Test or Die

We can all learn from direct marketers. They’ve dealt with ROI-based marketing for years.

Old-school catalogers may not yet grasp all the nuances of search engine marketing (SEM), but they do deal daily with maximizing ROI and profit from scarce resources (external mailing lists or internal customer lists). Catalog marketers deal with the scarcity of valuable customer lists or rental lists through testing. They always test new lists, new creative, new prices and new offers. Several catalog marketers shared with me recently that, depending on the season, they might allocate anywhere from 10 to 45 percent of their media and creative budgets to tests.

It may seem like a risk to allocate over 15 percent of a budget to testing, but catalog marketers know the biggest risk might be not testing at all. One catalog marketer emphasizes this point by sharing his “test or die” credo. He means if he wasn’t constantly testing new things, his business would die a slow, steady death. This truth puts a new face on the importance of testing, and it inspired this column.

The same analogy holds true for search marketers. In PPC search, it may hold even truer as PPC prices continue a steady upward spiral. Keywords you could afford top position on yesterday may cost more today. You won’t hit ROI targets unless you find a more efficient way to use those clicks, or a way to buy those clicks more cheaply. The only ways to buy the same clicks for less is to bid less (usually resulting in a lower position) or, in the case of hybrid auctions, get your CTR (define)up to a higher percentage (resulting in a better AdRank/higher effective CPM for Google).

Catalog marketers like the analogy of the psychic mailman. He waits outside the doors of your best customers’ homes. He has thousands of special product or service offers, and is poised to ring the doorbell only at the moment buyers start thinking about a specific product or need. If a customer began thinking about a new down ski parka, the psychic mailman would tear out the appropriate page from a catalog and hand it over.

Imagine if there were three or four different pages in that catalog that might work. A truly psychic mailman could pick the best pages to present.

Search engine marketing (SEM) can be your psychic mailman, delivering the perfect offer, tuned to the specific needs expressed by searchers seeking a product or a solution to a problem. Every time a Web site user types a term into a search box, he enters hunt mode — looking for information, a solution, or perhaps something to buy. The perfect offer, price, Web or landing page will change over time. That’s proven even in non-seasonal businesses. So a successful test provides search marketers with an answer to an optimization question that may only be appropriate for a short period of time.

Search marketing keyword placements are an investment. Past performance can help predict future results, but results can’t be predicted too far into the future.

Be sure to structure tests to provide the highest possible profit lift. The most common test is adding new keywords or engines to the campaign. The best way to test campaigns with new keywords or new search engines is to make a best guess as to copy, landing page and offer to get a baseline. You can do so quickly by starting at the top of search results (paying more for the fast data), or you can take a more moderate approach. Just because a keyword doesn’t hit your ROI target the first time doesn’t mean you should bid low, or trash it entirely.

First ask:

• Does your business have seasonality that could change conversion rates?
• Did you try the best landing page?
• Was copy written for the prospects and their state of mind?
• What was your competition doing at the time?
• Have you changed pricing?
• Has marketplace pricing for your product changed?
• Are you or your competition currently getting lots of PR or spending heavily in other media?
• Will your conversion rate change as you change position based on the search engine’s syndication network?You can test many campaign elements continuously. There will soon be opportunities to test even more variables based on demographic, psychographic and other data beyond simply which keywords users searched before arriving at your site.

  So remember: test or die.