Methodology

How Build a Ballot works

Build a Ballot asks Users a series of questions about key election issues such as housing, energy costs and healthcare. Based on Users' responses, Build a Ballot then calculates a Match Score with Candidates in their electorate by comparing how closely their views align on different policies.

Summary of the five stages below:

1. Identify the issues

We identify the major issues facing voters by reviewing current polling, news coverage, Party and Candidate announcements, civil society priorities and academic research.

2. Identify key policies

For each Issue Area, we develop a set of policies. We prioritise policies that are likely to effectively address the relevant issue, are relevant to public discourse, can be understood by most users and represent a genuine range of positions across the political spectrum

3. Assess Candidates

We assess Parties and Candidates against each Policy on the same 4-point scale that users engage with within the tool. For Incumbents, we draw on public records, voting history, and official Policy documents. For Non-Incumbents, we invite them to complete a questionnaire and provide supporting evidence. We share all Assessments with Candidates and parties, providing an opportunity for feedback before publishing.

4. Match voters

Once a User completes the Policy Questionnaire, Build a Ballot compares their responses against assessed entities' positions and calculates a Match Score for candidates in the lower house and parties in the upper house. We use percentages and colour-coding to display results: green (above 70%), yellow (40–69%), and red (below 40%).

5. Encourage further research

Users can then explore where Candidates stand on major issues, read Party summaries and Candidate biographies and pitches, rank Candidates in their order of preference, and screenshot or email their Personalised Voting Plan to bring to the ballot box. Build a Ballot is designed as a starting point — the final ballot order is always at the voter's discretion.

‍

Stage 1: Identify the issues

1.1 The criteria we use to select issues

At each election, we include eight Issue Areas in the Build a Ballot Policy Questionnaire. We prioritise Issue Areas that:

• matter to a broad range of voters, particularly younger voters
• elicit different Policy approaches from Parties and Candidates, so a match score is meaningful
• can be meaningfully addressed at the relevant level of government

We can't include every Issue without making the Tool overly complex. Our aim is to strike a balance between depth and accessibility.

1.2 Our process for identifying issues

We identify issues by:

• reviewing current polling and public opinion research
• reviewing news coverage, Party and Candidate announcements, and civil society "asks"
• reviewing relevant academic research and expert analysis
• consulting with academics, subject matter experts and independent organisations
• speaking with people "on the ground"

We use AI to monitor hundreds of sources and surface links to trusted publications. While AI Tools feed into our processes, we always verify, cross check and make final decisions ourselves. 

‍

Stage 2: Identify key policies

2.1 What we consider a Policy

A Policy is a specific, actionable proposal. We look for proposals that:

• are active — they involve a ‘change’ rather than maintaining an existing system
• name a clear mechanism — such as spending, tax, or regulatory changes
• can be measured — it should be clear if/when a proposal has been implemented

We try to strike a balance between oversimplified and overdetailed. We avoid general announcements or slogans as these aren't specific enough for voters to make meaningful comparisons, as well as implementation and administrative details as these often require specialist knowledge to interpret.

2.2 The criteria we use to select Policies

For each Issue Area, we select four Policies. We also select an additional four policies that do not address any of the core issues. We prioritise policies based on:

1. Political balance — policies should collectively represent a genuine range of political positions
2. Quality — we prioritise policies most likely to effectively address the relevant Issue Area. We consider quality to be a combination of;
   
   • Credibility — is supported by reasonable evidence (ideally from independent experts); is endorsed by multiple organisations and/or experts
   • Value — uses shared resources (eg. financial, environmental) mindfully. Where two policies are expected to deliver similar outcomes, we prioritise the less costly of the approaches.
   • Long-term stakes — considers the intergenerational impacts, i.e. consequences that extend well beyond a single election cycle
   • Civility — enables constructive and informed debate; does not actively harm communities
3. Relevance — are discussed and understood by the public
4. Understandability — are accessible and comprehensible

The first criterion is our top priority and sometimes sits in tension with the others — for example, we've included policies with limited expert support to reflect the full range of positions held by major Parties. Our process (2.3) is designed to ensure this tension is carefully considered and to mitigate potential bias.

2.3 Our process for identifying & selecting key policies

To identify and select policies, our process has six stages:

1. An internal research process, drawing on the sources listed in 1.2
2. A public Policy submission process, where we invite people to submit proposals.
3. Consolidation of policies into a shortlist
4. An expert review period
5. A focus group, to test how policies land with everyday voters
6. A final review and sign-off against the criteria in 2.2

Policy briefs

For each Policy considered — we write a brief outlining the problem, solution, and arguments for and against. We publish these on our website.

2.31 Internal research process

Drawing on the sources listed in 1.2, we build an initial list of Policy options for each Issue Area. We continually monitor for Policy announcements until the cut-off date, adding significant Policy proposals as they’re announced.

2.32 Policy submission process

Around two months before the election, we open a month-long public Policy submission period. We publish the Issue Area we expect to include in the Policy Questionnaire, along with example policies and a submission form, so anyone can put forward Policy ideas. We provide instructions covering our criteria (2.2) and quality standards.

Submissions must describe the Policy and the problem it addresses, and set out arguments for and against (where possible), with sources. We ask submitters to cite direct, primary sources where possible (such as government data, peer-reviewed research, or official reports), to favour Australian evidence where available, and to describe research findings accurately.

We review all submissions against our criteria and quality standards, and may follow up with submitters for further evidence or clarification. Submissions that meet basic quality standards will be published on our website where policy experts, researchers and non-partisan organisations are invited to provide feedback or note their endorsement or opposition.

2.33 Building the shortlist

Once the submission period closes, we consolidate our internal research and public submissions into a shortlist of up to ten policies per Issue Area, applying the criteria in 2.2. We publish this shortlist on our website.

2.34 Expert review

We send our policy shortlist to relevant subject matter experts and ask them to assess perceived quality (see 2.2) and flag any inaccuracies or missing context. We also ask a smaller number of generalist advisors — with expertise in democracy and public deliberation rather than a specific Issue Area — to review the shortlist as a whole against the criteria in 2.2, including political balance and clarity.

2.35 Focus group

We convene a focus group of everyday voters, recruited to reflect a mix of ages, genders, backgrounds and political views. We brief participants on our process, the Issue Areas and the policies, and ask them to narrow down the shortlist to four policies per Issue Area, based on the criteria in 2.2. Their feedback helps us understand where Users may perceive or interpret political bias, as well as emphasising relevance and understandability alongside quality considerations.

2.36 Final review and sign-off

We conduct a final review of the selected policies against the criteria in 2.2, checking both the political balance of the policies themselves and, where relevant, how Parties' known positions on them are distributed.

‍

Stage 3: Assess Candidates

For each Policy included in the Tool, we assess the Support Level of each Candidate or Party in order to provide a Match Score with Users.

3.1 Eligibility

All Candidates and Parties that appear on the ballot published by the relevant electoral commission will be included in the Build a Ballot Tool. However, not all Candidates and Parties are eligible to receive a Match Score with Users.

Note: just because a Candidate or Party is eligible, does not mean they will automatically receive a match score calculation (see 3.2).

3.11 Permitted Parties and Candidates

By default, all Candidates and Parties are eligible to receive a Match Score, unless they meet the criteria for ineligible (3.12) or flagged (3.13).

The process for being included differs depending on whether a Candidate is an Incumbent or Non-Incumbent (see 3.2), but eligibility itself does not depend on Party or political position.

3.12 Ineligible Parties and Candidates

Parties and Candidates are not eligible for a Match Score if they:

1. are flagged (see 3.13)
2. intentionally provide misleading information through the Build a Ballot process

3.13 Flagged Parties and Candidates

Parties and Candidates may be flagged in the Build a Ballot Tool if:

1. there is clear, public evidence that a Party has been set up to intentionally mislead voters (for example, registering under a name designed to be confused with another Party or cause)
2. they are proscribed terrorist organisations, or their representatives

In the case that a Party or Candidate is flagged, they will be ineligible for a Match Score and additional context will appear next to their name within the Tool.

3.2 Our process for assessing Candidates

3.21 Parties vs independents

We automatically assign Candidates running under a political Party the Policy Support Levels of that Party. We assess independents based on their individual Policy Support Levels.

Where a Candidate contacts us and can demonstrate — through evidence such as parliamentary voting records — that they hold a different Support Level to their Party (for example, having crossed the floor), we assess them individually on a case-by-case basis.

3.22 Incumbent Parties and Candidates

For Incumbents, we determine Support Levels using publicly available information, including:

• official Policy documents
• parliamentary voting records
• media appearances and interviews
• public statements and social media posts

We share our preliminary Assessment with each Party or Candidate and provide at least two opportunities for feedback before publishing.

If there is insufficient public information — and the Candidate or Party does not respond — they will not receive a match score and will be greyed out.

3.23 Non-Incumbent Parties and Candidates

We invite all Non-Incumbents with publicly accessible contact information to complete a Policy Questionnaire indicating their Support Level for each Policy, based on how they'd approach the issue if elected. We ask for supporting evidence, which may include official Policy documents, public statements, media appearances, social media posts, or relevant organisational endorsements.

We review all responses and evidence before publication. Where we identify inconsistencies or insufficient evidence, we contact the Party or Candidate and give them the opportunity to clarify. All statements provided to us by candidates will be published in the Transparency Centre on our website.

3.3 Determining Support Levels

Irrespective of incumbency status, we develop an Assessment for each assessed entity which includes an assigned Support Level for each Policy as well as supporting evidence. This evidence may be a combination of sources listed in 3.22, as well as statements submitted directly to Build a Ballot.

For each Policy, the assessed entity’s Support Level will be assigned using a 4-point scale that broadly maps to expected parliamentary behaviour.

Importantly, a Candidate or Party saying they ‘support’ or ‘oppose’ a Policy in our Policy Questionnaire does not automatically guarantee that we will assign that Support Level. If there is contradictory evidence (eg. voting records), we will take that into consideration when assigning Support Levels.

Because we weigh multiple, sometimes contradictory forms of evidence (voting records, public statements, questionnaire responses), assigning a position always involves a degree of judgement. We publish these decisions, and the evidence informing them, in the transparency centre on our website (see 3.4).

Where we can't determine a position, or a Party/Candidate chooses not to respond, we record "no information available", and that Policy is excluded from their Match Score calculation (see 4.1–4.2).

3.4 Public feedback and transparency

We publish our Assessments — and the evidence behind them — in a transparency centre on our website, and invite anyone who believes a Candidate or Party has been unfairly assessed to get in touch.

‍

Stage 4: Match voters

When a user completes the questionnaire in the Build a Ballot tool, they indicate their support level for each policy. Based on their selection, they receive match scores with their local candidates and parties. This match score is entirely dependent on the user’s submission and represents the percentage of policies that the user and candidate agree on.

4.1 Calculating a Policy Agreement Score

For each Policy, we compare the User's Support Level with the assessed entity's Support Level. If both are on the same side (both support or both oppose), the minimum agreement score is 0.75. If they're on opposite sides, the maximum agreement score is 0.5. This reflects the assumption that "somewhat support" and "strongly support" are closer together than "somewhat support" and "somewhat oppose."

4.2 Calculating the Match Score

Match Score = (sum of Policy Agreement Scores ÷ total number of policies) × 100

The Match Score is displayed next to each assessed entity's name, colour-coded as follows:

• Equal to or above 70% match = green
• 40–69% match = yellow
• Below 40% match = red
• No information available = grey

Candidates and parties will be displayed from highest to lowest match score, with greyed out candidates and parties at the bottom. Where two assessed entities have the same match score (or no match score), ballot order will be used to sort.

‍

Stage 5: Encourage further research

Build a Ballot is designed as a starting point for voters, not a comprehensive Policy resource — it's intentionally quick to complete the Policy Questionnaire to meet how most voters actually engage with an election. Once a User has their Match Score results, we surface further content to support deeper research before they finalise their vote:

• Additional resources — based on the ‘top issues’ selected by the User in the polling questionnaire, we will recommend additional resources such as scorecards and reporting from outside organisations
• Candidate pitches — a short, open-ended statement (around 100 words) in the Candidate's own words

Importantly, the final order of preference always remains at the User’s discretion. Users are prompted to rank Candidates in their own order of preference (in either ballot order or preference order), and screenshot or email their Personalised Voting Plan to take with them to the ballot box.

‍

Have a question? Want to provide feedback on our methodology? Get in touch: hello@buildaballot.org.au