Data > Opinion

[sandeshgour-blog]

All About Churn Users?

In this blog post we will discuss about churn rate - what is it, how it's calculated, good churn (yes you read that right!) and ways to get churned users back.

Terminology :   User : a person who is just browsing your site/shop/blog Customer : A person who actually performed a conversion i.e. bought/subscribed/signedup

What is churn ? Churned Users are the users who are no longer associated with you, in other words the users who have stopped using your product.

Why should we worry about churn? Churn rate provides insights on whether you are performing as per the expectations of your users or not. Also, majority of spends in a company goes into Customer Acquisition (CAC). Churned Customers are those who already bought from you once and hence are high intent users i.e. they need the product that is why they bought it in first place. Also, you have all the required data to reach out to them (their personal information like email id, phone number etc.). They are just one text/email away, which are more cost-effective marketing channels to acquire customers when compared to other channels like SEM and offline marketing.

How to calculate churn? In order to calculate churn rate, firstly we need the number of churned customers. For a subscription-based company, the calculation is pretty straightforward - the customers who end their subscription or do not renew their subscriptions in the future are churned. But for a product based company, a user/customer does not have any time-bound window for making a purchase, i.e. a purchase can happen at any point of time. In such cases, you need to look into the repeat purchase behavior of the customers, take a 95% confidence interval and decide a time period post which you will consider those customers as churned.

Basic Formula for churn rate : Churn Rate = Total Customers Churned / Total Customers at the beginning of the obervation period Here, Total Churned Customers is the number of customers who were churned during the observation period. For eg, A Subscription based company is looking to calculate Jan month's Churn Rate. Total number of Customers it has on 1st Jan is 1000. Out of these 1000 customers, 50 get churned by the end of the month. During Jan, company acquires 100 new customers, but out of these 100, 10 also churned by month end. So total number of churned customers = 50+10 = 60 Table:

According to the basic definition, Churn Rate will be = 60/1000 = 6.0 %

There is an asterisk attached to the above definition as it does not take into account many crucial factors. For example, churn rate includes churned customers out of newly acquired customers but does not consider number of new customers in total customers calculation.

A better way to calculate Churn Rate is described in this recurly blog(https://blog.recurly.com/2014/08/better-way-to-calculate-your-churn-rate) written by Devin. According to him, we must calculate Churn Rate on Per Customer Day basis. In the above example, 1000 customers at the start of month, each has 31 days, and we will assume that each new customer acquired will have 0.5 of it. So total number of customer days  = 1000 * 31+100 * 31 * 0.5 = 32550 Total Churned Customers = 50+10 = 60 Churn per customer day = 60/32550 = 18.43% Churn Rate Jan = Churn per customer day * 31 = 57.14 %

Even the above solution has an asterisk attached to it, as it assumes that all the new customers are acquired at a constant rate, which is why we use 0.5 for the calculation of customer days. But that's not true in reality, some newly acquired customers get more time to churn (users who acquired in first half of the month) when compared to others (users acquired on second half of the month). So, what is the best way to visualize churn ? The answer is Cohorts.Create rolling day window cohorts for each month! How to create and read Cohorts is topic for a whole new post, will elaborate it in upcoming posts.

At the end, whichever definition you choose, make sure you keep all your stakeholders informed. Make that definition standard across organisation to avoid confusion. Different departments may end up with different churn rates for the same month, if the definitions are different.

Good Churn? One of the purpose of this post is to enlighten you about the fact that not all churn is bad, some of it is actually good churn. Wait, how can losing my customers be good? Well losing the bad ones is! This brings us to the question of who are "bad customers", rather the "Not-So-Good Customers". Every customer is different, you cannot satisfy everyone's need because every person's requirement is different. You must be aware of your product's vision and what your target segment is, and act upon their suggestions, not of every other user or customer. The customer is surely not always right - and your product may deviate from its core vision if you pay heed to a small group.

How to get'em back? Try to find out why did your customers churn. There can be multiple reasons - like the product was not what they expected, or probably your website/app is not user friendly, delivery was delayed, need for inclusion of some more features, may be they are "Not So Good Customers" etc. Any Web Analytics Tools and Customer Surveys are the best ways to get the answer to this! Once you know why your customers churned, you can run reactivation campaigns to get them back - send out emails/sms, lure them via coupon codes, share upcoming feature releases and much more.

Again, this is not a one-time activity. Monitor this on a regular basis and DO NOT STOP LOOKING!

[sandeshgour-blog]

5 things to keep in mind when you start looking into data

Since, you are reading this blog you are already motivated to look into data. Below are the 5 basic things that you should keep in your mind before diving in:

1. Is the data right?

First and foremost thing is to check whether the data is correct?

Sometimes when we find an awesome insight from the data provided to us, we get so excited that we started dreaming that this is the insight for which I will get a freaking raise. But after a while you will find out that okay, it was all because one of your analytics tag was firing multiple times on a page where it was not even supposed to. All the effort, all the time spent digging out that insight, wasted.

So check the sanctity of the data. If it is provided by someone else, ask for the source and assumptions they made while fetching it.

2. Every metric should have a counter metric:

Here counter does not mean mathematical counter but it is used in the context of "opposite".

An individual metric can give you completely wrong picture, you should always take at least one counter metric while analysing. For e.g. Terry has a website. Terry hired an agency to operate his digital media ads. Website's visitor number increased by 20% after that. Terry was happy and ready to pump in more bucks to that agency. But Wait!! Bounce rate of the website is now 50% which was used to be well below 40 %. This means although visitors are increased but those are not your customers, for all we know they just unintentionally clicked on your popup ad.

3. Look at the big picture: 

Every metric must converge for revenue, may not today but connected

At the end, what everyone want is a Tesla, and for that you need $$$$$$$. So make sure everything, every metric improvement is another step towards profitability.

As mentioned in above example, if Terry was also looking into revenue number metric, he will definitely had found that although those ad campaigns is increasing their daily visitors but these additional visitors are not contributing much in Revenue.

4. Keep Track of metrics definition and choose wisely:

The definition of even basic metrics like customers, registration etc differs from business to business and it should be as different businesses has different requirements, different visions. The basic step people miss is to keep a log of that definition.

Which one of the following definition is correct?

1. Net Transactions = Total Transactions

2. Net Transactions = Total Transactions - Cancelled Transactions

3. Net Transactions = Total Transactions - Refund Transactions

4. Net Transactions = Total Delivered Transactions

All the above definition are valid, it completely depends on your business. And with the free flow of employees coming and going, it is entirely possible if someone misinterpret what is what. So Keep A Log of Metric Definitions.

5. Do Not Stop:

Most of the time we analyze data, create an awesome report, get useful insights , present that report to the board, get applause and move on. There is no point of all this cumbersome process of mining and analyzing data if you don’t keep track of those metrics in future. It should never be onetime task. You should keep monitoring metrics. Automate stuff, reports, alerts etc. There are many tools available in the market or make one for yourself according to your requirement. But DO NOT STOP looking into data.

[sandeshgour-blog]

Why should I implement Analytics : Developer’s Dilemma

In today's web centric world, God has given humans ability to track almost everything that is going on website with the help of various tools.

But blinded by huge amount of data and urge to get quick insights, we generally ignored the most crucial part of the process-"Implementation".

The biggest assumption we take while looking at the data is that "it is correct", is it ?

Correct implementation is depends on both analytics owner and the person writing implementation code.

Former knows how important it is to get implementation correctly, later however is more concerned about the site not analytics.

While making a change on site or adding a new feature, analytics implementation is least priority things when it comes to deadline of the project.

As an Analytics owner you have been given responsibility for analytics tool implementation, it is your duty to enlighten the tech person about the power of the data.

How? Give them insights. People love power.

Whatever change the team has done in past few days, show them the impact of their change on overall conversion or revenue or orders.

We made a simple change in final payment page by decreasing the size of the "Apply Coupon". Conversion increased. Reason: Less customers clicking on apply coupon, less customers exiting the site in search for company coupon on google, more conversion.

 Red Rule: Never give your customer opportunity to exit from checkout flow.

This impact analysis was not possible without proper implementation by your IT team.

Show them the metrics they changed, show them the power of their code.

Now tell them they will never be able to get the direction and amount of impact without proper implantation.

Implementation is like seed, if you don’t plant it right you may get poisonous fruits.

[sandeshgour-blog]

Million Dollar Idea

Million Dollar Idea

What does a million-dollar idea mean? Does the idea itself can earn you million dollars? How do we calculate worth of an idea? 

Well an idea by itself is worth NOTHING, not until its execution. Once the idea starts generating revenues one can state its monetary value. So the price tag of any idea is decided only after it has been executed. Fixing price of any idea is just like “luck”, good or bad depends on the consequences of your actions. Before the end results we cannot say whether your actions were lucky or unlucky, it is only after the final outcome one can reach any conclusion. If the outcome were in your favor then you were lucky if not, Unlucky. 

Let’s explain it in other way. A farmer gave his boy a seed and told him that seed is worth hundreds of dollars. The boy kept the seed hoping to utilize it when he need the money. One day he went to the market, to sell the seed and for his surprise no one is ready to buy that seed for even a dollar. He felt tricked by his own father. He went back to his father and ask why did you lie? His father explained him that it was not the seed that worth dollars but fruits, if the seed is planted and nurtured. So idea is worth nothing. But there is more to it. We can establish a value of an idea depending on its revenue generation, but what about failed ideas. There are certainly more failed ideas than successful ones. What do they worth? Do they worth nothing? What if the seed is unfruitful? You may not gain any money from them but you do gain more precious thing than that, experience. The boy did not get any money from the seed but he does learn the lesson.

Every idea is worth nothing and every idea is worth something.