Sifted Talks

August 25, 2021

Six insights from our talk on customer churn

The key to a successful business model is not only gaining customers, but keeping them. How can startups keep today’s demanding customers happy?

Poppy Koronka

5 min read

Gaining customers is key to growing your business — but equally important is being able to keep them or, in startup lingo, stop them ‘churning’.

In this Sifted Talks, we spoke to three experts to find out if startups should be using tech to retain customers and improve their experiences; which mistakes to avoid when attempting to engage their customers; and how businesses can differentiate their approaches to customer engagement at different stages in their growth. 

On the panel, answering these questions and more, were: Sakshi Anand, VP growth at fashion tech startup Otrium; Luis Martinez, global head of customer success at restaurant tech business Flipdish; and Brendan Yell, director at cloud communication platform Twilio. 

Photo credit: Dr Andrew Garthwaite
Photo credit: Dr Andrew Garthwaite

1. Why should you pay attention to churn in the first place?

Many startups, almost by nature, focus on customer growth first and foremost. If you’re starting from nothing, gaining customers is the number one priority. 

So, if you’re starting out, why focus on customer retention at all? Well, according to Anand, high customer retention brings up your lifetime value — which means that you can spend more on customer acquisition. She explains: 

“If we have a high churn, our lifetime value goes down. And if our lifetime value goes down, I can't spend more to acquire customers, because there is a relationship between lifetime value and customer acquisition cost (CAC). And if I can’t acquire more customers, and I'm losing customers, I'm not seeing the growth that I’m hired for. This is why this flywheel needs to work really, really well. It comes together — churn, customer retention and acquisition — to create growth.” –– Sakshi Anand, Otrium

2. Customer research can back up your in-app data

Anand says that every interaction on Otrium is a data point –– but figuring out why people don’t interact at all can also be key to finding out why they leave. 

Backing up these data points with customer research can also help flesh out a high-value customer. For example, finding out why customers don’t select ‘favourite’ brands on Otrium can help Anand turn those customers into high value ones. 

“Every interaction a customer makes on the app is something we learn about the customer. If a customer likes five brands, they are more engaged with us. [It’s important to understand why they aren’t engaging, too.] If customers haven't liked five brands but have visited many times, why is that? Do we not have [their favourite] brands? Working out why customers don’t engage helps us to figure out what leads to a high value action, and therefore what leads to a high value pipeline.”   –– Sakshi Anand, Otrium 

3. Work out your ‘triggers’ for customer churn

There are probably a few warning signs you can look for before a customer leaves, says Rodriguez. 

Flipdish uses a technique called ‘health scoring’, where each customer will be given a score based on how ‘successful’ or ‘unsuccessful’ they are with Flipdish — for example, how many orders they gain through the app. It has warning ‘triggers’ for potential leavers, too such as the customer opening the app less or getting fewer orders than the week before.

This allows managers to speak to those cases before they leave. 

“We build these health scores, which allows us to really quickly and simply see which customers are going to be more sticky. We then start modelling and figure out if a customer loses X amount of orders per week, there's a risk of them leaving. Then we deal with it.” –– Luis Martinez, Flipdish 

4. Know where your customers are in their journey

Alongside understanding what the trigger points for a customer leaving might be, our panelists also spoke about the importance of knowing where the customer is in their journey, saying that it’s a common mistake businesses make. 

A customer engaging with your service or product later than other customers might be a warning sign that they are going to leave, for example. 

“Map out your customer journey. Managing and knowing where your customers are in that journey is important. [Startups] may have mapped out the customer journey, but they don't know where [customers] are in that journey. So, have they read this report? Have they uploaded a task? Have they uploaded a profile photo yet? You can then actually act before they cancel, and hopefully re-engage them.” –– Brendan Yell, Twilio

5. Don’t just focus on why people churn — but why they stay

Collecting data from your customers can give you a broader understanding of what might cause churn. 

Rodriguez from Flipdish says that using the data you collect to understand what is going on with one particular account –– and applying the lessons learnt from success stories to individual customers can help them stay. 

“Really understanding what are the things that make your customer successful can help a great deal. For example, integration –– so a customer that is integrated with other platforms has a lower risk of churning because they went through the whole hassle of getting integrated. Understanding those metrics first can help understanding at the account level. What is the value of your account? How is that account growing over time?” –– Luis Martinez, Flipdish 

6. You can do this all with machine learning — eventually

Automation in business is almost a given in 2021 –– but when it comes to churn, is automation necessary? Is it even possible? 


Yell says that while machine learning can be extremely helpful in understanding and predicting when customers might churn and why, most startups simply don’t have enough data yet. 

It’s important, however, to analyse what data you do have early, so when you can make a model, you know what to base it on. 

“If you've got 100 customers or even 1000 customers, you probably don't have enough data just yet [to use machine learning]. You have to figure out what are the triggers and have a more qualitative way of looking at an individual case. Then you can build models based on that.” –– Brendan Yell, Twilio

To find out more about how to prevent customer churn, you can watch the full Sifted Talk here: