Straight Fire Search 🔥

Feature Overview

FamilySearch has four well used products that are currently searched individually. In today's world people expect to type anything in the search and get back relevant results regardless the format you are looking for. This is our second feasible attempt to move in that direction. The first attempt was so well received by people that we wanted to expand the ability to more. Account creations increased by 25% if people used the first version of this feature. 

Requirements

• Target beginners initially but make it expandable to experienced family historians.

• Update the search results so that repetitive information isn't always at the top.

• Allow people to update the search on the page.

• Allow this to be used logged in and logged out.

• Give people a next step instead of just leaving them hanging.

• Match new design system

• Determine how much information should be included with each result

• Encourage users to input less information in the search–our search works best that way.

Learning From the Past

ATTEMPT ONE
The search was created to provide value to people without creating an account. That turned out to be information about a surname, historical records, ancestors, and photos and stories. (See images above.)

KEY LEARNINGS
In a survey new users said the were interested in searching historical records. However, 41% of their time was spent looking at ancestor results, and ancestor results were 80% more likely to be clicked than historical records.

People didn't find the historical records very helpful because there wasn't very much information included in the search results. 

Hopes for the Future

THE PLAN
The first phase of this plan is to allow logged out and well as logged in users have access to an aggregated search experience and give them the next steps to go to results or advanced search. Future phases will focus on a guided process to viewing results logged out and saving their newly found ancestors.

UPDATING UI
A big focus for this phases is to update all the user interface design to match the new Material design inspired design system. In this process, we designed and documented new reusable components. The look and feel was created as well as what pieces it would include and how it would function.

Testing the Updates

PROTOTYPE–DESKTOP
We reduced the search  fields to only include the core elements. We allowed for more relevant information to appear at the top of the results page. We added an affordance to edit or update a search and matched all pages to the new design system. 

ROUND 1 TESTING
The core pieces of the flow, the search and results pages, tested well. But there were several problems with getting to the search page, who's information to search for, scrolling the results page, and updating a search. We also learned that users appreciated having an advanced search option available. 

PROTOTYPE–MOBILE
There were a few changes made to increase the scrollability of the results page but the rest of the prototype stayed the same. We wanted to see how users would use it on a mobile device.

ROUND 2 TESTING
People still struggle with getting to the search page, who's information to search for, as well as updating the search. The biggest flaw with the test is that people don't actually get to enter in what they really want so the results aren't as good as they could be. We need to build this and test with real data.

Hitting the Misses

ADDING  HELP TIPS
On the search page, we added help text if people entered a birth year that was within the last 110 years to let them know they won't find living people here. On the results page, we added a back arrow because most people said they would go back to update their search. We also added tips on each category to let people know what to do if they were not finding what they are looking for.

ADDING FILTERS
We decided last minute to allow people to add a few more details to their search on the results page. We want people to see some results before they start refining their search. So we decided to add filters on each category of results. Our design system uses chips and people found they worked well and were less intimidating the more fields. 

NEXT STEPS
We felt like we tested this product as cheaply as we could and go great results. Our next plan is to build what we have and test with a small percentage of users so we can really understand what is not working.