A confidence vote is a quick, simultaneous vote where each team member shows how confident they are, usually on a 1 to 5 scale ("fist of five"), that the team can meet a commitment such as the sprint goal. If most hands show 3 or more, you proceed. Any 1 or 2 means you stop and find out why before committing.
There are two different confidence votes, and teams often mix them up:
| Sprint confidence vote (fist of five) | Per-estimate confidence | |
|---|---|---|
| Question | "How confident are you we'll meet this sprint goal?" | "How sure are you about this estimate?" |
| When | End of Sprint Planning, sometimes mid-sprint | Every planning poker vote |
| Scale | 1 to 5 fingers | Usually a percentage or low/medium/high |
| What it catches | Overcommitment, capacity gaps, dependencies, morale | Guessing hidden inside an apparent consensus |
Both are covered below.
The sprint confidence vote (fist of five)
The Scrum Guide asks the team to craft a Sprint Goal and select work in Sprint Planning. It doesn't say how to check that people believe in it. A fist-of-five confidence vote is the most common lightweight way to do that.
The scale
Read this aloud the first few times so everyone uses the same scale:
| Fingers | Meaning | What the voter is saying |
|---|---|---|
| 5 | Very confident | "We'll make it. I'd bet on it." |
| 4 | Confident | "We'll make it unless something unusual happens." |
| 3 | OK | "Achievable, but I have a specific concern I can live with." |
| 2 | Doubtful | "I have a concern that I think will cause us to miss." |
| 1 | No confidence | "I don't think this sprint is realistic as planned." |
How to run it (about five minutes)
- State exactly what's being voted on. "Confidence that we'll meet the Sprint Goal, checkout supports saved cards, with the 34 points we selected."
- Everyone votes at the same time. In a room, on "three, two, one, show." Remotely, use a tool that hides votes until reveal, or have everyone type a number in chat and press Enter together. Sequential voting is just anchoring with extra steps.
- Read the result aloud. "Two 5s, three 4s, one 3, one 2."
- Talk to the lowest votes first (see next section).
- Act, then re-vote if you changed the plan.
What to do with 1s and 2s
A 1 or a 2 isn't a veto, and it isn't negativity. It's someone telling you about a risk the plan doesn't account for. Use a script like this:
"Thanks, Jun. A 2 is exactly what this is for. What would need to change for it to be a 4?"
The answer is usually one of these, and each has a standard response:
| What the low voter says | Typical response |
|---|---|
| "We're overcommitted, I'm out three days" | Remove or shrink the lowest-priority item; re-check capacity |
| "Story X depends on the platform team and they haven't confirmed" | Move X out, or replace it with a spike plus a dependency check-in |
| "I don't understand story Y well enough" | Five-minute clarification now, or swap Y for a ready story |
| "The goal is vague; I don't know what done looks like" | Rewrite the Sprint Goal in one testable sentence |
| "We said this last sprint too and missed" | Plan to recent velocity, not hoped-for velocity (try the sprint velocity calculator) |
| "Tech debt in that module will slow everything" | Make the debt visible as work; see estimating technical debt |
Rules of thumb for deciding:
- Any 1: don't start until you've talked it through. One person with no confidence often knows something specific.
- Two or more 2s, or the average is under 3: change the plan (usually scope), then re-vote.
- All 3s and above: proceed, and note the concerns behind any 3s so you can watch for them at the Daily Scrum.
- A wide split (5s and 2s): an information gap. The people voting 5 and 2 are looking at different facts. Get both to explain before changing anything.
Never average away a single low vote. "We're at 3.8, good enough" can hide the one person who knows the database migration won't work.
Common mistakes
- Making it a loyalty test. If a 2 gets a sigh from the manager, everyone votes 4 forever. The team's coach or Scrum Master should protect the low voters, visibly.
- Voting on the backlog instead of the goal. Confidence in "all 14 tickets" is a different question from confidence in the outcome.
- Never re-voting. If you cut scope because of the 2s, re-vote. The new number shows whether the change actually helped.
- Ignoring the trend. If confidence slides from 4s to 3s over several sprints, something's building up (often debt, burnout, or unclear priorities). Bring it to the retrospective.
Per-estimate confidence in planning poker
The sprint vote checks the whole plan. Per-estimate confidence checks each number inside it. Three people can vote 5 for very different reasons:
- The developer who built the same feature last quarter: "Definitely a 5."
- The backend developer: "API's easy, not sure about the front end. 5, I guess."
- The new hire: "Everyone else looked like 5."
The cards all match, but the certainty doesn't. A confidence rating next to each vote makes the difference visible.
Lightweight options
- Traffic light after reveal: green (done this before), yellow (some unknowns), red (guessing).
- Thumbs: up, sideways, down, all shown at the same moment.
- Percentage: 0 to 100%, attached to each vote.
What to do with low per-estimate confidence
| Pattern | Meaning | Action |
|---|---|---|
| Same estimate, high confidence | Real consensus | Accept |
| Same estimate, several low-confidence votes | Pseudo-consensus | Ask "what would make you more sure?"; clarify or spike |
| Spread estimates, high confidence on both ends | Different understandings of scope | High and low voters explain; re-vote |
| Low confidence across the board | Story isn't ready | Spike (timeboxed research) or send back to refinement |
| One low-confidence voter, everyone else confident | Knowledge gap, often a newer person | Accept, and pair them with someone confident on it |
Use it where it pays: stories of 5 points or more, unfamiliar tech, or fuzzy requirements. Skip it for routine 1 and 2 point work.
How Alignlee handles confidence (and the math)
Alignlee builds in both kinds of confidence vote.
Per-vote confidence slider. Every planning poker vote carries a confidence from 0% to 100% (default 50%). When votes reveal, Alignlee shows a confidence-weighted estimate alongside the individual votes. Each vote's weight is:
weight = 2^((confidence − 50) / 50)
| Confidence | Weight |
|---|---|
| 0% | 0.5x |
| 25% | about 0.71x |
| 50% | 1x |
| 75% | about 1.41x |
| 100% | 2x |
So a fully confident vote counts four times as much as a fully unsure one. No vote is ever ignored, and "?" votes are left out of the number. T-shirt sizes are mapped to numbers for the calculation (XS=1, S=2, M=3, L=5, XL=8, XXL=13).
Worked example. Four voters on a Fibonacci deck:
| Voter | Vote | Confidence | Weight | Vote x weight |
|---|---|---|---|---|
| Ana | 3 | 90% | 1.74 | 5.22 |
| Ben | 5 | 50% | 1.00 | 5.00 |
| Chloe | 8 | 20% | 0.66 | 5.28 |
| Dev | 5 | 70% | 1.32 | 6.60 |
| Total | 4.72 | 22.10 |
Weighted estimate = 22.10 / 4.72 ≈ 4.7. A plain average would be 21 / 4 = 5.25. Chloe's unsure 8 pulls less, and Ana's confident 3 pulls more.
How to use the number: it's a discussion aid, not an answer. Here it says "the confident people are leaning toward 3 to 5." The facilitator should still ask Chloe what made her think 8, because a low-confidence high vote is often the first sign of a risk nobody else has spotted. Then the team picks a card, probably 5.
Confidence Pulse mode. This is a separate room mode for the sprint confidence vote. Everyone rates confidence 1 to 5, votes stay hidden until reveal, and the team sees the average. Use it at the end of Sprint Planning as described above, and still talk to the 1s and 2s by name.
Votes are shown with names after reveal (there's no anonymous mode), so the psychological-safety points above apply.
Confidence voting outside planning
The same vote works anywhere a group commits to something:
- Release readiness: "How confident are you this is ready to ship Thursday?"
- Architecture decisions: vote after the proposal, before building.
- Retro action items: "How confident are you we'll actually do this?" Low numbers mean the action is too big.
For decisions more broadly, including consent and disagree-and-commit, see remote team decision-making. For keeping planning sessions focused, see sprint planning best practices and how to facilitate planning poker.
FAQ
What is a confidence vote in Scrum?
A confidence vote is a quick, simultaneous vote at the end of Sprint Planning where each team member rates, usually on 1 to 5, how confident they are that the team will meet the Sprint Goal. Low votes are discussed before the team commits. It's a common team practice, not part of the Scrum Guide itself.
How does fist of five work?
On a count of three, everyone holds up one to five fingers at once: 5 means fully confident, 3 means OK with a concern, 1 means not achievable as planned. The facilitator reads the result, asks the lowest voters what would raise their vote, adjusts the plan if needed, and re-votes.
What should you do if someone votes 1 or 2 in a sprint confidence vote?
Thank them and ask, "What would need to change for this to be a 4?" Usually the answer points to overcommitment, an unconfirmed dependency, an unclear story, or a vague goal. Fix that (often by cutting scope) and re-vote. Don't average a single low vote away.
What is a good sprint confidence vote score?
Most teams proceed when everyone is at 3 or above and treat any 1, or an average under 3, as a signal to change the plan. The exact threshold matters less than the habit of discussing every low vote and tracking the trend across sprints.
How does confidence voting measure team readiness?
It asks the people doing the work whether the plan is realistic, which surfaces capacity gaps, dependencies, and unclear stories that story point totals can't show. Repeating it every sprint gives you a trend: steadily falling confidence is an early warning worth taking to the retrospective.