Ballot Ready

Informed Electorate

Ballot Ready set out to create a more informed electorate for the U.S. election. Ollon helped build a platform where residents entered their address to see personalized candidate information for every race on their ballot. The system helped 750,000 U.S. voters access the information they needed to vote confidently across 20,000+ candidates.

Situation

Ballot Ready was on a mission to create a more informed electorate. They wanted to strengthen democracy by helping voters vote in a more informed fashion all the way down the ballot.

Solution

With the US 2016 election looming, and only 5 months to complete the project once funding was received, Ollon provided the development muscle needed to collect, catalogue and provide easy voter access to candidate information. We augmented the Ballot Ready internal team with three seasoned developers to meet these aggressive targets. Work commenced on an engine that automated many aspects of data collection, including the gathering and sorting of unstructured data from disparate sources. Candidate information was rapidly pulled together into a consistent, structured and geography specific format that was easy to access by voters needing insight to inform their decisions.

Results

For the 2016 US election, Ollon helped Ballot ready collect, organize and provide access to detailed information on more than 20,000 candidates within challenging timelines. The Ballot Ready website attracted more than ¾ of a million visitor voters prior to the election.

Technologies

NodeJs, MySql, Amazon Mechanical Turk, Ruby

Our Expertise in Action

GIS data focused

Every address in the country sits inside overlapping federal, state, county, and local districts, and Ballot Ready's platform needed to know exactly which ones applied to each voter. Ollon built tools to ingest district boundary files from many separate government sources, each in its own format, and normalize them into a common schema the platform could use. Resolving the overlap between jurisdiction levels was central to the work, since a single address could fall inside several boundaries at once. That normalized boundary data gave the platform the geographic foundation it needed to serve accurate information to voters nationwide.

Data platform

A single platform had to bring together automated data collection, boundary normalization, and address to candidate mapping so Ballot Ready's voters could get reliable ballot information from one place. Ollon built that platform ahead of the 2016 US election, combining the pieces into a system that organized candidate data across every level of government into a format built for public access. The platform delivered information on more than 20,000 candidates within Ballot Ready's five month project window. It went on to serve more than 750,000 voters in the run up to the election.

Team augmentation and staff scaling

With the 2016 US election five months out and funding for the project only just secured, Ballot Ready's internal team did not have the development capacity to hit that deadline on its own. Ollon augmented the team with three senior developers, working directly alongside Ballot Ready's staff rather than as a separate outside group. That added capacity let the team take on everything the work required, from automated data collection to the geography specific formatting voters would use, inside the five month window. The augmented team delivered organized information on more than 20,000 candidates before the election.

GIS/Mapping

Ballot Ready's platform needed to tell each voter exactly which candidates were on their ballot, based on nothing more than their home address. Ollon built the mapping logic that took an individual address and resolved it against the relevant federal, state, county, and local districts to determine which candidates applied. That address level mapping let the platform present each voter with a ballot specific to their exact location, not a general regional list. The mapping system supported more than 750,000 voters checking their ballots ahead of the 2016 election.

Unstructured data processing

Much of the candidate information available for the 2016 US election sat in inconsistent, unstructured formats spread across many separate government sources. Ollon built a system to gather that data and sort it into a usable form, drawing on Amazon Mechanical Turk to help process and verify the volume of raw records coming in. The result was candidate information restructured into a consistent, geography specific format that Ballot Ready's platform could serve to voters directly. That processing work supported detailed information on more than 20,000 candidates ahead of the election.

Large dataset handling

Ballot Ready's platform had to track more than 20,000 candidates across every level of government, then hold up under a volume of voter traffic the team could not fully test for in advance. A system sized for early development and testing was not going to survive hundreds of thousands of voters arriving in the weeks before the election. Ollon built the platform to handle that jump from test scale to election scale rather than assuming the smaller load would hold. The platform supported more than 750,000 voters through the run up to the election without buckling under that shift.