Paid Search Ad Copy Auditing

Hello, I’m Brian Rauschenbach. I am a principal at Add Three. We are a search engine marketing agency located in Capitol Hill in Seattle. Today I am going to talk to you about paid search ad copy ad testing.

We get a lot of questions from potential clients and current clients on sort of effective ways to optimize your ad copy on your paid search ads. We have this example today on a big head term for one of our clients and the keyword is babysitter. It is a keyword that is searched a lot on both the head terms and with the geo keywords against it. So, this is the search result that we’ve taken from Google. In this example, you’ll see that our ad copy actually has a localized listing against it, and our big competitor’s has local in the listing, but the Seattle city has not been picked up in the ad, even though we didn’t give Google a Seattle indicator.

So, a couple of the things that I want to talk about is sort of effective A/B ad testing and some best practices you could use around this. In this test here, we’re basically showing a couple different examples here that I want to highlight. One is localization of the ad. So, we’re doing it both by creating a geo ad group with Seattle as the ad group city location, and then what ends up happening is on a head term like “babysitter,” we’re still picking up a local region. Then we also use the ad copy behind it to say Seattle babysitters, where you’ll see the other ad in here is just taking local because it’s a head term and they’re figuring that this term is actually getting searched by a lot of people in different regions and they are just putting babysitter in the copy.

When you look at this listing in Google, the babysitter gets highlighted in both areas, and if you see down here in the actual description, the ad copy description, we usually do a title case. So, we will capitalize each of the first words all the way through and then usually use an exclamation point or something to sort of draw some emphasis to the ad.

One of the things that I like doing is having a pure A/B ad copy test. In our AdWords campaign, we usually have only one ad running against what we call our champion ad so that we can get out data pulled together faster and be able to report on which the wining ad is.

So, localization is setting up your geo ad groups so that you’re making sure that in your big local regions, where your conversions are the strongest, that you actual have ad groups specified for that.

Using the keyword insert tag is sort of a common practice. What we usually do is make sure that you are using that so that the title casing is being effective as well.

Strong call to actions in your ad copy. You’ll see here that we’ve got find and join free as our call to actions. On this ad, they’ve got search, fast and easy, guaranteed results as sort of a statement but can be considered a call to action as well.

Another area where we’ve seen a lot of success in ad copy testing is Google gives you 35 characters per line for your description text, and then it puts it in this long sentence. If you are in the top two listings, sometimes you will see the extended titles appearing where they figured out in their algorithm that it clicks better. Then if you see the ads that are on the side, they are usually in two lines. So, another thing that we’ve tested in the past with different clients is to shorten the actual description line. So, even though they are giving you 35 characters, it doesn’t mean that you have to use all 35 characters. In this instance, we might write an ad that just says, “Find sitters. Join free,” and test that against the champion ad.

One of the other things that we always try to do to is isolate the test so that we’re only testing the description. We’re only testing the title tag on the ad, and/or in some instances we’ve even tested brands without the Ws in it against the www. brand URL on there and have seen different result sets. So, I can’t really say that you should always have your brand sit at the root domain URL. In some instances, the www actually tests out a little better. A lot of that ad copy testing is done in the trade term buckets, and so we have those all separated out, so that would be the display URL piece here.

Another thing that we’ve seen recently is people are putting sort of a one- two step in their ads. So, if you can imagine this would be the first one and it would be “find sitters,” and then the second one would be a call to action too. Then you might just say, “Join for free!” These ads work better we found if you are vying in the third through fifth position, because they’ll get broken up on two different lines instead of showing up long on one line and then it usually doesn’t translate as well. That is just sort of another thing that we’ve been doing some recent testing with lately. We’re seeing it happen across other industries, like dating and people search, and they’ve had some pretty good success with it.

One of the other things that we usually do, like people always ask, “Well, how much data is enough data for this ad test to be successful?” What we found is that it varies from client to client, but usually if you are achieving at least 1% click-through rate on your ads and you pull in about 20,000 to 30,000 impressions, depending on what your conversion rate or where you have your conversion pixel firing, it is usually enough data right around the 30,000 impression mark to make sort of an informed decision on which ad is actually starting to edge out the other ad. We like to do this with just the two primary ads and not having five. Sometimes we look in campaigns and we see five different ad copies all being tested at the same time. That test to get those results back on those types of tests usually take two to three months, and in the meantime you are lowering your quality score on that ad group if you’ve got an ad that isn’t polling as well as the other one. Google usually will go through and start serving the highest click-through ad with preference anyways. Even if you are doing it as a 50/50 test, we’ve seen that in the past as well. So, my best practice is to just keep it a pure A/B ad test, so you always have a champion and a challenger ad.

Another area where people make a big mistake when they’re doing their ad testing is that if the is our primary ad right here, you want to duplicate this ad copy as your champion ad, and then you want to introduce the challenger ad and start those two pieces of ad copy at the same time. Once the ad copy has been approved by Google, usually happens within 24 hours, then you will pause the original ad so that you have a true start so there isn’t any layover data that is being transferred over to the test. What we’ve seen before in the past is ads that might be getting assists from other head terms or other keywords might get triggered in a conversion cycle, and one will show up in a database that happened on traffic that was maybe a week before you started your test. So that’s another really, really important piece that a lot of people overlook when they are doing their ad copy testing.

I’ll have a couple other examples that I’ll put in some notes on some other areas where we do a lot of social listening with tools like Radian6 and Social3i where we basically listen to what is happening in the social space to actually dictate what we are going to use for ad copy. One of the things that we might have heard in the past was that people really liked parent reviews or keywords like “trusted” for this type of business, so we have actually incorporated them into the ad copy and tested against that. So that is a good thing that you could basically use for what conversations are happening in Facebook and Twitter against your brand and incorporate that in to your ad copy.

That’s it.