Market Statistical Analysis (Campaign effectiveness)
It is statistical analysis tool for measure of campaign effectiveness. It will measure the significance level of the campaign impact. Normally most of the companies will use KPIs or Metrics to measure or track the campaign performance. Another method is statistical analysis of the campaign. It will give more realistic results. Statistical figures will give us assurance of quality of the result. As in data world accuracy and relevancy of the result key source of decision making. Now let’s move towards how to use this method. We can apply this method at two places. Before starting of the campaign or after the campaign completion.
- Control and Test group
From above picture we can find both groups are identical. It means we are doing comparison of the same group like apple to apple. But the difference between both is one is under controlled group and another one is under experiment. It’s very popular method to find the difference between control and experiment.
Suppose in a College, PGDM class two sections are there. In PGDM batch for marketing subject, one section got special classes from industry faculties. Another section only got normal classes from inside college faculties. While placement Interviews college management measuring the impact of special and normal classes. College expect from the special classes is student will have more exposure towards industry. Which will helps them during interviews like market trends, industry situation, scope etc.
On the other hand other section will don’t have this tool or knowledge. Now they will measure the impact by seeing the results of their interviews. It means who will crack more interviews special and normal class students.
Features of the control and test group:-
· Control group: — In this group everything will be constant. It means control group will not get any impact of the experiment or campaign. Company will not target this group.
· Test group: — This group will be the experiment group. On which they will apply the new product or in campaign. They will be the target point. They will get all poster, banners, advertisement, e-mails etc.
· We can only use this method when we have data or information regarding target and control group customers.
· It will measure the impact between both groups.
· Like above we seen the example of special (test) and normal (control).College management is trying to check. Is special classes is useful in cracking interviews?
2. Time Dimensions
It will useful when we don’t have data or information available regarding control and test group. In this we will do the comparison sales for different time. This technique method basically use to measure the impact of seasonal marketing campaigns. But this technique have disadvantage. Suppose except seasonal trending month. If we are measuring rest of the sales. In the market there is lots external which will impact the sales. Like natural, economic, technology up gradation etc.
When we are using time dimension to measure campaign effectiveness. Suppose during demonetization in Nov 2016, suddenly sale of the 4 G cellphone increased due to banned of 500 and 1000 notes. During that period Reliance introduced 4 G sim in the market. That time situation of the market was very dynamic. Economic change, technology change which had increased the sales of the mobile phones. So next year we can’t expect same volume of sales in that month. So we are doing comparison sales with time dimensions. Market situation will be always dynamic. So the time factor can be influenced by external factors.
Post Campaign Analysis
1. Control and Test group
In this analysis we don’t have data regarding customer target or control. So through customer behavior we have to find the pattern of purchase. Whether if there is increase in sales is normal. After campaign it’s increasing more as compare to earlier sales. If it’s increasing we have to find whether they are new customers or old customers. We can categories new customers as targeted and old customer as control group. Because campaign will only impact the new customers, old customers will be constant.
It is a method which used to check probability of respond from the customer. It’s statistical modeling tool which use to predict the response of the customer after the campaign. Either customer will buy or not buy. It will use logistics regression model to find out the probability of happening of the event.
The process of this method is randomly select a group of customers from normal population. In the same way randomly select the target group. After that you can start your experiment. In this process probability of selection control group and test group is equal. Which will remove chances of bias. So the result will be validated. After completion of the experiment we can find out the difference between control and test group.
Function of the Propensity model
This model will use the algorithms of machine learning to build a model. Which will be helpful in predictions of the result. It will be helpful on any population testing. For building this model they need to extract the features or pattern of the customer behavior. Which will be helpful in output labeling.
For finding probability score we need full-fledged data about the customer purchase pattern or behavior. On that basis we can do feature extraction and output labeling.