An election is only as credible as the ballot in front of the voter. A result can be technically flawless — every credential checked, every preference counted, every report signed off — and still attract complaints if members finish voting unsure whether they answered the right question, in the right way, for the right candidate. Confusion is expensive. It drives helpdesk calls, informal votes, abstentions and challenges to the outcome that consume weeks of board, committee and secretariat time.
For company secretaries, boards, unions, HR teams, associations and clubs across Australia and the Asia-Pacific, the practical challenge is rarely the count itself. It is the moment a member opens a link on their phone, between meetings or at the end of a shift, and has to work out what they are being asked to decide. Good design removes that friction before it appears.
Start With Plain Language, Not With the Technology
Ballot design decisions are often made late in the planning process, after the notice of meeting has been drafted and the constitution checked. Yet the wording of the question is the single biggest influence on whether a member understands what they are being asked to decide. Voting is an infrequent task for most people: a member may cast a ballot once a year, or once every three years. There is no muscle memory to fall back on and no opportunity to learn from a previous attempt. Every term, every instruction and every extra screen has to earn its place.
The strongest starting point is mirroring the documents members have already received. If the notice of meeting describes an item as “Resolution 3 — That the remuneration report be adopted”, the ballot should carry the same number and much the same wording, not a paraphrased headline. Consistency between the paper trail and the screen removes the small doubt — is this the same thing I read about? — that leads people to abandon a ballot or telephone the office.
Plain language does the rest. Where a resolution has a practical effect, say so in a line of help text: what happens if it is carried, and what happens if it is not. Avoid bundling unrelated decisions into a single question, because voters who support one element but not another are left with no honest way to answer. Choose one answer vocabulary — Yes and No, or For, Against and Abstain — then use it identically across the whole ballot and the supporting material. If abstaining is offered, explain whether it counts towards the result or simply records a position.
Terminology needs the same scrutiny. Phrases such as ordinary resolution, special resolution, enterprise agreement and special general meeting are familiar to the people who drafted them and opaque to many who must vote on them. A short explanatory line, written once and reused everywhere, does more for comprehension than a longer block of instructions at the top of the screen. The same applies to counting methods: where preferential voting is used, state plainly how many boxes must be numbered, what happens to a ballot that stops early, and whether an incomplete ballot will be accepted as formal. Members should never have to guess at the rules of the exercise they are part-way through completing.
Anatomy of a Digital Ballot That Works on a Phone
A ballot that works beautifully on a desktop monitor and badly on a phone is not a working ballot. Many members will open the link on a mobile device, often in a hurry and sometimes on a poor connection. The layout has to respect that reality before any consideration of features.
The opening screen should answer four questions immediately: what is being decided, who is eligible to vote, when voting closes (with the time zone spelled out, which matters for organisations with members across Australia, New Zealand, Singapore and the wider region), and roughly how long the process takes. A member who can see the end point is far more likely to start. From there, one question per screen, with the question text at the top and any help text directly beneath it, keeps attention where it belongs.
Options and candidates need consistent treatment. Names should appear in a stated order — alphabetical, or randomised for each voter — and the rules should say which, so that nobody can later argue the ballot was stacked. Candidate statements should be capped at the same length and displayed in the same format for every nominee. Where photographs or position titles are included, they should not create an advantage for one candidate over another. For preferential ballots, a simple numbered guide beside each option, showing the sequence as it is built, prevents the most common error: two candidates given the same number.
Before submission, a review screen showing every choice is essential, along with the ability to go back and change an answer. The final button should say something unambiguous — lodge my vote — rather than a generic next or submit that a nervous voter may hesitate over. After lodgement, the voter needs proof: an on-screen confirmation and an email receipt recording the date, time and a unique reference. The receipt should confirm that a ballot was cast, never how it was cast, so that verifiability and secrecy are both preserved.
Accessibility and resilience are not optional extras. Screen-reader labels, strong colour contrast, keyboard-only navigation, resizable text and clear warnings before a session times out all serve the same goal: nobody should be excluded by the interface. A visible support contact on every screen, and a save-and-return option for longer ballots, will recover voters who would otherwise give up part-way through.
Test With Real Voters, Then Protect the Result
No ballot should meet its electorate for the first time on the day voting opens. Testing with real voters — not with the board, and not with the project team — is the cheapest risk control available. Recruit a small group that reflects the actual membership: long-standing members who still prefer paper, younger members who will vote on a phone, shift workers voting at odd hours, anyone who uses a screen reader, and members dialling in from another state or country. Ask each of them to complete a specific task and watch where they hesitate. Hesitation is the finding; opinions are not.
A dry run follows. Run a dummy election with fictional candidates and a short voting window, then walk the full sequence: credential distribution, reminder messages, the ballot itself, the confirmation receipt and the final report. Check how reminders render on common devices and whether the link lands the voter directly on the ballot rather than an intermediate login screen. Smart reminders lift participation only when they are easy to act on, and a reminder that sends a member through an extra authentication step to reach the ballot wastes the goodwill it was sent to create.
Trust is built in the same unglamorous places. Publish the voting period, the counting method, the eligibility rules and the name of whoever is conducting the count. Where a ballot is contested, high-stakes or subject to close scrutiny — enterprise agreement votes, board and committee elections, motions put to a divided meeting — engaging an independent returning officer and keeping a documented audit trail answers most challenges before they are raised. A clear record of who voted, when, and how the result was certified protects the organisation as much as it protects the outcome.
None of this depends on a complicated platform; it depends on disciplined configuration of whichever platform is chosen. Whether the process is run in-house by a company secretary or delegated to a returning officer, the principle holds: a digital ballot should be designed for the least confident voter in the electorate, not the most comfortable one. That means careful wording, testing on a phone in poor light, plain explanations of the counting method, and a receipt that proves a vote was lodged without ever revealing how it was cast. Members rarely notice that work. They simply vote, and they trust the result.
