A technology-focused company, then and now

As the global leader in Family History, Ancestry is a technology company with a very human mission: to empower journeys of personal discovery that enrich lives.

2023 marks the company’s 40th anniversary. Over the last four decades, Ancestry has continuously innovated, evolving from magazines to CDs to online platform and mobile. “Technology underpins everything we do,” said Ancestry’s Director of Engineering Partha Sarathi. “Our goal is always to deliver the best experience to our subscribers to enable them to quickly and easily learn more about who they are and where they come from.”

To enable this, Ancestry moved to the cloud in 2017 to better manage updates and [use containers] containers. Ancestry customers continue to benefit from technology, including Ancestry’s award-winning mobile applications, the use of AI-powered Handwriting recognition technology and machine learning.

“I know that we are able to impact our key business metrics in a positive way with Statsig. We are definitely heading in the right direction with Statsig.”

Partha Sarathi

Director of Engineering, Ancestry

Leveraging experimentation for rapid growth

In 2018, Ancestry discovered that a legacy dynamic configuration tool no longer met the demands of the business. As a result, Partha and his team started looking for a solution to help launch dark and improve experimentation.

Ancestry wanted the ability to quickly ramp up and ramp down traffic, but a single end-to-end experimentation and analytics platform didn’t exist yet. As Partha states, “Growth and experimentation are two sides of the same coin.” The search for a solution included using Split for delivering experiment treatments and evaluating Optimizely. At the same time, the Marketing organization ran a small number of highly customized tests on their own tools.

Despite trying different A/B testing and feature flagging solutions, Partha says “The culture of experimentation wasn’t there yet.” Teams only ran a few tests per year, limited primarily due to the lack of consistency in tooling that didn’t lend itself to automated and easy experiment setup and evaluation across multiple teams.

Introducing a modern experimentation and analytics platform

In 2021, increasing experimentation and launch velocity became a high priority for the business in order to innovate more quickly and measure what was contributing most to the growth of the product. Partha began discussions with the Statsig team in late 2021 and kicked off a proof of concept (POC) in early 2022. Partha brought in product managers, engineers, and analysts to stress test Statsig as a cross-functional product solution that would meet their needs across feature management, experimentation, and marketing use cases.

“We only had so many analysts. Statsig provided the necessary tools to remove the bottleneck,” says Partha. Through the POC, Partha found that Statsig offered something for everyone. The agile Growth team couldn’t afford to occupy their three already very busy analysts with manual SQL queries and number crunching. In addition, Statsig offered the rich analytical tools product teams were looking for and on the engineering side, Statsig provided a way to natively measure impact thanks to the Holdouts feature, something that wasn’t possible with previous tooling.

The POC was successful and in May 2022, Ancestry hit the ground running with Statsig powering their modernized experimentation stack.

Experimentation as table stakes

With Statsig in place, running experiments to quantify the value of new product changes and using progressive rollouts to limit exposure of new features was a must. Two problems Partha wanted to solve were gaining the ability to slice and dice experiment results by segment or cohort and up-level existing experimentation programming with meta-analysis across experiments with Holdouts.

With the ability to drill down and group results by country, operating system, or even custom attributes set by the Ancestry team like free vs. paid customers users and high-engagement vs. low-engagement users.

This increased sophistication allowed the Growth team and Ancestry leadership to answer the question, “What are the benefits we’re reaping?” from new features. “If I go to Statsig, I see a bunch of experiments and a bunch of tags being created. So you can really see that people are starting to use it for everything that they’re doing. Any new feature that they are launching, any new experiment that they are launching, they’re using Statsig,” says Partha.

Evolving experimentation with a modern end-to-end solution

For the engineering side, the ability to use the Holdouts (especially global holdouts) is key. The “really big unlock” according to Partha is that the Statsig can automatically apply holdouts across new features, without engineers having to think about them. This is crucial from an engineering standpoint to make holdouts widely used.

The other key is the partnership between Ancestry and Statsig during and after deployment. According to Partha, Statsig’s dedicated and responsive support team has empowered the Growth team to “help us do it right the first time around.”

Over time, experimentation at Ancestry increased as did incremental improvements in their products to meet and exceed business targets. Key business metrics have been impacted in a positive way, and Partha’s colleagues are very pleased with the analytical capabilities of Statsig’s Ultrasound feature to see what’s really moving the needle.

Despite being early and in the learning phase with Statsig, Ancestry’s roadmap includes a mandate that every new feature has to be delivered via experimentation and has to have a holdout. In addition, senior leaders have weekly syncs to review analytics, as well as the results from experiments run in the last 8-9 months.

“Building a culture of experimentation is important. The journey of 1,000 miles starts with the first step. So, if this is your first step it’s good to have the right shoes on. If you know that you’re going 1,000 miles, you cannot go with flip flops.”