To obtain meaningful results, you have to commit to a meaningful volume. A few thousand pieces here or there will lead to upsetting results and might prematurely end your efforts in using direct mail as a acquisition channel.
They don’t work and don’t let cost lure you in. Always use a credit bureau and actual credit attributes.
the coverage between the 3 prime bureaus are not the same. Even if the same prospect can be found across all bureaus, the depth of their credit profile are quite different.
If you are fortunate to have a team of data scientists, building a response model is not enough. You need to consider building a risk model. The most desperate applicants are often time the most risky.
Always use firm offer of credit and not ITA (invitation to apply). With a firm offer you can use languages that are direct and draw people’s attention.
For a variety of reasons, you might have opted out A/B testing. It’s not too late. Subtle differences in wording or color variances could lead to a significant increase in response rate.
If you haven’t build any landing pages, it could be hurting your pull through rate. To establish trust with your applicants, try aligning your direct mail creatives with the look and feel of your landing page. Also, have a look at your ads in google or bing, the language in your ad copy should also be as consistent as your CTA (Call to Action) on the mail piece.
If you haven’t mailed the population that you’ve mailed before, you should. Try a different offer or a different creative. The amount of responders will be no less than the initial mailing. Remember, you never know when your customers will need your product or service.
Your product offering may or may not work during a certain time of the year. Consider pulling back during tax return season and pump up the volume leading up to the holidays if you are offering a credit product.
Different shops have different core competencies. Some shops are great at fulfillment and shipping, some are great at printing. You might see a pricing difference, and a performance gain.
Bonus:Always test outside of your boundaries. Reserve 5% of your volume for these out of bound testing. Otherwise data you use to calibrate your program or subsequent models will become a self-fulfilling prophecy and die a slow death.