! Same ad group, same keywords. I discovered this technique by accident (obviously!), but I use it now occasionally as a sort of statistical barometer. Hopefully this underlines the need to collect enough data to make good decisions regarding your ads, keywords, etc. in your Pay Per Click Campaigns. Clearly at this point there is not enough data to make a decision on these ads. Making decisions too early may hurt you in the long run in the quest for finding the best converting ads. Be Careful!

If you think you’re Adwords campaign might be wasting money, contact me for a free Adwords evaluation. I’ll be your second pair of eyes to double check your campaign for these kind of issues. Because when you optimize your Google Adwords campaign, you want someone optimizing it to make money for your business, not Google’s.

Ad Optimization….and A/B testing.

admin — Wed, 11 Nov 2009 17:56:56 +0000

Everyone would like to have their PPC campaigns operate more efficiently — with better conversion rates and lower cost per conversion. For most, this means repeated manipulation of keywords, bid prices, regional targeting and ad scheduling.

Those things are important and they all need to be done, however what is often over looked is ad testing. (This is often referred to as A/B testing, but actually A/B testing can apply to other PPC components as well, like landing pages, etc.)

For obvious reasons, ads are an important part of any PPC campaign. They are the beacon for potential customers to find your site and make a decision on your product or service.

What is not so obvious, especially for beginners, is how important ad testing is to the overall effectiveness of your campaigns and how much customer traffic can be affected by small ad changes.

How do ads affect other components?

Quality Score.

One of the cornerstones of AdWords is the quality score. Briefly, quality score is sort of a multiplier or factor that determines your effective bid price when competing for position with other advertisers on AdWords. If your quality score is 4 you have to bid much higher (say twice as high or more) than someone that has a 10 quality score for the same ad position.

Ads are part of the equation.

While no one outside of a few Google employees knows the exact quality score equation, it roughly boils down to the combination of keyword, ad, and landing page. We could devote an entire article to quality score….and we probably will in the future, but the ad affects the equation in two ways. Relevance and click thru rate.

Relevance.

Google measures keyword relevance in at least two areas. They measure how relevant the keyword is to your website (landing page primarily) and how relevant your ad(s) is to the keyword/landing page combination. There is actually a lot of mystery regarding Google’s actual process here, but main point is that they make this calculation, so it pays to make your keywords show up in your ads and on your landing page.

Click Thru Rate.

This is where your ad can have the greatest influence on the overall performance of your PPC campaign. Google (and Yahoo, Bing, Facebook) seem to primarily measure performance/quality by the click thru rate. Of course they have economic incentive to do just that. The more clicks, the more revenue they bring in.

The click thru rate affects your campaign in two ways. First, on AdWords it helps determine your quality score, so your actual cost per click can actually go down if your CTR goes up. And actually the same applies for Facebook and others as well. So you want to keep your CTR rates as high as possible (with some obvious caveats, but we will save those for another day).

How to test your Ads.

Since Google AdWords accounts for 80%+ of all ppc advertising, we’ll focus on that, but the basic techniques apply to all PPC advertising.

The first thing, yet by far the most important, is to have multiple ads running for a particular ad group. No brainer? You would be surprised to see how many accounts have only one ad running in some, if not all of their ad groups.

Now you are probably saying to yourself, "not me, I have at least two ads running for each ad group, I remember setting them up". Maybe you do, but maybe you don’t….at least not from a practical standpoint. How is that?

Google likes to help….whether you want it or not. When you set up a campaign there are a number of settings for you to select. Under the Advanced Settings section you will find a setting like this:

Ad delivery: Ad rotation, frequency capping. The default setting is Optimize.

Optimize or Rotate.

Professional AdWords Managers rarely, if ever, set the ad rotation to Optimize. Why? Because the Google optimization algorithm has to make some assumptions that you really don’t want to make if you want to truly find the best performing ads.

Want happens is that Google monitors your ad performance and quickly decides which ad is the better performing one and then essentially routes all the traffic to that ad. It does that by determining which ad has the better click thru rate…..which is not necessarily what you want. For a couple reasons.

Premature Evaluation.

Sorry, I couldn’t resist. There is a general rule of thumb that says you should have at least 30 clicks to properly evaluate an ads performance….and actually 100 clicks would be much better. When the Google ad optimizer is running it tries to pick the best ad with fewer clicks….even as few as 10. So you have the danger that it will pick the wrong ad as the best one. Once it does it effectively shuts down the other ads you have in the ad group. So even though you think you have three ads running for an ad group you may only effectively have one ad really running.

Conversions.

The real issue is that Google is probably optimizing on the wrong statistic. What is really important to a business is the CPA (Cost Per Action) in its PPC Campaign. The ads with the highest CTR (Click Thru Rate) are not necessarily the best converting ads. In fact I would say more than 50% of the time, the lower CTR ad has the better cost/conversion or CPA. So once again, you may have set up three ads, but you will never know which one is the better converting one because the Google ad optimizer is showing the best CTR ad 90% of the time.

Real Life.

In the next blog post, I will give some real life examples and sample ads….stay tuned.

In the meantime, if you want a little more information on testing ads for PPC, check out this link:

http://www.ppchero.com/landing-page-testing-how-to-test-and-what-to-test/#more-4326