Setup Considerations: anonymity and visibility

What this achieves

Feedback is the most socially delicate thing in the product, and every decision about who reads it is settled at the moment somebody writes rather than afterwards. This page covers those decisions while they are still decisions, so that what you promise your people matches what the screens actually do.

Decision 1: what you tell people about anonymity

Humavera offers anonymity in two different places, and they behave differently. Say the right thing about each.

Where	What the product does
Continuous feedback	Send anonymously is chosen per note by the person writing it. The recipient sees Anonymous instead of a name, and so does a manager reading the team feed. The author still sees the note in their own Given list.
Feedback cycles	Anonymous Responses is set on the cycle, before it launches. Reviewers are told on the form that the subject will see ratings and comments but not their name.

Neither is chosen by the person receiving the feedback, and neither can be changed after the fact.

The honest sentence to give your people is that anonymity hides a name from the reader. It does not make a note unattributable in a team of four, and it does not stop a comment describing a piece of work only one person did from identifying its author.

Example: an anonymous note to Daniel Okonkwo about last week’s forecast meeting narrows to whoever was in the room. Saying “anonymous” and meaning “nobody will ever know” is a promise the product has not made.

Decision 2: the default visibility, and whether you say anything about it

Continuous feedback carries a Visibility of Private, Manager visible, or Public, chosen per note. It opens on Private.

Option	Description
Private	The narrowest setting. Marked Private in the feed.
Manager visible	Marked Manager visible in the feed.
Public	The widest setting. Carries no marker in the feed.

The default is the setting most feedback will actually carry, because most people leave defaults alone. Decide whether your guidance to employees says anything about this at all — and if it does, be clear that changing it from Private puts a note in front of somebody’s manager, which changes what the note is.

Example: at HC Corp, a Constructive note to Tom Hargreaves set to Manager visible is a criticism his manager reads. The same words set to Private are a conversation between two people. The switch is one click and the difference is not small.

Decision 3: who may request feedback about whom

The request screen lets anyone request feedback about Yourself or about Someone else. There is no restriction in the interface on who may be named as the target.

That is a policy question the product leaves to you. Requesting feedback about a colleague is legitimate — a manager gathering input before a review, a project lead after a piece of work. Done quietly and at volume, it is something else.

Settle three things and put them in writing:

Decision	Why it has to be now
Whether requesting feedback about a third party needs the person to know	Deciding after somebody objects means deciding it about a specific incident.
Who is expected to respond to requests, and how often	A reviewer asked constantly stops writing anything considered.
What happens when somebody declines	The product notifies the requester and does not keep the reason. If a reason matters to you, it has to travel another way.

Decision 4: what nobody can undo

This is the shortest list on the page and the most important one.

Action	What cannot be undone
Sending continuous feedback	There is no edit and no withdraw. The note stands as written, at the visibility chosen.
Submitting a cycle review	The interface states responses cannot be edited once submitted.
Completing a check-in	Notes, agenda, mood, and action items are locked. Action-item completion can still be toggled.
Launching a cycle	Drafts cannot be re-edited, and the target scope is locked from creation.

Tell people this before they write, not in a support article afterwards. “Read it back before you send it” is the whole of the training this needs.

Decision 5: what a check-in record shows to whom

A one-to-one record is not one document with one audience.

Field	Who sees it
Agenda	The manager and the employee. Seeded before the meeting, so the employee reads it in advance.
Manager Notes	Does not appear on the employee’s own view of the check-in.
Employee Notes	Shared. The employee writes here on their own screen and the manager reads it.
Mood	The manager and the employee.
Action items	The manager and the employee.
AI preparation	The manager. Not shown to the employee.

This differs from a career conversation note, which the employee reads in full apart from AI preparation. Two adjacent records, two rules. Managers need to know which one they are writing in.

Decision 6: the smallest group you will report on

Feedback reporting carries the same arithmetic as survey reporting, and the same rule should govern both.

The product’s own analytics flag participants with fewer than three submitted responses as low participation. That is a floor for reading an individual’s result, not a threshold for publishing a group breakdown.

Example: HC Corp’s Engineering department at 34 supports a departmental average. Finance at 9 and People at 6 do not — a departmental score across six people, next to the department’s name, is close to publishing those six people’s ratings. The rule you set for surveys applies here unchanged.

What is expensive to change later

Decision	Cost of changing later
Anonymity on a cycle, once launched	Cannot be changed. Reviewers were told one thing on the form they filled in.
A cycle’s target scope	Cannot be changed. Cancel and recreate, and lose the responses.
What you told people about who reads their feedback	Very high. A correction here reads as having been misled, and people stop writing anything candid.
The default visibility guidance	Low to publish, high to reverse once a year of notes exists under the old guidance.

What happens next

Write the guidance down before you launch a cycle or encourage people to start giving feedback, and make it say the same thing the screens say. Then build the templates and design the cycle, where the reviewer mix and the peer minimum turn these promises into something the arithmetic can actually support.