With more than 19 million customers worldwide and 78 million domain names, GoDaddy is the world’s largest domain name registrar – more than 4 times bigger than its closest competitor. The company prides itself in staying true to its founding values as a successful startup: hustle, adapt, listen, repeat. Being able to listen to its customers and adapt accordingly has been central to keeping GoDaddy as the market leader.
Data Points
- Saved $200K by not building or maintaining their own experimentation platform.
- Improved sign up rates by 8% via experimentation.
Background
Not surprisingly, the mantra trickles down to their engineering teams as well. Key to adapting, listening and repeating, is the ability to carry out customer-engaging experiments, and being able to respond quickly to customer feedback, a need recognized by Björn Levidow, Sr. Director of Product Management. In this role, Björn leads experimentation efforts at GoDaddy.
Across the company, we had 15 engineers building different experimentation and feature flagging platforms… that’s a lot of money we were spending, and just from that point we’ve already realized the ROI with Split.
Björn Levidow
Sr. Director of Product Management at GoDaddy
Buy vs Build
Björn wanted GoDaddy to be able to carry out experiments “at the velocity of thought”. He envisioned creating an air traffic control for experiments, a standardized platform and a repository of all experiments for all the disparate teams at GoDaddy wanting to learn from their customers.
By Björn’s estimates, creating such a platform would have needed 3 engineers working full-time on the project for over a year. In addition, building a platform would require continuing maintenance work, the opportunity cost of diverting highly talented (and expensive) resources, and the cost of driving adoption across teams.
Hence Björn began to look for vendors in the space, and sought recommendations from the teams that had looked into experimentation platforms. GoDaddy’s acquisition of Main Street Hub, a Split customer, introduced Björn to Split.
Choosing Split
Of the 7 vendors Björn looked at, Split was the only one built specifically for engineering teams. At Main Street Hub, Split’s experimentation platform had been crucial for developing and testing the most optimum machine learning models for sales qualification, where Split’s ease-of-use and developer-friendliness allowed for rapid iteration and a fast adoption across teams.
Split’s simplicity meant GoDaddy was able to set up Split experiments with only 30 minutes of work. Through the evaluation period, it became clear that Split provided the most trustworthy results, and had the highest statistical rigor. Split’s fine-grained customer segmentation capabilities, targeting rules and phased rollouts allowed for the greatest flexibility in experiment design.
For example, an engineering team can first run an experiment in a non-core geography to test out new ideas before expanding the change to more users. Furthermore, Split’s REST API surface enabled integrations with key workflows, easing adoption across teams.
Split tells us exactly what’s happening and gives us trust that the experiment is being run correctly.
David Meyer
Sr. Product Manager at GoDaddy
In addition to enabling GoDaddy to carry out “experiments at the velocity of thought”, Split’s feature delivery capabilities emboldened the teams to take yet more risks and increase output. With the ability to limit the blast radius, roll back changes quickly, and find the root cause of issues in minutes, developers feel more confident in delivering more features.
Business Impact
Not only did GoDaddy realize immediate ROI by avoiding the engineering effort and opportunity cost of 3 FTEs, the team was able to realize tangible revenue improvement via experiments. The Sign Up team, responsible for user authentication, experimented with the password requirements flow on mobile and was able to improve sign up rate by 8%.
We had the intention to change the password field, but we didn’t have a way to know how good it was effectively and to be able to control it within our product team. Split allowed us to do all that and more.
David Meyer
Sr. Product Manager at GoDaddy
Performance experiments on page load times led to an improvement in clickthrough rates, which correlated directly with customer experience. By leveraging Dynamic Configurations, an ability for engineers and product managers to make feature adjusts post-release without code, the team is able to minimize the engineering time needed to make changes to their features, and thus maximize engineering productivity.
Future Usage
Today, GoDaddy is aggressively expanding the use of Split across the company by driving a culture of experimentation, building Split into core experimental platforms, and evangelizing Split to all of their global subsidiaries.
The team is exploring means to understand cart abandonment better and weave in experiments that can lead to a reduction in cart abandonment – a project that would have a transformational impact on the company’s top line growth.
Inspired by an internal hackathon project to make their Application Performance Monitoring tools more relevant for a feature flagging world, they’ve also begun testing Feature Monitoring, a Split add-on to get instant alerts if there’s a performance degradation at the beginning of an experiment or a feature release. Björn envisions a future where 100s of experiments are run on the Split platform every week, helping GoDaddy make smarter product decisions every time.