See who may leave while there is still time to talk.
Most HR reports name people after they resign. Retention Radar looks at records you already have - attendance, leave, time in the company, overtime, recent exits nearby - and explains why someone is flagged. It is a prompt for a conversation. It is not a reason to punish anyone.
An early warning you can explain, and a follow-up you can write down.
HR runs the check when they choose. Every score shows its reasons. Nothing is emailed on its own, and the score is not shown to the employee or their manager.
A band, plus how sure the picture is
The reasons in sentences, not a mystery number
A stay chat, a 1:1, a growth plan, or a workload look
You can dismiss a false alarm, with a reason
HR gets a list they can act on. Leadership sees the pattern. Managers are not handed a hit list.
This version is an HR screen. The point is a timely conversation, not a score that follows someone into a pay meeting without context.
A register of people to talk to, with the reason on the page.
You press Run when you want a fresh picture. The overview shows how many people sit in high, medium, and low, which departments are warmer, and which reasons come up most. Open a person and you see each factor: what was measured, how much it counted, and a sentence you can read. Then log what you will do - a stay conversation, a one-to-one, a pay review, a growth plan, a workload look - or dismiss the flag if it is wrong, with a reason and a date to hide it until.
- Search and filter by band, department, reason, and time in the company
- A trend of recent checks for that person, so one bad week is not the whole story
- Suggested things to discuss. Nothing is done automatically
Attrition after the fact is a history lesson. This is earlier than that.
Leadership wants to know if a team is tired, if strong people have not moved in years, or if exits are clustering. They do not want a secret score used to block a promotion. The page says so in plain language: an early warning for a conversation, not a basis for a harsh decision. Small teams are hidden on the department view so a group of three is not exposed. Individual mood answers are never used.
- Department view with tiny teams locked, numbers hidden
- You choose how much each factor counts, and the next run uses that
- Opening a person’s detail is recorded, and so is a download
They do not get this list. HR does the sensitive looking.
In this version, managers and employees cannot open Retention Radar. That is deliberate. A manager should not discover their team as a ranked risk board. If HR decides a conversation is warranted, they log it - a one-to-one, a workload review - and own the follow-up. The product will not email the manager a score overnight.
- No manager view and no employee view in this version
- No automatic messages, warnings, or performance plans
- Gender, age, religion, family, and medical details are never inputs
Why last quarter’s attrition report is already late.
Companies can list who resigned. They struggle to see, while the person is still at their desk, which patterns usually show up first - and to do something other than worry.
An attrition report after the quarter
The people on it have already left
A check of people who are still employed and not already exiting
A sealed “AI resignation score”
HR cannot say why, so they either ignore it or misuse it
Reasons, weights, and a confidence note when data is thin
Manager dashboards of individual mood
Breaks the promise that a mood answer was not a surveillance tool
Department mood totals only, and only when enough people answered
A spreadsheet of “flight risks” in someone’s laptop
No reason, no follow-up, no record of who opened it
A register, a logged next step, and a record when a person’s detail is opened
One bad week of lateness equals “they will quit”
Not enough history, and no comparison with their previous weeks
Attendance needs enough working days, and it is compared with the previous window
Automatic emails to managers naming “risky” people
The conversation starts as an accusation
No automatic messages. HR decides whether a conversation should happen
The “we should have talked sooner” moments this is for.
Each one is a workplace pattern the product can point at - and explain.
The attrition report arrives after the resignation.
The check looks for patterns while the person is still employed and not already in an exit.
A single “risk score” that nobody can explain in a 1:1.
Each factor shows its sentence, its weight, and how much it added. The pieces add up to the score.
Someone is flagged because a part of HR you do not use showed “zero”.
If there is no performance history, or not enough attendance yet, that factor is left out. It is not treated as a failing mark. Confidence drops instead.
A new joiner of three weeks looks “at risk” because there is no history.
People who are too new, already leaving, or inactive are left out of the check. You can set how new is too new.
Mood answers that were promised anonymous get reused to score a person.
Only a department’s overall mood shift is used, and only when enough people in that department answered. One person’s mood is never read.
A false alarm sits on someone’s name for a year.
Dismiss it with a reason of substance and a date. It stays out of the “needs attention” list until then. The score is still kept.
HR sees a red name and does not know what to do on Monday.
Log a follow-up: stay conversation, manager 1:1, pay review, growth plan, workload change, or something else you name. Mark it done.
Leadership wants a department story and HR has twenty individual guesses.
The overview has a department view, the common reasons, and risk by how long people have been with you.
A score gets used to deny leave, promotion, or a rating.
The screen states it is not a basis for adverse decisions. Nothing in the product auto-starts a warning or blocks a promotion.
Six reasons this is not a black-box “AI will resign” score.
If you cannot see why someone is on the list, you should not be having the conversation.
Earlier than the exit interview
The list is people who are still here. Someone already resigning is excluded, because that warning is no longer early.
Every point has a sentence
Tenure windows, worsening attendance, odd leave patterns, a rating drop, a long time without a move, overtime, a new manager, probation trouble, a team mood shift, nearby exits. You can read them.
Missing data is not a punishment
If you do not run performance reviews, or there are not enough attendance days yet, that factor is left out. The score uses what you actually have, and shows how complete the picture is.
Anonymous stays anonymous
Individual mood answers and private feedback are not inputs. Department mood is a team total, hidden when the group is too small.
HR can write down the next step
A flag without a follow-up is gossip. Stay chats, 1:1s, growth plans, and workload reviews live on the same page, with an owner and a due date.
You run it. It does not run you.
There is no overnight job and no automatic email. HR presses Run. Old results can be marked stale so you know when to run again.
What you can actually do.
Overview after you run a check
- Counts for high, medium, low, and not enough information, compared with the previous run
- Average score, when it was run, and how many people were included
- Department tiles. Very small departments stay locked with no numbers
- Which reasons are showing up most, and how risk sits by time in the company
- A line for who was left out: already resigning, too new, or in a department you excluded
The people list
- Score, band, confidence, top reasons, and whether the score moved since last time
- Filter by band, department, reason, tenure, and whether a follow-up is open
- Dismissed flags stay hidden unless you ask to see them
- Download the filtered list if you are allowed to export
Why this person
- Each factor in words: what was seen, how much it counted, and whether data was missing
- A trend across recent runs
- Conversation starters matched to the factors - suggestions only
- Opening the detail is recorded
What the check looks at
- Time in the company, including the windows when people often leave - around a year, around two years, and just after probation
- Attendance getting worse, and heavy overtime
- Short-notice or single-day leave, or a long stretch with no leave at all
- Latest performance rating and whether it dropped - only if you use performance reviews
- Months since a promotion, role, band, or pay change, with extra weight if a strong rating has gone nowhere
- A recent manager change, probation that was extended or is overdue, a shift in the department’s mood, and recent exits in the same team
Follow-ups and false alarms
- Log a stay conversation, a manager 1:1, a pay review, a growth plan, a workload change, or other
- Owner, due date, notes, and mark it done
- Dismiss a flag with a real reason and a date to suppress it
- Nothing is sent to the employee automatically
Settings HR can explain
- Turn the check on or off, and set how new is too new
- Choose how far back to look, how often a run is allowed, and how long old runs are kept
- Move the lines for medium and high
- Switch factors on or off and change their weights. At least three stay on
- Exclude departments, and set how big a team must be before it appears
- Changes apply the next time you run
Professional and Enterprise. Same radar. HR only.
Starter does not get a locked people-risk tile. Both higher plans get the same check, the same explanations, and the same follow-ups.
Starter
Retention Radar is not on Starter - on purpose.
- No locked “who might leave” tile
- Mood Pulse, if you are on a higher plan, remains the feeling check - this is a different page
- Move up when HR is ready to own an explainable people-risk list
Professional and Enterprise
Same Retention Radar on both plans.
- Run a check, read reasons, log a follow-up, dismiss a false alarm
- Department view with small teams hidden
- Weights and thresholds you can change
- HR only. Not shown to employees or managers in this version
What makes the approach different.
A conversation aid
The first thing on the page is that the score is not grounds for a harsh employment decision. The product will not start one for you.
You can see the maths in words
Weights are visible. If performance data is missing, it says so. You are not asked to trust a sealed number.
Private attributes stay out
Gender, age, religion, marital status, disability, medical, and family details are not factors. Anonymous feedback is not re-identified.
Small groups are not exposed
A department with too few people does not get a public risk number. The same idea applies to department mood: too few answers, and it is not used.
HR chooses when to look
There is a cooldown so a double-click does not fire two runs, and a note when results are getting old. There is no background send.
The sensitive views are recorded
Running a check, opening a person, changing settings, logging a follow-up, dismissing a flag, and downloading are written to the change record.
Retention Radar - questions buyers actually ask.
Stop learning someone was unhappy from a resignation email.
Book a short walkthrough: run a check, open one person, read the reasons, and log the conversation you would actually have. The disclaimer on the screen is part of the product: this is an aid, not a verdict.
Trusted HRMS for Modern Businesses
