Which Test First?

Increasing numbers of experienced search marketers are realizing that paid placement search is far from a set-it-and-forget-it channel. The competitive search landscape shifts constantly, searchers change habits, and the search engines’ bidding structures, algorithms, and interfaces evolve. To stay on top of the landscape and ahead of the competition and to continually improve ROI (define), a campaign must change as well. You must know which changes to make and what you have to test.

Regardless of whether you manage your own campaigns or have your agency manage campaigns for you, there are always tests you should be running. But since all tests involve the time and costs of planning, executing, and analysis and the to-do list of possible campaign changes, enhancements, and tests is always long, it’s an ongoing challenge to decide which campaign features to test first.

Where do you begin?

Start testing the campaign segments that can make a significant difference in profitability. Whatever part of your campaign will make you the most money is the element you should test first.

With that, I’ll go through the cost-benefit analysis of different types of testing. I’ll start with the engine-side experiments — experiments that deal with your keywords and how your ads appear in the engines. Then I’ll move into experiments that touch on the site itself. Finally, I’ll look at testing out new search technologies.

Engine Side Experiments

Keyword expansion. The most commonly recommended engine-side experiment is keyword expansion. Yes, keyword expansion is an experiment. Some keywords will work well, others will fail.

Before you revisit keyword expansion, think about whether you’ve already hit the point of diminishing marginal returns. At some point, the additional work to generate keywords further down the long tail, and tuning the creative for those new keywords (assuming you aren’t using dynamic keyword insertion), isn’t worth the costs of the labor and testing involved.

Meanwhile, broad and phrase match (or in Yahoo, Advanced Match) are designed to capture tail keywords you haven’t explicitly included. Some engines are better than others in the way their systems implement broad matching, but at a certain point, ads and campaigns are sufficiently relevant for the searcher that keyword experimentation isn’t the best use of your time.

Creative testing on power keywords. All engines now reward a high predicted CTR (define). If the engines think your ad will be well clicked, they’ll let it appear at a higher position, without a higher bid cost.

That’s why power keywords, the keywords with high click-through volumes and rates, are more valuable than ever. Effective creative on a power keyword makes that keyword work even harder, both at bringing more visitors to your site and at getting the better predicted CTR you need. This is why testing creative on power keywords — an often-overlooked test — can yield dramatic results.

Start creative testing on ad groups that contain power keywords, because this allows you to capture material gains (assuming the experiments are successful). Then, work your way down your keywords based on total inventory (popularity of the term), keeping in mind your typical conversion rates.

For example, if two keywords are equally popular and one already has a higher conversion rate, test new creative on the highly converting keyword first. Any gains are multiplied by the high conversion rate. That multiplier is almost like compound interest.

Campaign reorganization. Campaign reorganization is another often-overlooked test. It may be as simple as moving keywords out of ad groups that contain too many unrelated words. It could also be something quite complicated, like an entire restructuring of an existing campaign based on data collected about conversion rates or click quality.

Restructuring power keywords is, of course, more likely to achieve a greater positive outcome than restructuring your tail keywords.

Site-Side Experiments

Landing page testing. Landing page testing and tuning is the most popular site-side testing, and with good reason: a poor landing page experience can cause the site visitor to use the dreaded “Back” button.

When a paid-search visitors hit the “Back” button (which happens a lot), you’ve just paid for visitors who got nothing out of their visit — and who didn’t convert through your site. What’s more, they may have clicked the “Back” button because you gave a poor branding experience, which is never something you want. To ensure landing pages are effective, you’ve got to test them.

The rules for prioritizing landing page testing are the same as for keywords. Start testing on the campaign segments that can make a material difference in profitability, then work your way to the tail.

Offer testing. Offer testing is an offshoot of landing page testing that relates specifically to the offers you provide on the landing page. In some cases, offer testing needs to be done in conjunction with creative testing, because searchers expect to see offers that are relevant to the ad they clicked on.

Technology Enhancements

Every month there are seemingly more technology enhancements you can bolt onto a site or use to handle clicks. Based on tests from my internal teams, some of the latest enhancements show significant promise.

However, use the same top-down rule when evaluating any add-on technology. Ask yourself, will that technology have a material impact on the campaign?

Remember, review your campaign with a top-down perspective, and prioritize experiments based on their likely impact on the campaign.

A final point: In cases where you’re working with an agency, many of these tests require coordinating the agency and in-house teams. Make sure everyone brings unique expertise and knowledge to the testing process for the best results.

Meet Kevin at SES San Jose on August 20-23, in San Jose, California.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.

Related reading

caroline-reynolds-digital-breakfast
sears-site-search
hp
penguin-4-0
<