Choosing what to test in your marketing efforts.

In the last issue I wrote about the importance of strong "creative" for newsletter marketing efforts in 2006-2007. Sadly, however, I concluded I didn't have the "magic bullet" to allow you to produce better creative.

Yet I can provide a personal checklist of things to test and things not to bother about when you look into revamping your marketing efforts.

Remember this when you test

Newsletter universes are almost never large enough to allow for statistically relevant testing. Cells of 5,000 are probably the most common, but they aren't big enough to give more than an "indication." So look for dramatic results, plus 30 percent or more from the test cell, before you think about declaring a test a winner.

One veteran publisher told me that, if you don't believe this, split 10,000 names A/B and mail each cell the identical package and be amazed at the different response rates you might well get.

Be cautious. If you are testing any really dramatic changes, be cautious. Don't mail all 40,000 of your new package; send perhaps 10,000 of those and send the rest your current control. If the test is a winner, of course, you roll out ASAP to the rest of the universe. But if, for some reason, it bombs totally, you haven't put all your eggs in that basket.

Elements to test

* An entirely new and dramatically different package.

* A sample issue package (if you haven't done one in a while, or ever).

If you can afford it, a forced free trial (expect it to cost 10 times the control package). There are almost certainly folks on your prospect lists who simply can't be sold a product they haven't seen ... not even if Woody Allen is writing your sales letter.

* E-mail marketing. You could test this by offering things like special reports to subscribers you already have ... and their e-mail addresses as well.

* An illustrated letter. Show the premium you're writing about. Or the reference folder that comes with your subscription. "Pictures sell," Bill Jayme used to say, "newsletter marketers are almost the only people in direct mail who don't know this."

* Teaser copy. Make it heavier and more dramatic than you've used before.

* Price. Test your price only if you are willing to make a dramatic difference--not "Save $40." One test I was involved in dropped the price from $387 to $277, and it produced a clear winner at the lower price.

Elements not to test

* First Class vs. Standard Class. 1C almost never works better. Last year USPS reported that total revenue from 3C exceeded 1C for the first time.

* 9" x 12" and other oversized envelopes.

* Postal indicia. Metered 3C works best for business offers, postage stamps for consumers.

* 4-page vs. 6-page vs. 8-page sales letters. If you can convey your selling message in four pages, expanding to six or eight usually won't do more than recoup the additional printing costs.

* Letter format. 8 1/2" x 11" pages vs. 11" x 17". Some claim two separate sheets will outpull the 11" x 17" but I haven't really seen it.

* Brochures. I won't go as far as Craig Huey who says he's never seen a brochure that didn't depress response. An exception might include a complicated service in which the prospect will get several different benefits and options. The brochure tells, the sales letter sells.

* Premiums. If you have used them, a new one usually doesn't make much difference. Also, a special report that didn't sell very well to your current subscriber base won't be a good premium either.

* Insignificant things like second colors.

Even if the experts are right and "creative" is worth no more than 10 percent to a marketing package, testing various elements of your package should bring another 10 percent to the bottom line--or not. So, get testing, but make sure the elements you tweak are worth testing.