Designing for Variable Data

WHY BOTHER with variable data? A lot of sales people hit their customers with lines like: "It always gets a better response than static mail pieces."

Red Alert! Red Alert! Failure is imminent!

Too many printers and designers think that variable data is automatic and a simple process that is going to solve their economic problems. If you do not develop a strong analytical and practical deductive reasoning process for implementing variable data campaigns, you will not be helping your economic problems; you will be intensifying them. The percentage of variable data campaigns that show 0 percent gain or even 0 percent response rates is larger than most people think.

The great variable data campaigns did not typically happen on the first attempt. They evolved and continue to improve as the printer/designer and print customer better understand the end customer's needs and response patterns. The good printers/designers also learn from one campaign how to get to the best solution a bit faster. The key to success is test, test, test. Find out what works and what does not.

Design with Results in Mind

If you are designing a piece that will use variable data in it—regardless of whether it is a printed piece, an e-mail, a PURL (personalized universal resource locator) or a text message — you need to design for the results (i.e. increased sales, reduced acquisition costs, higher levels of customer retention, improved customer loyalty, improved customer satisfaction). So when you design for variable data, make sure that anything you change produces some type of result. There is only one way to know that this is really happening and that is why you should also design to track and measure the results (we are back to test, test, test).

Many people base the success of their results on a mythical response rate of 2%. Somehow, it has gotten out that 2% is the average response rate for a typical mail piece, and anything higher than that is considered a success. The Direct Marketing 2010 Statistical Factbook is out, and the new average response rate for cold calling (that is sending a mail piece to someone who is not a customer currently) is now 1.9%—but the average response rate for a mailing to existing customers is 16.8%.

The important thing to remember here is that these are averages and if you properly set up a way to measure the results you can know what the response rate would be with or without variable data (the details of that will have to be covered in another article). Response rates are impacted by many customer specific characteristics: the past relationship with the customer, the value proposition for the customer, the offer and call-to-action, the culture variables in the customer base (age, gender, race, political party, pizza topping preferences, parents with kids at home, etc.) and a host of other characteristics.

By now you may have realized that this is a follow-up article to "Preparing for Successful VDP," which appeared in the January issue of IPG. In that article there was a discussion on how important the data is when planning a variable data project; here, the focus is on designing a piece that uses the data in a meaningful way. It is important to recognize that clean, accurate data is the foundation for building a successful campaign.

Obvious or Subtle?

One of the big questions that comes up is should the variable text and images be obvious or subtle? That depends on the objective of the campaign.

If a printer wants to make a statement to a potential customer on the targeting nature of variable data, the plan might be for each member of the sales force to take a picture of the front entrance of their top 25 customers' buildings. When the brochure is printed for each unique customer and they take it out of the envelope, they are seeing the picture of their building, not the printer's. When you are trying to sell the idea, you want them to see every point where you are personalizing the information. But more typically when you are creating a campaign for the customer's customer you likely want the variable nature to be subtle.

One of the best examples of a successful variable data print campaign was a piece designed for a furniture store. The sales force saw a pattern in the way couples walked around the showroom floor each Saturday. The couples would come in and start walking around separately, but by the end of their bedroom or living room search they would get together and focus in on two or three different options. The sales person would make note of these pieces and after getting their contact information would e-mail the printer with the data and furniture options. Images of everything on display were already stored in a digital asset file (created for the furniture catalog books). So all the printer had to do was drop the correct picture into the pre-planned brochure layout.

The store sent the brochure with a personalized thank you letter and outlined six of the top-selling furniture groupings. When the couple received the brochure, they were impressed with their good taste, since two of their choices were in the top selling groups. The brochure did not say "we know you were looking at these two groups." It made them feel like it was a general brochure and it just happened to have their two top choices as show pieces. All the variable text was in the personalized thank you letter.

The most important point to realize was that the furniture company sold twice the amount of furniture each week they sent out the brochures. When they skipped a week, sales dropped. The printer was charging 400% profit on this job, and the customer did not care because they were selling more furniture as a result. The brochure design was nice, the photos were professionally done, but it was the right selection of the images that was helping them sell more furniture.

Study Your Target Audience

A sample rule map for selecting a picture for a proposed campaign for an automobile motor club promotional mailing allowed research to show that only about 50% of the tows done by a typical towing company were for people that are members of an automotive motor club. So the other half was targeted for club membership, using the logic that they would more likely be open to a membership offer if they knew it could have reduced their towing cost by one third.

There were three demographic characteristics being used to personalize the images for this campaign—gender (two brackets), age (five brackets) and race (four brackets). There was also personalization of the text, but the rule map only shows the plan for the photos. If you look at the final results you'll see it requires 40 images to target the audience. This means, for example, a white female college student would get a brochure with a picture of a white female in her twenties sitting on the side of the road with a car broken down and waiting for a tow.

The variables are targeted on the distance that they had been recently towed and the approximate cost. You would not want to target the price exactly; the goal would be to create what appeared to be a coincidence based on their recent towing experience. The customer will think how uncanny this mailing is after they just had a car towed a few weeks earlier.

One more important thing that went into the design of this variable data piece is a careful study of why people belong to motor clubs. Typically they fit into one of a few categories: they want to save money; they want a reputable company when they are stuck at a strange location; or they want a local towing company that can respond quickly. The motor club needs to know why people belong. Knowing more about their customers helps them to better target more of the same people.

Use Emotional Imagery

So designing for variable data campaigns is not just about a pretty picture or a colorful layout (although they are still important). It is about targeting the image and variable text to elicit a response out of the customer. People react to emotional imagery, financial factors and personal preferences. The more you know the better you can target the message.

Another key element found in high-level variable data is the use of personal information (this would be when you want to make it obvious) in the opening paragraph. If a grocery store was sending you a mailing and has been tracking customers for the last 15 years with their VIP cards for weekly specials, they know what you like and what types of sales you typically take advantage of each week. When they are trying to send out a loyalty campaign, something as simple as saying how many years you have been a VIP customer makes a statement that you are seen as unique.

A good rule of thumb is to try to bring in at least two or three pieces of variable text into the first sentence or two (name and address do not count here). It is like when you are sitting at an awards dinner, and they are getting ready to announce the top award for the year. If they say "our winner tonight has been with our company for more than 20 years," they have narrowed the field. Then they say "he served in the military," further narrowing it down. And if they finally say "he has eight grand children," then you know it can only be one or two possible people. It only takes two or three pieces of personal information for the customer to think someone has taken the time to write a single letter directly to him or her.

So why bother? Because if you use your creative talents to personalize the images and text in a variable data campaign you will get a lift in response rates, in sales and in customer loyalty for your customers. If you do it right, the customer will see the value you bring to the table. It is not just the photos or the background or the data—it is the blend of data that makes it all work. You control the end results. No it is not easy. If it was, it would be a commodity, and the markup would be marginal.