3 BizOps lessons from DoorDash

Jessica Lachs, VP of Analytics and Data Science recently did a podcast with Lenny. Here are 3 hard-hitting lessons from it that I think are applicable to almost every tech business.

Lesson 1: Do NOT measure your customers’ health using a composite score

This is one of my pet peeves especially in Customer Success orgs. If you look at the leading software that are used by CS teams, a very important module is some version of a ‘Customer Health Score’. It’s typically made up of a slew of metrics including number of logins, product usage, account growth, customer support tickets, product surveys, renewal and expansion history, and

The problem with having an “exhaustive list” to make up a customer’s health score is that they become incredibly noisy. They are probably okay for reporting at an overall customer base level to see changes over time, but they are HORRIBLE at providing insight into what a team member should focus on.

Jessica mentions that they had a similar experience at Doordash with their Merchant Health Score. They calculated it as a composite metric to see the likelihood of the merchant being on the platform and getting orders.

“It included many things like the number of orders, number of active hours, whether the merchant had a full menu listed, what % of their SKUs had images, and many other metrics. All of these factors were then given some weights and then a comprehensive Merchant Health Score was created. Someone found that one of the merchants had a score of 0.35! Ummm, so what does that even mean?|

It was so hard to understand what it was and how to move it, that it became meaningless. “

The lesson for BizOps teams is basically this: avoid composite metrics for important things like customer retention. A simple metric that’s intuitive to understand is more valuable than a perfect, comprehensive one.

If you do want to use a few different metrics, figure out 2 or 3 factors that are most important to the health of the customer and just focus on those before moving on to the next ones. This simplification will make your org move faster and in the right direction.

Lesson 2: Every initiative should be measured in a common currency

Businesses must choose which projects to pursue from many options. For instance, to grow DoorDash, should the focus be on lowering pricing or lowering delivery times or signing more restaurants or onboarding more dashers (delivery persons) or should they make improvements in the login flow?

It’s very hard to decide. By having a universal impact metric, DoorDash actually is able to compare the business impact of each potential project and decide to allocate resources accordingly. For DoorDash, this universal currency is Gross Order Value (GOV). So they run mini experiments, and quantify how each project is likely to have an impact on GOV and then decide on which ones to focus on in the next few months.

Lesson 3: Focus on edge-cases and hidden customer experiences

Most metrics are designed as averages. But this often means that edge cases are left out of the focus of the team. For example, DoorDash has orders which are never delivered. A very small % of consumers ever face this but when they do, it’s absolutely the worst experience for them, increasing their likelihood of churn dramatically. They are also very expensive for DoorDash and severely impact the unit economics. If we only looked at traditional “average based” metrics like average time to delivery, they just wouldn’t turn up. So DoorDash has a metric called Never Delivered defined as the % of orders that were never delivered, and there are teams both in product and ops whose sole goal is to eliminate this.

Another interesting point was that there are often customer experiences that are hidden from us. For instance, login errors. There are users who are probably stuck in the login screen and never even enter the product to place an order. These don’t show up in the data but it’s important to be aware of them and solve them proactively.

See the entire podcast here: