Ch1

맨 처음 페이지

Churn-reducing strategy	Core concepts/customer metrics	Book chapters
1. Product improvement - Make more of the best features. - Makes the best features easy to find	- Cohorts of metrics based on product use events identify engaging and disengaging product features. - Identify optimal feature-use ratios.	3, 5, and 7
2. Engagement marketing - Promote the best features. - Targeted product insights.	- Metric cohorts provide benchmarks for healthy levels of product use. - Segment customers with metrics.	5 and 7
3. Pricing and packaging - Differentiate pricing to provide value without discounting - Monetize valued groups of product features.	- Unit cost and unit value metrics identify customers getting high/low value on the product. - Correlations show relationships between the use of different features/content.	6 and 7
4. Customer success and support - Help customer in need. - Identify falling customer proactively. - Provide training at key points in customer journey.	- Metric cohorts benchmark healthy use levels. - Forecast customer risk with regresssion and machine learning. - Metric-driven customer conversations. - Account tenure cohort analysis measures risk along the customer journey.	5, 8 and 9
5. Channel Tartgeting - Identify your best customer channels. - Find lookalikes.	- Category cohort churn analysis with confidence intervals. - Identify the best/worst sales channels and demographic/firmographic indicators of success.	10

 

p.xv

When it comes to running a subscription business, churn rates are a matter of life and death!

 

 

p.14

An important way in which real-world data problems differ from training is that in the real world, the job never ends: as soon as one analysis of churn is complete, new product features or content are created, requiring reanalysis. Or an entirely new type of data to enhance the original may become available. Also, there are constant changes in the business environment, such as competition and changing economic conditions. Such changes can require reanalysis, even if the product isn’t modified.

 

p.24

But note that when we talk about cancellations as events, we are talking about the date and time the cancellation change is entered into the system, not necessarily the time the service contract ends when the user has time remaining. For this reason, a cancellation event is not the same as a churn. The churn occurs when the subscriber completes the current term without sign- ing up for a new one, often allowing a short grace period. Consequently, a cancellation event does not necessarily mean a churn will occur, because the customer still might re-sign before much time has passed.

 

p.26

Look for events that are as close as possible to the value created by using the service, even when that value cannot be measured directly.

 

p.30

As I mentioned in the discussion of events, the behaviors that are most closely related to the value delivered by the service are most important. But choosing the measurement to make is also crucial. Here are three metrics that I have found to be especially effective in the fight against churn: 

-  Utilization—Metrics that show how much of the service the customer uses. If the service imposes limits on some types of use, a utilization metric shows what percentage of the allowed amount the customer took advantage of.

-  Success—Metrics that show how successful a user is in activities that have different outcomes.

- Unit cost—Metrics that relate to the price the customer pays for the quantity of the service consumed or used.

 

p.31

While measuring the number of active users is a good metric for fighting churn, an even better one is shown in figure 1.3. This is the license utilization metric calculated by dividing the number of active users by the number of seats the user has purchased. Many SaaS products are sold “by the seat,” meaning the number of users allowed (this is called the licensed number of seats).

 

p.34~35

So why does the relationship to churn show that more detractors are good when you look at the detractor count in figure 1.5, and that more detractors are bad when you look at the detractor rate in figure 1.6? 

The answer is that the total number of detractors in figure 1.5 is related to the total number of promoters shown in figure 1.4 because Broadly customers who receive a lot of reviews overall are likely to receive more of both good and bad reviews. 

 

When you look at the impact on the relationship between the number of detractors and churn in the simple way in figure 1.5, it conflates two underlying factors driving the metric: having a lot of reviews (which is good) and having a high proportion of bad reviews (which is bad). 

 

When the proportion of bad reviews is analyzed alone, you get the more useful result shown in figure 1.6. This illustrates why success and failure rates can be so effective for understanding churn.

 

p.35

When trying to understand churn, it is important to consider not only the amount of service that customers use but also how much they pay.