Online voting tools for associations: what to look for

Most boards start the search the same way: someone types "voting app" into a search engine, gets a hundred results, and cannot tell which of them would survive a member asking, three weeks later, how the board knows the count was right. The products differ far less on features than on one thing: whether they were built for polling an audience or for deciding something. Below is a short list of requirements to judge them on, and how the three common approaches score against it.

The seven requirements

1. A voter list you control. The tool must let you upload the members who are entitled to vote, from your own register, and freeze that list before the vote opens. Anything that simply counts whoever shows up has no way of telling a member from a spouse, a guest or a forwarded link.

2. A personal link or code per voter. One link per member, not one link shared with everybody. That is what makes "one member, one vote" enforceable. It also lets the tool see who has not voted yet without seeing how anyone voted, and it stops a member voting twice from two devices.

3. Secret ballot, per item. Elections and personal matters usually have to be secret, while the budget is normally taken openly. The choice belongs to the chair on the night, so the tool must offer it item by item, and it must genuinely separate the vote from the voter rather than promise not to look. The guide on secret ballots online sets out what that separation looks like.

4. Voting weights. In a housing association, a co-operative or a federation of clubs, a vote follows units, shares or member numbers rather than heads. If the tool counts one row per response, it cannot represent an owner with three flats or a branch with forty members. See the guide on weighted voting.

5. Proxies. A member who cannot attend gives their vote to someone else. The tool has to let one person cast several votes, apply any cap the bylaws put on the number of proxies, and record which votes were cast under a proxy.

6. A record. After the vote you want a document showing the eligible list, how many voted, the distribution for, against and abstaining, the majority rule applied, and timestamps for opening and closing. It should be produced by the system, not typed up afterwards, and it should not be editable by the person who ran the vote.

7. A price that matches your usage. Most associations vote once or twice a year. A tool priced as a monthly seat licence for a company will cost far more than the meeting is worth. Pay-per-vote or a small annual fee fits the pattern better.

Two further points are easy to forget. The first is where the member data is stored and for how long, which matters if your members are in the EU. The second is whether a member without a smartphone can still be given a vote, on paper or through a code read out over the phone, and have it entered with the right weight.

The three approaches

Survey and poll tools such as Google Forms, Microsoft Forms or Doodle. They are free, everyone already knows them, and they fail requirements 1 to 6. A form is either anonymous, in which case you cannot show that only members voted and only once each, or identified, in which case whoever owns the form sees every answer next to a name. There are no weights, no proxies and no tamper-evident record. The detailed comparison explains why no combination of settings fixes this.

Meeting software with a hand-raising or poll function such as Zoom, Teams or Google Meet. Better for running the meeting, no better for the vote. The poll counts whoever happens to be in the call, which is not the same as the eligible list. The raised-hand count is a headcount, so an owner with three flats breaks it immediately. There are no proxies, and the export is a spreadsheet rather than a record. Hand-raising does have one genuine advantage over both other options, which is that everybody can see it happen, and that is exactly why it cannot be secret.

Dedicated voting systems. These are built around the eligible list rather than the audience, so the seven points above are the product itself rather than add-ons. The trade-off is cost, one more tool to learn, and a dependency on a supplier at the moment of the meeting. Evote is one of several such systems: free for up to 50 voters, and 19 euro per vote above that. Whichever you pick, let the seven requirements decide rather than the price, and try it on a small, harmless vote before the general meeting.

When the cheap option is fine

• Finding a date for the meeting, or signing up for it.
• Non-binding consultations: which activities members want, whether there is interest in a summer party.
• Collecting nominations and proposals before the notice goes out, where you want to know who submitted what.
• An uncontested show of hands at a small meeting, on an open item, where nobody disputes the count.
• A small board deciding openly between meetings, if the bylaws allow it; see the guide on board decisions between meetings.

The line runs through consequence, not through size. If the outcome binds the members, spends their money, elects someone, changes the bylaws or could end up in front of a lawyer, use a system built for voting. If it is a preference poll, a form is fine, and it is honest to call it a poll.

Questions to put to a supplier

• Can we import the eligible list ourselves, with a weight per member, and lock it before the vote?
• Can an item be secret while the next one is open, decided during the meeting?
• How are proxies handled, and can we cap them?
• Can the chair open and close each item live and see the result before announcing it?
• What exactly does the record contain, and can the administrator edit it afterwards?
• Where is the data stored, for how long, and is there a data processing agreement?
• What happens if a member cannot get in ten minutes before the vote closes?
• What does one general meeting a year cost, all in?

Any supplier who cannot answer these in plain language has told you something useful. Run a test vote with the board first: import a list of ten, make one item secret, give one member a proxy and a double weight, and read the record it produces. Fifteen minutes of that answers more than an hour of demo.