CRM and Customer Analytics - Appunti

Defection Rate or Churn Rate: Measure how many customers left the company

Retention and defection rate: example

Survival Rate: Measured for cohorts of customers

We take as a reference always the initial number of customers in the starting period

We are interested to know how many customers since period 1 are still our customers in period 3, 4, 5.

Survival rate: example

Attention! In the Survival Rate the denominator is always the same, in the Retention rate the denominator changes all time.

Survival rate decreases over time, while retention rate increases over time, it is a normal thing. (*weird pattern: schema strano)

Customer-Based Value Metrics:

Size of Wallet

Share of Category Requirement

Share of Wallet

Lesson 9: TESTIMONIANZA AMILON

Lesson 10: TESTIMONIANZA GRUPPO POLI

Lesson 11: Customer-Based Value Metrics:

are metrics that are quite interesting, because they allow you to understand what is the value of customers and especially these statistics are also quite useful to understand

brand or firm within its entire base of buyers

Individual Share of wallet iSOW: Can you give an indication on which percentage of your total expenditures on groceries you spent at (Retailer) in the last three months? If I spend €1000 in 3 months and €700 are spent with brand/retailer A, A has 70% of my SOW.

Aggregate Share of Wallet:

SOW vs SCR: For frequent purchases SCR is more employed than SOW. The other way around for less frequent purchase.

Segmenting customers for Share of Wallet and Size of Wallet:

Perception-based measures of the Customer Experience:

CUSTOMER SATISFACTION (scale from 1 to 7): is a survey metric

NET PROMOTER SCORE -NPS- (Reichheld 2003 and Bain & Company)

Issues/Problems with Net Promoter Score?

Not every customer answer

Is a subjective measure

We get score on those who are more likely to respond to the survey, not all customer base

Each customer may have different conceptions of good and bad score (for example 8 can be considered a high)

Suggestions to compute the Net Promoter Score:

PRACTICE QUESTIONS

Question 1

Periodicity: 8,2 (su excel =dev.st.p(16, 36, 24))

04/01/2015-20/01/2015= 16

20/01/2015-25/02/2015= 36

25/02/2015-21/03/2015= 24

Average inter-visit time: 16+36+24/3 = 25,3

Question 2

Totale A: 17

Totale B: 14

Totale C: 11

Totale C1: 18

Totale C2: 18

Totale C3: 6

Ascr: 17/36 = 47%

Ascr: 14/24 = 58%

Ascr: 11/24 = 46%

Solution:

Question 3

Retention rate:

Period 1: 57% (4000/7000 x100)

Period 2: 62.5 % (2500/4000 x 100)

Period 3: 80 % (2000/2500 x 100)

Period 4: 87.5 % (1750/2000 x 100)

Period 5: 90.8 % (1590/1750 x 100)

Defection rate:

Period 1: 42.8 % (7000-4000 = 3000/7000 x 100)

Period 2: 37.5 % (4000-2500 = 1500/4000 x 100)

Period 3: 20%

Period 4: 12.5%

Period 5: 9.1 %

Survival rate:

Period 1: 57% (4000/7000 x100)

Period 2: 35.7% (2500/7000 x 100)

Period 3: 28.57% (2000/7000 x 100)

Period 4: 25%

Period 5: 22.7%

Question 4

Share of Wallet and Size of Wallet

Segment 1:

We have to maintain these customers and go on because they have a large SOW

Segment 1:

They have a large share of wallet (SOW) compared to other segments, but already have a very high SOW. The point is to maintain these customers and ensure that they stay with our company.

Segment 2:

They have a quite good share of wallet, but the problem is that their size of wallet is low. We need to work on making them more loyal to our company.

Segment 3:

They have the potential to spend money with us, but they are not capable of taking advantage of the money they have. Our competitors may be better than us in this aspect. We should create incentives and offers to encourage them to spend more money with us.

Segment 4:

This segment has a very small wallet. The retailer does not invest much time in pushing this segment to buy from us because the efforts will not be worth it.

Share of Wallet and Share of Category Requirement:

Share of Wallet (SOW) is computed in value, while Share of Category Requirement (SCR) is computed in volume. SCR is more commonly used for frequent purchases.

Purchases while SOW is more employed for non-frequent purchases

Question 5

NPS = Promoters – Detractors

169 promoters - 486 detractors = -317

169/10 consumatori = 16.9% promoters

486/10 consumatori = 48.6 % detractors

NPS = 16.9% - 48.6% = -31.7% = 32 %

Is important to know the benchmark of competitors companies, for example in this case is important to know that the net promoter score is the same in the other competitor’s companies.

Strategic Customer-Based Value Metrics: RFM METHOD

- RECENCY

- FREQUENCY

- MONETARY

RFM CLUSTERING - SEGMENTATION:

RFM METHOD:

importance of variables

The importance of R F M might vary depending on the industry of reference. It is given by the rapidity of customer response rate drops.

Lesson 12:

RFM METHOD: different methodologies

- Standard sorting

- Advanced sorting

- Employing pre-identified cut-offs

- Employing Cluster Analysis

(k-means)

- STANDARD SORTING

We have 40000 customers divided in 5 equal parts, we start sorting (*iniziamo

1. For recency:
• 5 means that the segment is the best segment in terms of recency
• 1 means that this segment is the worst segment in terms of recency
2. For frequency:
• 5 means that the customer segment at the top is very frequent
• 1 means that the customer segment at the bottom is less frequent
3. For monetary:
• 5 means that at the top customers spent a large amount of money
• 1 means that at the bottom customers spent a low amount of money

It is possible to have a customer with a score of 511 or 513, etc. There are a lot of different combinations and situations, it is not always the same.

Customer is not recency 1 - Customer is very recent 2

Low frequency 1 - High frequency 2 / Low monetary 1 - High frequency 2

ADVANCED SORTING - RFM METHOD: new

Options in the online environment

Lesson 13: Examples of RFM segmentations:

• EXAMPLE 1
• EXAMPLE 2
• EXAMPLE 3

First Purchase = It's the LOR, how much time elapsed since the first purchase of the customer

Which is the best Customer here?

Cluster 5, highest monetary and frequency, is the good news for the company the low recency, means that customers are active recently

Cluster 1 and Cluster 2, which cluster is better for reactivation?

Cluster 2, is a more suitable target to reactivation

Which are the external variables that enrich the profile of clusters?

Age, subscription to the newsletters, downloaded the app, inter-visit-time

CRISP-DM

Cross-Industry means that the process is valid for any industry/sector

Data mining means cluster analysis

Why there should be a standard process?

Framework for recording the experience

• Allows projects to be replicated
• Aid to project planning and management
• "Comfort factor" for new adopters
• Reduces dependency on

“stars” Initiative launched in late 1996

Over 300 organizations contributed to the process model

Published CRISP-DM 1.0 (1999)

1. BUSINESS UNDERSTANDING
   • Studying the business and identify business goals
   • Identifying key stakeholders
   • Finalizing clear data mining goals (*)
   • Develop a project plan (*)
2. DATA UNDERSTANDING
   • Data collection and description
   • Checking technical issues
   • Identifying potential outliers
   • How to classify databases: the type of information included
     • a) CUSTOMER DATABASES, is a data bases that company has inside
     • b) PROSPECT DATABASES
     • c) CLUSTER DATABASES AND ENHANCEMENT DATABASES, information to beenrich your information data bases
3. DATA PREPARATION
   • Data selection and cleaning
   • Missing values imputation
   • Variable creation
   • Integration of datasets and formatting
4. MODELING
   • Select modelling technique
   • Generate test design (you might want to work with a small sample first)

• Modeling Estimation:
  • Splitting data into training sample, validation sample (and test sample)
  • Try different model formulations
  • It is an iterative process
• EVALUATION
• DEPLOYMENT

Main digital touchpoints:

• Email marketing
• Websites
• Social Media
• Mobile APPS
• Online ads: Search Ads, Display Ads, Social adv
• Organic Search results

In digital marketing you work with different goals and different objectives, you might want expose your user to your brand, to your message and to your content. You might want to persuade your user to visit your website, click on your email to engagement with your content.

Conversion: employs an action from your users that has to do closer to a purchase (*avvicinarsi ad un acquisto) or a purchase indeed.

Advocacy: promoting your brand or your products with other people

WHAT IS A CONVERSION?

Macroconversions: purchases

Micro conversions: downloaded a coupon or mobile app, subscribe a newsletter

DIGITAL ANALYTICS: By using digital