1. By checking click rate, we can't simply say that the advertising works for the sales promotion. The website might recommend the advertisement to people who would like to click the advertisement rather than all of the people, so as to increase the advertisement cost paid by agencies.
4 Steps for Business Analytics
1) Build data infrastructure: Ensure we can get individual level data
2) Data exploration: find the basics that indicate the effect is real.
3)Prediction: customer behavior analysis
4) Optimization
Be individual customer centered, rather than sales centered or product centered.
1. Track every customer at the most granular data level. Some customer is negative valued. (from Store level to Customer Level)
2. Find behavior pattern and use past pattern to predict future pattern. And be forward-looking, focus on the future, rather than the past.
3. Think broadly and develop every potential customer.
50s'- Store level
70s'- Household level : price/quantity, frequency/ timing, product assortment vs purchase behavior
80s'- Modern Marketing: store scanners, membership card to track customer level data. Provide coupon for related products, so as to track the behavior.
90s'-00s': Internet provides the viewing data on website. The combination of internet viewing data and purchase data helps to the behavioral analysis. What drives the sales ? Why people stopped buying ? What are they thinking during the purchase decision moment ?
It's very important to link the online data and offline data by multiple methods, such as sending the coupons to customers, or analyzing the membership card data and so on.
Advertisement Attribution Problem (ROI of Advertisement)
Last Click is not the right way to attribute the advertisement value, but the whole path.
Media Optimization Problem/Channel Cannibalization Problem (How customers utilize multiple media platform)
GRP=Reach*Frequency (Gross Rating Points)
Reach as many customers as possible, and frequency is the volume of the content.
Findings: Channels don't cannibalize each other in the long run but slightly do on a given day.
How valuable are price discounts ?
Consumers bring additional values through their social network
2/3 value of discounts created by direct sending and 1/3 by indirect sending.
网友评论