Family tree building @ Findmypast

Problem

Starting a family tree from scratch is difficult: users need to figure out how to use the information they have at hand to find their next clues, then start to use historical records to piece together a picture of their past. Findmypast's family tree maker breaks down these barriers for users through an onboarding flow and the provision of automated record suggestions, but users still struggled to establish a tree building habit beyond their first 3 generations. The challenge: figure out how to help new users establish a tree building habit and and improve their retention in the process.

What we did

After an initial discovery, we established three key user experience problems:

• Direction: after arriving at the tree canvas with clear intent to discover unknown ancestors, they didn't know how to find the next generation back from the family members added during onboarding.

• Access: they quickly discovered a free trial is necessary in order to find any ancestor using historical records, a big ask before experiencing any value or delight.

• Usability: most new users arrived on mobile devices and couldn't navigate the next phase of the experience easily because it had been designed for desktop only.

We watched genealogists build a tree from scratch and identified a common pattern in their approach to discovering unknown ancestors:

• They first found the birth records of known family members in order to discover their mothers' names

• Then located the mothers' records in order to verify them and their partners' names in marriage records

• After adding the new parents, they then repeated the process again until a relative could be found in a census record, where rich details such as addresses, occupations, employers, and neighbours helped them unlock a whole host of research avenues.

Once users got to that census horizon, we observed a flurry of activity and excitement. So this became our experience goal: getting users from Level 1 to Level 2.

Our experiment results showed 60% of new tree builders engaged with suggested relatives on their first day, and that those users were then twice as likely to come back the next day and over the course of their first few weeks. We also saw that removing paywalls on the early tree builder records vastly increased the number of new users who reached 'Level 2' (as illustrated in the paywall graph below) presenting the next problem to solve: getting users to Level 3.

Our experimentation laid the groundwork for Findmypast's subscriptions team to take the concept and build out our new subscription package offering a free tier, a tier that enables new users to build with only record suggestions, and an everything tier.

What we achieved

We then built quick-fire tests to figure out how best to help people get to the next level. Our tests ranged from light touch explainer modals, to interactive tasks and automated relative suggestions. This helped us quickly get a view of the impact on users and the effort required to scale up and productionise. One of our most effective and easy-to-implement solutions was our second experiment, Suggested Relatives, which simply showed users the ancestor they could add to their tree by engaging the record suggestion we surfaced for them through an automated behind-the-scenes search.

Our next challenge was access. The family history hobby involves cycles of discovery that gradually increase with complexity, so our hypothesis was that allowing users enough free access to complete a first cycle would increase the volume who could make it to the next stage.

We worked with the licensing team responsible for royalties and the engineering team responsible for the record entitlement service to scope out what sort of free access variables were feasible, an ambitious proposition considering the company had not tried free access before. After a outlining a set of options to senior leadership, we achieved sign off on a proposal to make birth and marriage records free to new users, the records we knew were critical to advancing to the next stage of the hobby.

Then we had one major challenge left: redesigning the process of receiving a record suggestion, reviewing it, and merging it into your tree so that it could create a new relative for you and produce new actions. You can see the original experience below on the left, and the new mobile-first experience we built on the right.