Make Your Ads Better: Three Powerful Techniques

Tomorrow I’m giving a talk at the eMetrics summit in San Francisco. A search marketing perspective can often seem “refreshing” in the context of the discipline of marketing in general; it even seems to go against the grain a bit. I’ll be telling the eMetrics audience a bit about how I use core stats in paid search to iterate much more rapidly than most analysts are used to. In other words, I won’t be talking about the fancy stuff you get from analytics reports, but the bread-and-butter stuff, like CTR and even initial Quality Scores (yes, that’s a statistic!).

The majority of data analysts look for patterns after the fact, gathering information from large volume corporate campaigns and developing strategies in response to research. While the hunger for more rapid iteration is starting to trickle into many larger companies’ web strategies, the lightning-iterative environment of paid search seems hectic and loosey-goosey by comparison. We are the proverbial street-smart Fat Tony (or Joey Metallica) by contrast with straight-laced Banker John, who believes in data almost to a fault.

In the SEM field, a debate continues as to how much automation we should be using, and how much human interpretation is necessary. It’s an important debate, because with all of that rapid iteration, the majority of human analysts are far worse performers than computers. But not all are. And computers are still notoriously poor at theory-building and inferences in complex environments… at least the computers that use the pea-shooter algorithms developed by lone developers or small teams at small software companies.

Recall that Google actually has amazing computing power and highly intelligent algorithms. In that environment a human analyst can leverage Google’s power, for free, and do very well.

In such an environment, we know that new campaigns are under the gun, facing a potential cold shoulder from Google’s quality score algorithm. You must be relevant from the start, gain high CTR’s on your ads (all else being equal) in the early going, or face higher costs and even a spiral of failure.

Three classic techniques can get you to better ad copy more quickly, without brute force testing of many variations. By “better,” I mean higher CTR. We’ll need to make Google happy in the early going to build up quality signals that will give you an account-wide benefit. Later, you can drill down more carefully on ROI related measures.

Dynamic keyword insertion in title. The first technique is fairly simple. Unless you know better from the beginning, try dynamic keyword insertion in title, but still try to pay attention to decent alternate text in cases where the matching won’t work. Yes, I generally warn clients not to overuse dynamic keyword insertion because it ups CTR’s too much without an ROI benefit in many cases. But to start, it’s helpful to have it in your arsenal. The basic theory is we don’t know what people are thinking yet, but we do know they’ll click on a title that matches their query.

So, for a product offering archived, remastered football coverage (hypothetical example), you would use:

{KeyWord:Football Classics in HD}

If someone types in “packers videos,” the title would say “Packers Videos.” At least you haven’t hurt your chances.

For body copy, plain and navigational beats flair and color most times. Ad text that says “Does Sterling Sharpe look better in HD?” will fail (even if the query were about Sterling Sharpe) compared with clear text that says something like “Free trial offer – classic NFL game archive in HD. Experience it now!” The latter explains the offer, doesn’t try to be clever, and contains a call to action. (Actually, twice, because mentioning a free trial is a quiet call to action.)

Use “skinny persona research” to govern your early ad copy choices. Or if that’s even too much research, just do less of it, or none at all, until you stop making ridiculous mistakes due to flawed research. Fat Tony doesn’t have time for the in-depth research conducted by Banker John. But what he can do is use his head, and pay attention to ‘lite’ stats about his business all the time, as they emerge. An inexperienced attempt to relate to video gamers, for example, might try overly hard to relate to the “vernacular” of youth: a challenging and often unrewarding effort in the paid ads space. A ham-fisted attempt to squeeze demographic assumptions (or “research”) into the tiny ad space might take the assumption that a certain game is targeted only to youth, primarily to Asian youth, or (let’s say a video hockey game) primarily to male white kids who predominantly grew up in Moose Jaw, Saskatchewan.

It’s definitely important to think about personas in writing your ads, and these might change from ad group to ad group. But the key is not to let persona thoughts override winning ad tone and tried-and-true navigational and psychological triggers. And when so many would-be creative writers actually insert wrong personas into ads, the maxim “less is more” applies in most cases to the targeting effort. Use a light touch here.

You can’t ignore what the market may be looking for, but don’t make a mess of the targeting process. For video gaming in general, your first thought might be “suburban kids.” Your second thought, though, might be “not only kids.” Fat Tony knows that the average age of gamers is over 30. So, your market is basically gamers, and anything that might entail. The same thought process goes for “niche” games and content. People of all ethnic groups, believe it or not, like hockey. Nearly 50% of football fans are women. Do football fans live in the city or country? Both. White kids like Eminem, and they also like 50 Cent. “Niche” video games find a broad market, and even if you could identify the niche audience, would you speak to them differently? Generally, only if they speak a different language. Do some people live in low income areas or areas that have lower broadband penetration? Sure, and you might someday finagle a way to incorporate that insight into the demographic targeting settings that govern your bidding, as long as you know for sure if those insights mean your targets will be more or less likely to buy your product as a result of their low-income, slow-DSL plight. But you ain’t gonna squeeze all of that insight into your brief text ads. So you probably shouldn’t try.

What we do know is that 100% of the people looking at your ad are searchers. Most searchers dislike ads that seem to be laser-targeted to someone else. At the other end of the spectrum, most searchers dislike ads that are overly generic. Many searchers will recoil at ads that seem to be written in “ad agency” style with an overly stylized “tagline” feel. Searchers who are misled and feel duped once they arrive at a landing page will stab at the back button, making your trickery penny wise and pound foolish as quality score algorithms punish you with high minimum bids.

Possibly the most important thing to do in your early ad tests, then, is not to trip over your own shoelaces trying to be too clever. Your plan needs to be built around the granularity of your prospects, the psychological triggers common to many searchers and customers, and avoiding mistakes that turn searchers off. That sounds funny to say when, depending on the query, upwards of 98% of people won’t click on your ad. But within your 1-2% sub-sliver of ad clickers, all else being equal, you should focus on developing consistent techniques to reach the 2% or 3% level faster as opposed to languishing at 0.5%. (In some niches, of course, you will reach 8-10%. I am talking about the high volume stuff, mainly.)

You’ve only got 25-35-35 characters and a display URL to work with here. Even so, writing these ads is a sensitive task and early success is vitally important to the economics of your campaign. Find a winning strategy inside of a month, and you are into the initial stages of refinement far ahead of competitors who take months to study their options.

Good luck.

P.S. Rapid iteration won’t work with low-volume campaigns. Because of that, it’s even more important to start well on lower-volume accounts, because it will be impossible to test your way to the best answer in any reasonable time frame if you start poorly.

Andrew Goodman is the founder and principal of Page Zero Media and author of Winning Results with Google AdWords. The Paid Search column appears Mondays at Search Engine Land.