The Illogical Song: What Makes Sense vs. What Works

It's official: J.C. Penney scrapped its every-day-low-price strategy for the rollercoaster world of markups and markdowns. Merchants, marketers and graphic artists waited with bated breath, hoping J.C. Penney could make this strategy work. Imagine the time it would save! Price changes and errors would be a thing of the past. What's the price? Same as yesterday. FTP the files to the printer. Have a good evening.

If you ask consumers what they prefer, many would claim a consistent low price. Who has time for tacit haggling? It only makes sense. But does this strategy work? Is that what customers really want?

As direct marketers, you tend to be logical and math oriented. Ironically, a key to your success is accepting the disconnect between what makes sense vs. what works. You have to ascertain the difference between what customers say they want vs. what they respond to.

It turns out that customers don't just want a deal, they want a special deal. Limited time, limited quantity, limited attention span. If it doesn't seem special, your customers aren't interested. Worse yet, they forget about you.

I recently showed a consultant friend some sales results where giving away a 59 cent tchotchke produced a 20 percent lift in sales. He was "astounded that this worked," which of course implied he thought me to be an idiot to test it in the first place. (Fortunately we're friends.) The offer wasn't logical. When it comes to offers, you need to jettison logic from time to time and stop overthinking them.

Here are some objections that need to be abandoned:

• We don't want to attract customers who are only looking for a deal. Wal-Mart makes a fortune from customers only looking for a deal. Sure, you as a merchant need to make money on the proposition, but if customers believe they got a deal, and the return on investment works for you, isn't that good?
• We don't want to attract customers who will buy only once. I'm a database marketer and have long stressed the importance of building your customer file with quality names, however, even with the best of efforts, about 50 percent of your customers buy from you once and never return. If you get an acceptable ROI on the first sale, you're good. Since databases allow you to segment at the purchase level, test whether these customers are loyal or fair-weather friends. You can know whether these customers will be repeats to whom you can remarket or if they're one-and-done.
• We don't want to encourage our customers to wait for a deal. This objection misses the point of an offer. An offer encourages a percentage of consumers who wouldn't have otherwise made a purchase to buy from you. This means that most of your customers would have purchased without the offer. The key is to structure the offer so that the lift in sales covers the additional expense of the offer, leaving extra profit.
• We don't want to degrade our brand. People typically point to Apple in this regard and say that it seldom if ever makes special offers on its core products. Former Apple executive Ron Johnson applied this strategy to J.C. Penney with catastrophic results. If you can make a no-offer-ever strategy work, you have a product or service that's either unique or of exceptional quality. On the other hand, if a customer can find a similar product of similar value at any number of competitors, you'll have to use offers.
• We don't know if the offer will work. Of course you don't know. That's why you're testing it. Let the customer decide whether it works. With that said, you can determine whether it has a chance by running a simple ROI test. A good rule of thumb is to start with a 10 percent to 15 percent lift. If this range of lift in sales doesn't cover the added expense of the offer, you may want to rethink your promotion. On the other hand, if you only need a 5 percent lift in sales to make it work, you have a lot less to lose.

Once you've decided to test your offer, apply the 10-50-100 rule. Test with 10,000 names in your database. If that works, up the ante to 50,000 names. If it still works, roll out to 100,000 names. Of course, your quantities may vary to get a valid statistical read, but the principle is the same. Test, retest bigger then roll out.