Key takeaways
- →Dot voting narrows a long list quickly but cannot tell you how strongly anyone feels or who would refuse to live with the winner.
- →Ranked voting surfaces second choices and intensity, which is what produces a decision a group will actually stand behind.
- →Give each voter roughly one dot for every three options, with a floor of three dots and a ceiling of six.
- →Never rank more than about seven items; beyond that people rank the top three carefully and put the rest in arbitrary order.
- →Open dotting is contaminated by whoever votes first, so collect dots simultaneously and reveal the totals once.
- →The default sequence is dot vote to reach a shortlist of four or five, then rank that shortlist to choose.
The honest answer to dot voting vs ranked voting is that they are not competing methods, they are consecutive ones. Dot voting is a filter: it takes thirty options and gets you to five in about ninety seconds. Ranked voting is a decider: it takes those five and tells you which one the group as a whole can genuinely live with.
Treating them as alternatives is where workshops go wrong. Teams dot vote a list of twenty ideas, see a top result with nine dots against a runner-up with eight, and declare a decision that a third of the room quietly disagrees with. The dots never asked whether anyone objected, so nobody objected.
What follows is what each method actually counts, the specific numbers that make each one work, the two biases that ruin an open dot vote, and how to run the shortlist-then-rank sequence inside a single sixty-minute session.
How each method actually counts
Dot voting is approval counting with a budget. Each person gets a fixed number of dots and spends them on whichever options they support. The total under each option is the number of people who thought it was worth some of their limited allowance. That is a measure of breadth: how many people mildly approve.
Ranked voting asks each person to put a small set of options in order. Depending on how you count, you either award descending points to each position or eliminate the lowest-placed option and redistribute. Either way the result is a measure of order: what the group prefers relative to what else is on the table.
The distinction matters because these two measures disagree more often than people expect. An option that everybody places second and nobody places first will lose a dot vote outright and win a ranked vote comfortably. In most workshops that option is the right answer, because it is the one nobody has to be talked into.
✓ Pros
- •Dot voting handles thirty or more options without slowing down
- •Dot voting takes ninety seconds and needs no explanation
- •Dot voting shows clusters, which often reveal a theme nobody had named
- •Ranked voting exposes the option everyone can live with, not just the one with loud fans
- •Ranked voting makes a near-tie resolvable instead of leaving three options bunched together
- •Ranked voting captures intensity, so a strongly opposed option cannot win on mild support
✗ Cons
- •Dot voting cannot express opposition at all - there is no negative dot
- •Dot vote results are frequently within noise of each other and get read as decisive anyway
- •Open dotting lets the first few voters anchor everyone who follows
- •Ranked voting collapses above about seven options as people stop ranking carefully
- •Ranked voting takes two to three minutes plus explanation and confuses at least one person
- •Ranked results are harder to display live, so the room has to trust the count
What dot voting is genuinely good at
Dot voting earns its place at the moment a group has generated more material than it can discuss. After a brainstorm you might be looking at thirty-five sticky notes and twelve minutes left. Discussing each one takes ninety minutes you do not have; dotting them takes ninety seconds and cuts the list by three quarters.
It is also the only prioritisation method that works reliably on ideas nobody has read yet. Ranking requires you to hold the whole list in mind and place items relative to each other, which is unreasonable with more than a handful of options. Dotting only requires you to notice the ones you like as you scan.
The clustering effect is a genuine bonus. When dots pile onto four notes that all describe the same underlying problem in different words, you have learned something the individual counts never told you. Read the clusters before you read the winner.
Where dot voting misleads is in the precision people read into it. With twelve voters and four dots each, the difference between seven dots and five dots is two people changing their mind, which is not a mandate. Treat anything inside about a 20 percent spread as a tie and carry all of it forward.
What ranked voting is genuinely good at
Ranked voting exists to stop a group choosing something that a substantial minority actively rejects. Because every voter places every shortlisted option somewhere, an option that sits at the bottom of half the ballots cannot win, however enthusiastic its supporters are.
That property is what makes it the right tool for decisions with consequences. Choosing which of five roadmap themes to fund, which of four office moves to make, which of five candidates to shortlist: these are decisions where the group has to live with the outcome, and where a narrow plurality winner will be relitigated within a fortnight.
It also produces a usable second place. When the winning option turns out to be blocked by budget or timing, a ranked result tells you immediately what to do instead. A dot vote leaves you re-running the whole exercise.
The cost is real and it is cognitive. Ranking five options against each other is a genuinely harder task than ticking the ones you like, and it takes people two to three minutes rather than ninety seconds. Explain it in one sentence, cap the list at five, and do not apologise for the extra time - it is buying you a decision that holds.
The two biases that ruin an open dot vote
Dot voting on a physical wall is sequential and visible, which introduces both of the classic voting biases at once. The first is anchoring: the earliest dots set a reference point for what a popular option looks like, and later voters calibrate against it rather than against their own judgement.
The second is the bandwagon effect. Once an option carries a visible cluster, adding a dot to it feels like joining something, and adding a dot to an empty note feels like wasting one. The result is a distribution that is far more concentrated than the group's actual preferences.
Seniority compounds both. If the head of department dots first and everyone watched them do it, you have not run a vote, you have run a nomination with confirmation attached.
The fix is simultaneity, not honour systems. Have everyone place dots at the same moment with their backs to each other, or collect votes digitally so nothing is visible until every ballot is in. Session Flo's live voting activities keep results hidden until the round closes for exactly this reason, and the shift in distribution when you do it is usually obvious the first time you compare.
The sequence that works: shortlist, then rank
Cluster before you vote
Spend four minutes merging duplicates. Voting on a list where the same idea appears three times splits its support three ways and can knock the group's actual favourite out of the shortlist entirely.
Set the dot budget from the list length
One dot per three options, with a minimum of three and a maximum of six. Eighteen options means six dots. Nine options means three. More dots than that and everyone approves of everything, which flattens the result into uselessness.
Dot silently and simultaneously, ninety seconds
No discussion during the window. Announce that all dots are spent at once and that nothing is visible until the timer ends. If you are on a wall, have people write their dots on a slip and place them all in the last ten seconds.
Take the top five, including ties
Read the spread, not just the order. If options four, five and six are within a dot or two of each other, carry all six forward. A shortlist of six is manageable; excluding the group's compromise candidate on a one-dot margin is not.
Discuss the shortlist for six minutes, not thirty
Give each shortlisted option ninety seconds of advocacy from whoever wants to speak for it. This is where the objections that dots cannot express finally surface, and it is the most valuable six minutes in the whole exercise.
Rank the shortlist and read the second place
Everyone orders the five. Show the winner and the runner-up together, and say out loud what the fallback is. Naming the second choice at the moment of decision saves a meeting later when the first choice runs into a constraint.
Running dot voting well
Top Tips
- Decide before the vote what the result authorises. 'The top five go to the discussion round' is a promise you can keep; 'we will do whatever wins' is one you probably cannot.
- Ban stacking multiple dots on one option unless you specifically want intensity, in which case say so. Silent stacking rules are the most common source of arguments about the count.
- If the room includes people who joined the topic today alongside people who have owned it for years, run the dot vote anonymously. Newcomers vote for what they actually think when nobody can see their dot land.
- Read the empty notes as well as the winners. An option with zero dots after a full round is a genuine signal, and it is the cheapest thing you will ever cut from a backlog.
- For remote or hybrid groups, never dot on a shared board where cursors are visible. Collect the votes and reveal once, or the people in the room will have effectively voted first.
- Record the full count, not just the shortlist. Six weeks later somebody will ask why an idea was dropped, and 'it got one dot out of forty-eight' ends that conversation immediately.
Running a ranked round well
Top Tips
- Explain the mechanism in one sentence: 'put these five in the order you would want them done, first to last.' Any longer and you will spend three minutes on method instead of decision.
- Cap the ballot at five options, six at the absolute limit. People rank their top two or three with care and then fill in the rest arbitrarily, which adds noise dressed up as data.
- Show the distribution, not only the winner. An option ranked first by four people and last by five is a warning, and it looks identical to a safe consensus choice if you only display the totals.
- Do not let people abstain from placing an option. A partial ballot quietly becomes a vote against everything left off it, which is not what the abstainer meant.
- Run the ranking after the advocacy round rather than before. The entire point of ranking is that it captures informed preference order, and preferences move once people hear why others disagree.
- If the top two finish within a couple of points, say so and pick on a stated tiebreak such as cost or speed to start. Pretending a two-point margin was decisive is how a decision gets reopened.
When neither method is the right tool
Voting of any kind assumes the options on the table are the right options. If the list was generated in ten minutes by the four loudest people, tidying it up with dots produces a well-organised version of a bad list. Generate silently first, individually and in writing, then vote on what comes back.
Voting is also wrong when the decision is not the group's to make. If a single accountable owner will decide regardless, say that plainly and use the session to gather input and surface objections instead. A vote whose result is then overruled costs more trust than never voting at all.
Finally, some decisions turn on information nobody in the room has. A group can vote confidently on which of five features to build and be completely wrong because none of them have seen the support data. When you notice that happening, stop the vote, name the missing input, and schedule the decision rather than manufacturing one on the spot.
Merge duplicates before voting
The same idea written three ways splits its own support and can miss the shortlist entirely. Four minutes of clustering protects the whole exercise.
Hide results until the round closes
Visible running totals turn a preference measurement into a popularity contest. Reveal once, all at the same time.
Treat close counts as ties
Anything inside a 20 percent spread is noise with a typical workshop group. Carry all of it into the ranked round.
Always name the runner-up
The second-ranked option is your fallback when the winner hits a constraint. Say it out loud at the moment of decision.
Frequently asked questions
How many dots should each person get?
Roughly one dot for every three options, with a floor of three and a ceiling of six. Twenty-one options means seven, which is above the ceiling, so give six. The reason for the ceiling is that once people can approve of a third of the list, almost everything gets some support and the result stops discriminating between options.
Can people put more than one dot on the same option?
Only if you want intensity rather than breadth, and only if you announce it before the round starts. Stacking lets a small group of enthusiasts push a single option to the top, which is sometimes exactly the signal you want and sometimes a distortion. Whichever you choose, say it out loud, because unstated stacking rules cause more arguments than any other part of the method.
Is ranked voting worth the extra time for a small team?
For a genuine decision, yes, even with five people. Ranking five options with five people takes about three minutes including the explanation, and it is the difference between a plurality winner and a choice everybody can live with. For a quick prioritisation that nobody will act on this quarter, dot voting alone is fine.
What do you do when a dot vote produces a three-way tie?
Take it as the intended result rather than a failure. A flat top means the group genuinely does not prefer one of those options over the others on the information available. Give each of the three ninety seconds of advocacy, then run a ranked round on those three alone. The order almost always resolves once people have heard the arguments.
How do you stop senior people dominating the vote?
Make the votes simultaneous and invisible until the round closes. Sequential open dotting lets the first voters anchor everyone after them, and seniority makes that effect much stronger. Digital collection solves it cleanly; on a physical wall, have everyone place their dots in the same ten-second window.
Does dot voting work for remote and hybrid groups?
It works well provided you collect votes through a tool rather than a shared whiteboard. On a shared board, remote participants see cursors moving and in-room participants often vote first, which reproduces every bias you were trying to avoid. Collect, close, then reveal the totals to everyone at the same moment.
Keep reading
- how to run dot voting step by step — The full mechanics of a dot voting round, including budgets and timings.
- running a live ranking exercise — Setting up and reading a ranked round with a group in the room.
- collect first, then vote — Why generating options silently before any vote improves what ends up on the ballot.
- the anchoring effect in group voting — The evidence on how early votes shape later ones.
- turning workshop output into decisions — What to do with a shortlist once the room has voted on it.