Most searches for a 360 feedback template end with a document: a page of competencies, a five point scale, and two open questions. That part is not hard, and a usable version of it is below. The part that decides whether the exercise is worth running is everything around the document. Who rates whom, what each rater is told about who will read the answers, how the answers get gathered from nine people without twelve reminder emails, and what the subject actually receives at the end.
Get those wrong and the template still works perfectly. The output is just a set of polite fours that nobody learns anything from.
What a 360 template is made of
A 360 review differs from a manager review in one respect only: the same questions go to several groups of people who see different parts of the same work. The structure follows from that.
Rater groups. Self, manager, peers, and direct reports where they exist. Some organisations add a fifth group for clients or partners. Each group answers the same core questions, which is what makes comparison across groups possible, and this is the whole point of the format. A gap between how a person rates themselves and how their peers rate them is the most useful single output of the exercise.
Competency blocks. Four to six areas, each with three or four statements rated on a scale. Six areas at four statements is twenty four ratings, which takes a rater about eight minutes. Twelve areas takes half an hour and gets abandoned.
Behaviour statements rather than traits. "Explains a decision so the reasoning can be followed" is observable. "Is a strong communicator" is a personality judgement, and different raters will mean different things by it.
Two or three open questions. These carry most of the value and most of the risk, because free text is where a rater can be identified by phrasing alone.
A relationship question at the start. Which group the rater belongs to and roughly how long they have worked with the subject. Without it, the results cannot be split by group, and a split by group is the reason to run a 360 at all.
The template, written out
Rate each statement from 1 to 5, where 1 is rarely and 5 is consistently, with a sixth option for no basis to judge. Label every point on the scale rather than only the ends, and keep the no basis option available, because a rater forced to guess produces noise that looks like data.
Delivery
- Finishes what was agreed, in the time that was agreed
- Says early when something will be late, rather than at the deadline
- Keeps quality steady when the work is under pressure
- Picks the important work over the urgent work
Communication
- Explains a decision so the reasoning can be followed
- Writes so a reader outside the team can understand it
- Listens through an objection before answering it
- Gives bad news directly and without delay
Collaboration
- Offers help before being asked
- Disagrees in a way that keeps the discussion useful
- Shares information others need without being chased for it
- Gives credit accurately
Judgement and ownership
- Takes a decision with incomplete information rather than stalling
- Says plainly when something was a mistake
- Escalates at the right point, neither too early nor too late
- Changes position when the evidence changes
Developing others, for anyone with direct reports
- Gives feedback close to the event rather than saving it
- Creates room for people to take on work above their level
- Is available when someone is stuck
Open questions
- What should this person keep doing, and why does it matter?
- What is the one change that would have the largest effect on their work?
- What is something this person may not realise about how their work lands on others?
That is twenty ratings for an individual contributor, twenty three with direct reports, and three open questions. It fits on one screen per block and takes under ten minutes to complete honestly.
Choosing raters, and how many
Between five and eight raters per subject is the working range. Below four, one strong opinion dominates and group anonymity is impossible. Above ten, the aggregate stops moving while the effort of collecting and compiling keeps growing.
A defensible split for an individual contributor is the subject, the manager, four to five peers, and any regular internal collaborators outside the team. For someone who manages people, three or four direct reports as well.
The selection method matters more than the exact number. Let the subject nominate a list longer than what is needed, then have the manager or the coordinator finalise it, removing and adding names. Subject nomination alone produces a friendly panel. Manager selection alone makes the exercise feel like evidence gathering. The combination avoids both readings, and it takes one conversation.
Two rules save trouble later. Ask nobody to rate more than three subjects in a round, because the fourth set of twenty ratings gets answered on autopilot. And keep the same panel across rounds where possible, since a change in rater set looks exactly like a change in performance when the numbers are compared next year.
Privacy, and what each side actually sees
Here is where announcements and reality most often part company. Three arrangements are in common use, and the difference between them is worth stating to raters in plain words before they answer.
| Arrangement | What the subject receives | What the coordinator sees | Where it fails |
|---|---|---|---|
| Aggregated by group | Scores averaged per rater group, open text edited for phrasing | Everything, including who said what | Groups smaller than three make averages identifying |
| Fully anonymous | Scores and text with no group labels at all | Identities not collected | Cannot tell a peer view from a manager view, which removes the main insight |
| Attributed | Scores and comments with names attached | Everything | Ratings compress towards the top and open text turns bland |
Aggregated by group is the usual choice, and it needs a minimum group size to hold. Three responses per group is the common floor. If only two direct reports answered, fold them into the peer group or report them as a single unlabelled block, and say in advance that this is what happens so it does not look like a leak.
Free text is the harder half. Averages protect an identity; a sentence referring to a specific meeting does not. The coordinator either edits open comments into substance without the phrasing, which takes real time, or tells raters clearly that their wording will be passed on as written. Both are legitimate. What damages the programme is promising the first and doing the second.
One more line to include on the form: whether the manager sees the raw text, and whether any of this reaches a pay or promotion decision. If 360 results feed a pay decision, raters will work that out within one cycle and the ratings will rise for everyone. Development and pay are best kept in separate exercises with separate forms.
Collecting it without chasing nine people
The logistics are the reason so many 360 programmes run once and stop. One subject with seven raters is seven submissions. Ten subjects is seventy, arriving over two weeks, each belonging to a particular person and a particular rater group.
Three things have to be true of whatever is used to gather them.
The first is that each submission carries its subject and rater group as structured data, not in a free text field where somebody types a name three different ways. In practice this means one form per round with a question naming the subject, or a duplicated form per subject. Duplicating is cleaner to compile and only sensible if duplicating a form is quick.
The second is that progress is visible without opening every response. What the coordinator needs at a glance is which subjects are short of raters and which raters have not started, which is a status on each response rather than a row in a list. A tool that gives each response an owner and a stage makes this a glance instead of an audit, and the response management features matter far more here than the form editor does.
The third is that a reminder can go to the people who have not answered, and only to them. A reminder sent to the whole list annoys everyone who already did the work, and it is the single most common reason a second round has a worse response rate than the first. Sending to a filtered selection from the response list solves it.
Compiling is the last step and should be planned before collection starts. Export the ratings as columns, average them per group, and produce one document per subject. If the export does not come out in a shape that can be pivoted, this becomes a day of copying between sheets. Checking the export format on a test round of two people is ten minutes well spent, and the same shape of work shows up in other multi rater processes such as hiring panels.
What the subject receives, and the conversation after it
A compiled report that lands in an inbox with no conversation attached is the most common way a 360 programme quietly ends. The document is read once, the two lowest scores are remembered, and nothing changes.
The report itself should be short. Per competency block, the self score next to each group average, so the gaps are visible without arithmetic. Then the open text, grouped by question rather than by rater, which stops the reader from trying to reconstruct who wrote what. Four pages is plenty. A twenty page pack with charts for every statement buries the two findings that mattered.
Order matters more than completeness. Lead with the areas where every group scored high, because those are the things worth protecting and they are usually invisible to the person. Then the largest gap between the self rating and the group ratings, which is the most actionable item in the whole exercise. Blind spots, where others rate lower than the subject does, are where behaviour actually changes. Hidden strengths, where others rate higher, are where confidence changes.
Then hold a conversation within a week of sending the report, with the manager or a coach rather than the coordinator. The purpose of that hour is to leave with two commitments, not to relitigate individual scores. Anyone who arrives wanting to argue about a rating has usually received a report with too much detail and too little structure in it.
Last, plan the follow up before the round closes. A 360 with no second round is an opinion poll. A second round twelve months later, with the same questions and roughly the same panel, is the only way to tell whether the commitments held. Write the date in the calendar while the first round is still fresh, since the effort of restarting from nothing next year is what kills most programmes.
What to change first
Before writing any questions, decide two things and write them on the form: the minimum group size for reporting, and whether the results reach a pay decision. Those two sentences change the answers more than the question set does.
Then run one test round with two subjects and the full rater panel, all the way through to the compiled document. The problems will be in the collecting and the compiling, not the template, and they are cheap to fix at that size. A tool that tracks who has answered without opening each response makes the real rounds manageable, and Halict can be tried on a single round before committing to anything.
Q1. How many raters does a 360 review need?
Five to eight per subject works for most teams. Fewer than four makes group level anonymity impossible and lets one strong view dominate. More than ten adds collection and compiling work without changing the averages much, and it increases the chance of raters answering on autopilot.
Q2. Should 360 feedback be anonymous or attributed?
The common middle option is to report scores averaged by rater group with a minimum of three responses per group, so the subject learns how peers differ from managers without knowing who said what. Fully attributed feedback pushes ratings upward and makes open comments cautious.
Q3. Can the same template be used for managers and individual contributors?
Mostly yes. Keep the core competency blocks identical so results can be compared across the organisation, and add one block about developing others for anyone with direct reports. Changing the shared questions between roles makes any comparison across the company meaningless.
Q4. Should 360 results affect pay or promotion?
Keeping them out of pay decisions produces more honest answers. Once raters understand that their input moves someone's salary, scores compress towards the top within a single cycle. If the results must inform a decision, say so on the form rather than letting people discover it later.
Q5. How long does a round take to run?
Plan roughly four weeks: one week to confirm rater panels, two weeks open for answers, one week to compile and hold the conversations. Leaving the form open longer rarely raises the response rate, because the people who intend to answer do so in the first few days.