The latest article on customer health can be found here.
If you run an online technology business like many of the Totango customers you already know that the growth of your company relies, more than anything else, on delivering lasting value to your customers. Acquiring new customers is essential, but if many of them leave, your business is a leaky bucket that will never grow.
It is this realization that compels many online software businesses to build customer success teams tasked with building long-lasting, value-based customer relationships. Obviously, key KPIs for these teams are churn and renewal rates. These are essential indicators to the level of success your customers – and in return your business – are having in receiving/delivering value.
But it’s not enough to systematically drive customer-success. Good customer success teams also pay close attention to leading indicators that help them identify:
1) Which customers are in trouble, so they can proactively reach out and intervene for the better
2) Which segments of customers are struggling and which are doing well, so the team can focus resources and efforts where they are needed the most
3) Which customer-success team members are tracking more risk account, to balance load and improve overall team performance.
It is these needs that drive customer-success organizations to define customer-health KPIs and track them on an ongoing basis.
In this post I’ll go over the steps needed to build such a KPI and how to leverage them for action.
Step-1: Defining ‘First Value’ and the customer journey
We’re basically looking for a data-driven, objective way to answer the simple question “How is customer X doing?”.
Let’s walk through this using Dropbox as an example.
To assess the health of a customer we first map that customer into one of two stages:
1) Onboarding: Any new subscriber that signs up for Dropbox.
2) Established: Those that have already used 25M of space on their Dropbox, or those that have connected two or more endpoints (PCs, phones) to their Dropbox.
The former group tracks all users that subscribed, but not really make use of Dropbox. Everyone starts like that, but obviously if you stay in this situation for long you are not doing well as a customer.
The latter, holds all established customers that already experienced that magical “First value point” and started syncing significant amounts of material with their Dropbox. They have become an effective customer and we want to make sure they stay this way.
Every new customer is onboarding when they just sign up, but we expect them to get to their first value point in the first week of their subscription. So if a new subscriber is onboarding for 3 days, we raise a flag and label them “At Risk”
In the Dropbox case, we may say that established customers are showing signs of trouble when their Dropbox shrinks. The user is getting less files served and hence less value from their Dropbox. We define “At Risk” when the customer’s Dropbox shrinks by, say 10%, in any given month.
You’ll need to come up with a similar map for your service. It will likely be more intricate than our simplified Dropbox example, but you should aim for something simple. Another good real-life example is CloudBees’ customer lifecycle example.
The key point is to keep things simple while still having a way to map each customer to a stage and risk level.
Step-2: Collecting data
We now need to collect the usage data necessary to drive this sort of analysis. If you use Totango, you just hook up your application to the Totango APIs and let it do the rest of the work, in real-time on all your users. If not, we can do some manual work and some excel kung-fu to get there.
To know if a customer is onboarding or established, we need to know the size of their Dropbox, and the number of endpoints it has connected.
We also want to know the date the user signed-up, so that if 3 days have passed we can raise the “At Risk” onboarding flag. Finally, we measure the relative change in the customer’s Dropbox size for change of more than 10% on a monthly basis.
To get this data, you’ll probably need to do some raw-data scraping from internal databases. If you don’t have access to the data or the skills to pull it out, you’ll need to find a friend in the dev team to help you out (It’s usually quite easy to do, so small bribes can be very effective for this :)). Either way, you should end up with an excel spreadsheet that looks more or less like this:
NOTE: Obviously, for a service at the scale of Dropbox, maintaining an excel spreadsheet for the entire customer-base is not a realistic proposition. If you too have a large customer-base where this is impractical, you may want to run this analysis on a subset of users (e.g. those on the Dropbox for Teams plan).
Similarly, the real value of this analysis is when you do it on an ongoing basis and track trends and improvements. So eventually you’d want to automate this process.
With the health of individual customers at your fingertips, there are a number of things you’d want to use it for:
1) The customer success team should build different swimming-lane processes for the four groups. Onboarding projects are proactively managed, At-Risk customers monitored closely and established customers doing well are occasionally reviewed.
2) Track progress over time – Cohort and trend analysis. This is a great way to measure the impact of the success team as well as the effects of changes in your product or service delivery are having on customer success.
3) Analyze correlation and refine health score – The best health indicators are a good leading predictor of churn. You would want to check that correlation occasionally and adjust your health score if that’s not the case. A good way to do this is by looking at the relationship between churned customers and those that are at-risk for a prolonged period. If most customers that end up churning are precedented by a long period of risk, your health indicator is doing a good job in predicting unhealthy accounts.
There’s a lot of talk about actionable metrics for customer success. The key is to define a metric that’s correctly aligned with the value end-users realize from your service. That’s why simple pageviews, visit or even time do not mean much in this regard.
Building a health-score may not be obvious or easy (though I do hope this post helps), but the benefits are enormous. Not only will your team have more actionable data to work against, but it will help answer the most important question a pro-customer organization has: “How are my customers doing?”