Ask any Statistics/Probability/Math Question
Like any other statistical modeling, Response modeling involves the axiom "garbage in, garbage out". The goodness of the model depends upon the data that is being used for building it. The definition of responder becomes incorrect at the time when the data is wrong. Also, we will obtain disappointing response rates when the model used is not updated, or is bland and uninspiring in its creativity. Compared to credit risk models, response models are more dynamic. There is a negative effects of direct marketing on some customers, where the level of spend of the customers is reduced.
When a person in a stable situation (unstable when ill or is jobless) has to pay the obligations in credit risk it is fairly stable over the time. It is likely and favorable that people who have never been felonious on the loans will continue to pay the bills on time. A lot of things are there to see whether a person is responding to mails, and a few of it can be considered in a model. Response rates can be highly affected with a slight shift in the demographics of an area. Potential customers can be turned off with the over use of catch-phrase. All these reasons lead to re-building marketing models rather than the credit risk counterparts.
From a wide variety of people involved in a business input is required for response models. For a successful campaign the marketing manager, the mail shop, the statistician, the financial group, the artist in charge, are all required. Most importantly, the product is worthwhile for the customers. As there are so many people involved a bad decision or any compromise can result in a failure before the first mailing .In a response mailing coordination of people and production can be huge.
The Goodness of Conventional Predictive Analytics
The similarities between responders in a direct marketing campaign can be identified by response modeling. And also statistical models are created so that it can be applied to the new data for determining whether the prospects will respond or not.
The following things can be done by response modeling in the business:
- Marketing costs is decreased.
- New customers can be acquired more cost effectively.
- The sales and response rates can be increased by 20% to 80%.
- 30% of your customers that account for 70% of your profit can be identified.
- Identifying those who are most likely to respond.
- The profitability of your customers can be maximized.
- A better understanding of prospects and customers can be gained.
- It can be learned which of your customers are being undersold.
- Market segments which are the best can be discovered.
- The most profitable prospects should be targeted.
At times when marketing campaigns avoid purchasers a missed opportunity is created.
If the wrong customers are predicted to be the ones who are likely to respond, a wrong decision of contacting them takes place. So all these customers should be sorted out from the contact list. Even if the response rate of the campaign may be high, the overall positive impact of it is low. When a campaign pushes the customers away more damage and less good happens. The danger of wastage of most dollars that are spent occurs in each campaign that considers and targets response model. Thus one need to predict correctly those customers who are eager to buy only if contacted.