Originally posted on LinkedIn Pulse.
Earlier in my career, I had the opportunity to work in the Customers for Life team at Salesforce. We had a singular mission: plug the leaky revenue bucket and improve customer retention. This role was during the aftermath of the great recession and, while I'm not able to share the confidential metrics, I can say that an elevated churn rate at that time was a clear and present danger to the business growth of Salesforce.
So, how did we stem the tide? We decided to take a thorough approach to detect, analyze, and prevent churn. Before we worked on prevention, we had to figure out the actions that Salesforce customers take that lead to higher customer retention rates.
In this post, I’ll share ideas that I’ve collected over 15 years working for high-growth B2B SaaS companies (Salesforce, Host Analytics, QuanticMind, and Tray.io). The good news is that advances in technology are making getting ahead of customer churn challenges easier than ever. There is no one size fits all answer. I'll offer a mindset and approach that you can tailor to any business.
Create a churn risk detection system
Back in 2008 at Salesforce, we first decided to cast a wide net to understand the signals that indicate probable churn. We aggregated data such as customer demographics (# employees, industry, titles), products and services purchased, products and services usage, support tickets submitted and a myriad of other data. For product usage, we evaluated metrics like last login date, the frequency of logins, and how many active users there were within a customer account. The data problem was a lot harder than we initially realized because it was spread across many systems (CRM, product database tables, and learning management systems to name a few). It took us more than six months to just assemble the data. Most organizations don’t have that long.
What followed was the realization that SaaS businesses need to manage customer churn proactively. In turn, that led to a new category of Customer Success Software.
Years later, I had the privilege of working with one of the pioneers in this space: founder of Gainsight, Jim Eberlin. Jim co-founded Host Analytics, and while there he saw the opportunity to create a system to monitor customer success. It consisted of combining product usage data with qualitative customer satisfaction survey data (largely net promoter score or NPS) with CRM information like customer issues and product purchase data.
Integrating these disparate sources of data was a challenge for Jim and continues to be a challenge for most of the players in this category. And no wonder. There are many survey companies like SurveyMonkey, Qualtrics, Promoter.io, and GetFeedback. Similarly, there are many CRMs in the market. If you calculate all the permutations in tools across businesses, you suddenly realize the enormous resources required to integrate thoroughly for a 360-degree view of customer health. I saw it at Salesforce, and I would see it again, though briefly, years later at Tray.io.
The difference is when we ran into this data aggregation problem at Tray.io, we realized rather quickly that we could use our own product to integrate and automate the flow of data across systems to provide a comprehensive view of customer health. I wrote a detailed post about how it works on our blog. For example, our customer satisfaction survey data from Promoter.io is piped into Salesforce. This automated process puts the most current data in front of our customer success team, sales team, and marketing team. Additionally, we integrated our other sources of customer data like Salesforce, Intercom, and Amazon Redshift - to name a few. Now we have all customer health signals viewable in real-time.
Analyze and understand the leading indicators of customer churn
Coming back to the leaky bucket brigade at Salesforce, through a lot of data crunching we uncovered that customers who invested in training had a 56% higher likelihood of renewing. Similarly, we found that if customers integrated Salesforce with at least one other application, the combined entity was more “sticky.” Further, customer investments in proper implementation and support also showed statistically significant increased customer renewal rates.
At Host Analytics, we determined that there were plenty of cases of false positives in our churn detection model. In some situations, we’d detect a decline in user login frequency. When we looked closer, we realized that there were other issues at hand that didn't affect our churn rate like seasonality and employee churn within our customers. In other cases, sometimes a less than stellar NPS score by one person at a big account didn't materialize into churn risk. When you have accounts with 50+ logins, one unhappy user in a sea of happy users, doesn't mean that you'll lose the account. So, the moral of this story is keep on refining your models. Back-test them. Use the basic engineering feedback loop analysis to reduce error rates over time.
At Tray, we use Periscope Data as our BI tool, and have integrated all data sources to have full visibility into customer health. Over the last ten years, the analysis part of this process has gotten about 10X easier than before. Call it the miracle of APIs and flexible and easy to use Cloud software.
Determine your churn prevention strategy
For Salesforce, we had to change the way our services were packaged and sold to reduce churn. The goal was to increase the attach rate for services that would help our customer be more successful. Our attach rate for training and services was low and for a good reason. We didn’t have the packaging, pricing, or offer that fit Salesforce’s subscription business. We were still approaching services like the on premises software industry. For example, we were still trying to sell 3 to 5-day classroom training sessions.
So we invented Success Plans that bundled the right amount and type of services for customers across all segments of Salesforce’s business. This packaging was a feat considering that Salesforce sells to companies of all sizes: 5 person startups to 100,000 person enterprises. It's hard to define any package that could fit five go-to-market segments.
Our sales organization wasn’t trained nor incentivized to talk about training, support or consulting. We had to define Success Plans as a subscription so that it would fit into the sales channel and relieve quota. Once you can ignite a 1,000 person (now 5,000+) strong sales team to care about a service or product, attach rate goes way up.
We also had to evolve the Salesforce brand message from “it’s so easy to use that you don’t need training” to “training is a good investment to ensure that you get the most from your Salesforce subscription.” This message took a lot of marketing elbow grease to change.
We also had to improve new customer on-boarding processes among many other improvements that were part of our churn prevention strategy.
At Tray, we added alerting in Slack for each new customer satisfaction survey submitted by a customer, which allows us to get ahead of any customer issues. For prevention, we use a variety of strategies (though it’s probably best not to dive down that rabbit hole just yet).
Your prevention strategy will differ. It's about taking a data centric approach to reveal that factors in your business that produce happy customers.
Align all parts of your organization and roll-out improvements
Your support team is not solely responsible for improving renewal rates. It's the responsibility of every person at your company. While I was part of the swat team to figure out how to fix the leaky bucket, it was the change management and rollout that made it happen. Believe it or not, I enabled or trained almost every sales team at Salesforce on how to pitch Success Plans. This effort meant flying to Toronto, London, Dublin and many meetings at HQ in SF. There was a legion of organizational changes that Customers for Life leaders (Maria Martinez, Eric Kelleher, Dean Robison, et al) implemented.
It's likely that your rollout will be different. Make it a big deal. Take the time to train people. Make sure that changes take root by getting involved at each stop of the happy customer train.
Added benefit: Use newfound visibility into customer health to help your sales and marketing teams
At Tray.io, it became apparent that setting up these churn detection processes also helps the sales and marketing teams.
For instance, one of the mantras we had back when I worked for Salesforce.com was to "empower your customers to sell for you." This was a high bar because it meant we needed to identify clients that were successful, willing, and enthusiastic to participate in customer programs like case studies or sales reference calls.
But, if we already know which customers are reporting 10 out of 10 in customer satisfaction, we can more easily recruit them into customer programs. Meanwhile, our customer support team is satisfied, because we don't have to bug them for our happiest customer list. You can read this blog post How to integrate NPS into Salesforce CRM to learn how to set-up a “happiest customer detection” system.
In short, set-up an integrated system to sense churn risk and implement a holistic prevention strategy
This is all to say that after years of trial and error, we're finally on the verge of ending or at least significantly reducing, preventable customer churn. The key is integrating your systems and processes to get a full view of customer health and then implementing strategies that prevent churn. At Salesforce, the result of this effort paid off in terms of increased customer satisfaction and higher renewal rates. At Tray.io we’re getting ahead of it before it happens.