# The Future-Manager Diligence Standard, Eight Criteria to Pass, Hold, or Fail

*Grade what you know about your future manager on eight pass or fail criteria and reach a Pass, Hold, or Fail verdict that a second grader would match.*

- Canonical URL: https://www.refolk.ai/candidates/guides/future-manager-diligence-standard
- Pillar: Offers and negotiation
- Format: Standard
- Published: 2026-11-04
- Last reviewed: 2026-11-04
- Reading time: 22 min
- Keywords: how to vet a hiring manager before accepting an offer, backchannel a future manager, ask to meet the team before accepting offer, questions to ask former employees about a manager, red flags in a new manager before accepting a job, talk to future manager after offer

## Key takeaways

- A future-manager file passes only when it holds at least three sources who worked with the manager in the last three years, two of them independent of the employer and one a former direct report you found yourself.
- Gallup attributes 70% of the variance in team engagement to the manager, measured across business units, which justifies diligence but cannot grade one specific manager.
- About one in five established US LinkedIn users retro-edits a job they have already left, so a public-profile attrition count is a floor: a high count is strong evidence and a low count proves little.
- In Refolk's index, 12% of a 25-profile Engineering Manager sample held other titles such as CTO, so title search alone will miss or misplace members of a manager's team.
- Leverage peaks after the offer is made, and practitioner sources describe 48 to 72 hours as a commonly granted extension, so every Hold should map to one named call or extension request.
- Voluntary review-site ratings over-represent extreme opinions, so reviews corroborate a verdict but never carry one alone.

This standard is for anyone holding an offer who needs more than the hiring manager's own pitch to decide whether the person they would report to is someone they can work for. It grades the evidence you hold on that manager against eight pass or fail criteria, and it produces a Pass, Hold, or Fail verdict that two people reading the same file would agree on. When a criterion is unmet, it names the call or extension to ask for before the offer deadline.

## Why grade a future manager, and what can this standard prove?

You grade the manager because the manager is the biggest single reason one team differs from another, and the hiring pitch is the one source that has a stake in your answer. The standard checks whether your evidence is enough and independent enough. It does not predict how you personally will get on with someone.

The headline number deserves care. Gallup says managers account for 70% of the variance in team engagement. That figure rests on its Q12 meta-analysis of more than 180,000 manager-led teams. Beck and Harter's original wording was "at least 70% of the variance in employee engagement scores across business units". The unit in that sentence is the business unit, not the person.

> **Note:** What the 70% figure does and does not support
>
> The 70% figure explains differences between business units, so it justifies doing diligence on the manager. It cannot score one specific manager or predict your own experience. Only evidence about that manager can do that.

Candidates already behave as if the manager matters:

- In a UK survey of 2,000 employees, about a fifth said they would turn down an offer if they knew the new manager had a bad reputation. That survey is roughly twelve years old.
- In a LiveCareer poll of 1,000 US workers, 40% said they had quit a job because they distrusted their manager.
- In Bayt's MENA survey, 86.1% chose a supportive manager over faster pay growth.

Advice on backchannels is mostly written for employers checking candidates. Candidate checklists say "know your manager" but never say how much evidence is enough or what counts as a source. This standard fills that gap. It uses one term of art. A **reverse backchannel** is an informal reference check that the candidate runs on the manager, through people the employer did not hand over.

## What are the eight criteria and their pass lines?

Each criterion has a single pass line, stated as something you can point to in your file: a dated email, a meeting that took place, a named source with an exit date. If you cannot point to it, the criterion is unmet. Whether the manager seemed nice does not change that.

### 1. Manager of record confirmed in writing

- **Pass line:** A written message from the employer, dated after the offer, naming the one person you will report to on day one.
- **What it proves:** You are checking the right person.
- **How it lies:** The reply names "the team" or "for now" someone else, or the confirmation is dated before a reorganisation you have not heard about.

### 2. Post-offer one-to-one held

- **Pass line:** At least one conversation with the manager of record has taken place after the offer, not just been scheduled, and you asked every question on your list in it.
- **What it proves:** You have the manager's own account of the five behavior dimensions and of their seat, given once the selling pressure has eased.
- **How it lies:** The call turned into a sales pitch and your questions never got asked, or the person who joined was a peer or someone from human resources.

### 3. Independent source count

- **Pass line:** At least three counted sources who worked with the manager, covering at least two relationship types. At least two of them were not supplied or suggested by the employer, and at least one of those two is a former direct report you found yourself.
- **What it proves:** The picture does not depend on who the employer chose to put in front of you.
- **How it lies:** A "former report" the manager pointed you to counts as supplied, even if you made the contact yourself.

### 4. Source recency

- **Pass line:** Every counted source worked with the manager within the last three years, measured from the source's exit date or the end of their overlap with the manager.
- **What it proves:** You are hearing about the manager as they are now, not as they were.
- **How it lies:** A source who left long ago gives what one employer-side guide calls a "four-year-old answer". Log them as background only.

### 5. Behavior-specific answers on named dimensions

- **Pass line:** On each of five dimensions (feedback, credit, conflict and escalation, workload and boundaries, growth and exits), at least two counted sources each gave a specific instance: the situation, what the manager did, and what happened.
- **What it proves:** You have evidence you can compare across sources, not adjectives.
- **How it lies:** "Great to work with" scores zero. Gem warns that references come back positive by default, so a mixed reference is usually a bad signal.

### 6. Team attrition count from public profiles

- **Pass line:** A dated list of people whose public profiles show them reporting to this manager and leaving. Each entry is tagged confirmed or unconfirmed, and at least one source or the manager has explained every confirmed departure.
- **What it proves:** Whether people leave this manager, and why, in numbers that do not come from the employer.
- **How it lies:** Retro-edited titles and private profiles hide departures, so the count can only err low. See the attrition section.

### 7. Manager's seat stability

- **Pass line:** You know the manager's start date in their current role. You asked them directly whether their own role, scope, or reporting line is expected to change. No source contradicts the answer.
- **What it proves:** The person you are evaluating is likely to still be your manager after you arrive.
- **How it lies:** The manager is stable but their boss is on the way out. Ask about the line above them as well.

### 8. Search confidentiality kept

- **Pass line:** Nobody at your current employer was contacted or told. No current direct report of the manager was contacted cold. Every source was promised non-attribution, and you have quoted no source to the employer.
- **What it proves:** Your search, and the people who helped you, are protected.
- **How it lies:** A former report who still has close friends on the team passes your questions along. Keep questions about behavior, and never say who else you have spoken to.

## How many sources count, and which ones?

Three sources count if each worked with the manager in the last three years, at least two are independent of the employer, and at least one is a former direct report you found yourself. That puts the threshold at the top of Gem's range of 2 to 3 and borrows AVRA's rule that relationship types must be mixed.

The employer-side guides disagree on how many sources are enough. They agree on what makes a source worth hearing:

| Guide | Minimum sources | Required relationship types |
|---|---|---|
| Gem | 2 to 3 backchannel | Prior employers preferred |
| AVRA | 7 | 3 bosses, 2 peers, 2 subordinates |
| Bain Capital Ventures | Fewer but more relevant | Past managers |

Some rules carry over to the candidate side unchanged:

- Set a minimum source count and do not round up.
- Mix relationship types: report, peer, and skip-level where you can.
- Prefer people who have already left.
- Ask every source the same questions.

One rule flips. Employer guides warn against contacting the candidate's current employer because the candidate is the one at risk. When you check a manager, the people at risk are the manager's current reports. So your caution points at them, not at yourself.

#### Which sources count toward the threshold

Horizontal axis runs from Employer-supplied to Self-found. Vertical axis runs from Left 3+ years ago to Within 3 years.

| Quadrant | What it means |
| --- | --- |
| Recent but supplied | Counts toward the three, but not toward the two independent sources |
| Recent and self-found | Full weight, and the only kind that satisfies the former-report rule |
| Stale and supplied | Do not count it, log as context |
| Stale and self-found | Background only, ask them to introduce someone more recent |

*Only a source you found yourself who worked with the manager in the last three years counts at full weight toward criterion 3.*

### Why skip-level sources are harder to find

A skip-level is your manager's manager. In Refolk's index of professional profiles, the skip-level layer in engineering is much thinner than the manager layer, and it clusters in fewer places. That makes an independent skip-level source more expensive to find, and in a concentrated market it puts your confidentiality at more risk.

| Pool | Total in Refolk's index | Exact-title share (n=25) | Sample note |
|---|---|---|---|
| US Engineering Manager | 43,206 | 88% | 16% Bay Area |
| UK Engineering Manager | 10,940 | 88% | Meta top employer, 3 of 25 |
| US Director of Engineering | 22,742 | 92% | 44% Bay Area |

**1.9 - US Engineering Managers per Director of Engineering in Refolk's index**

Derived from 43,206 Engineering Managers and 22,742 Directors, so there are about half as many skip-level contacts and they are more concentrated.

The samples behind the share columns are small (25 profiles each), so read them as direction, not measurement. My practical reading is this. Do not hold up your file waiting for a skip-level source. Count one when you find one, and meet the threshold with former reports and peers.

## How do you read a team attrition count from public profiles?

List the people whose public profiles show them reporting to the manager and then leaving, date the list, and treat the number as a floor. A high count is strong evidence. A low count proves little.

No published procedure exists for this count and no error rate has been measured, so state your method in the file. The known blind spots all push the count the same way:

- **Retro-edits.** NBER working paper 35546, which covers 29.4 million US profiles, found that roughly one in five established US LinkedIn users goes back and edits the title or description of a job they have already left. For users under 30 the rate is 38.1%. The edit typically lands about four years after the job ended.
- **Private profiles and lag.** The same research treats profile counts as an approximate lower bound, because some users keep their profiles private and some update late.
- **Title drift.** In Refolk's index, 22 of 25 sampled Engineering Manager profiles held the exact title. The other three held titles such as CTO and Managing Director Engineering.

**12% - Share of a 25-profile Engineering Manager sample in Refolk's index holding a different title**

Matching on title alone will miss members of a manager's team or put them in the wrong place.

How to run the count:

1. Collect public profiles that list the employer, the manager's function or team, and dates that overlap the manager's tenure.
2. Match people to the manager by org-chart evidence (team names, a named manager, shared projects), not by title alone.
3. Record each person's exit date and where they went.
4. Tag an entry **confirmed** when two sources agree on the reporting line, for example the profile plus a call. Otherwise tag it **unconfirmed**.
5. Date the list, and bring every confirmed departure to at least one call to get the reason.

This standard sets no numeric attrition threshold, and that is deliberate. Without a measured error rate and the team's size, any cutoff would be invented. What gets graded is whether the confirmed departures are explained.

> A low attrition count proves little. A high one, read as a floor, is hard to explain away.

## The procedure: from offer to verdict

The procedure has eight steps. It runs in about a week if you start the day the offer arrives, and the steps that involve the employer go first because they take the longest. The order follows Ask the Headhunter, which says to wait until the offer is in hand, not Gem's employer-side advice to start backchannels early. The exception is the public-profile count. It is solitary and invisible, so you can start it during the final round.

#### Offer to verdict, by who you are dealing with

1. **Recruiter** - Written manager of record and deadline, about 1 day plus 1 to 2 days
2. **Hiring manager** - One-to-one and team meeting, 2 to 5 days
3. **Alone** - Attrition count from public profiles, starting with thirty minutes
4. **Sources** - Three or more calls of 15 to 30 minutes, same questions each time
5. **Grader** - Eight criteria, Fail triggers, verdict

*Lock in the employer-facing steps first, because leverage peaks after the offer and the clock is already running.*

#### Future-manager diligence, offer to verdict

1. **Confirm the manager of record in writing** - Email the recruiter for the name of your day-one manager. Done when a written reply dated after the offer names one person.
2. **Lock the deadline and ask for extra time early** - Get the deadline in writing and ask for any extension now, tying a request for a full week to a specific reason. Done when the date is in writing.
3. **Request the post-offer one-to-one and a team meeting** - Ask the hiring manager directly, not human resources. Done when the one-to-one has been held and every question was asked.
4. **Count attrition from public profiles** - Working alone, list everyone who reported to the manager and left, with dates. Done when the list is dated and every entry is tagged confirmed or unconfirmed.
5. **Source independent contacts** - Find at least three people who worked with the manager in the last three years. Done when two are independent and one is a self-found former direct report.
6. **Run the calls with one question set** - Hold 15 to 30 minute calls using identical questions on all five dimensions. Done when the notes record instances, not adjectives.
7. **Corroborate with review sites** - Log relevant reviews against criteria as corroboration only. Done when no review is counted as a source.
8. **Grade the file Pass, Hold, or Fail** - Mark each criterion, check the Fail triggers, apply the verdict rule. Done when the verdict and the action for every unmet criterion are written down.

### Deadlines and extensions

Exploding offers can carry deadlines of 24 to 72 hours, so ask about the deadline the same day the offer arrives. Practitioner sources describe 48 to 72 hours as a commonly granted extension and say that a full week needs a specific reason behind it. Notre Dame's advice is to ask a week or more before the deadline. No survey measures how often employers grant post-offer meetings or extensions, so ask early and get the answer in writing.

> **Rule:** The one-to-one is with the manager of record, arranged directly
>
> Ask the named manager yourself, not human resources. A meeting with anyone else, or one that never gets to your questions, does not satisfy criterion 2.

**Manager of record and deadline, then the one-to-one request**

```
To the recruiter:
Thank you for the offer. I am excited about it. Before I decide, could you confirm two things in writing?
1. The name of the person I will report to on my first day.
2. The decision deadline. If any extension is possible, I would like to ask for [48 hours / until DATE] so I can finish a conversation with my future manager that I am arranging now.

To the hiring manager:
Thank you again. Before I sign, I would value 30 minutes with you on how you run the team, and, if possible, 20 minutes with two or three people I would work with. Would [DAY or DAY] work?
```

*Send the first message to the recruiter on offer day and the second to the hiring manager once the name is confirmed. Replace the bracketed parts.*

### Preparing the one-to-one

In the one-to-one, ask the same five dimension questions you will ask your sources, then the seat-stability question. If you applied through [Refolk](/candidates), the fit score on this posting already lists the gaps between your history and the role. Take the two largest gaps into the meeting and ask how the manager would handle them in your first six months. The answer gives you a specific data point on growth, from the one person whose answer you can check against the calls.

**Source call script, identical for every source**

```
Opening: I am considering an offer to work for [MANAGER]. Nothing you say will be attributed to you or repeated to anyone at [EMPLOYER]. I will not mention that we spoke.
Context: When did you work with them, in what role, and when did that end?
1. Feedback: Tell me about the last time they gave you or a teammate hard feedback. What did they say and what happened next?
2. Credit: Tell me about a piece of work that went well. Who presented it upward?
3. Conflict and escalation: Tell me about a time the team disagreed with them or with another team. What did they do?
4. Workload and boundaries: Tell me about the busiest stretch you saw. How did they handle hours and priorities?
5. Growth and exits: Tell me about someone who was promoted or left. How did they handle it?
Attrition: I count [N] people who left the team between [DATES]. Is that right, and do you know why they left?
Seat: Have you heard of any change coming to their role or their manager's?
Close: Is there anything you would want to know if you were me? Is there anyone more recent I should speak to?
```

*Read the opening word for word. Ask every question in this order on every call, and write down instances, not adjectives.*

## How does this go wrong? Failure modes and false positives

Most bad verdicts come from evidence that looks independent, recent, or complete and is not. Each failure below names the false positive and the check that catches it. Two graders will only agree if both run these checks.

1. **Reorg false positive (criterion 1).** The person who interviewed you is not the person you will report to. *Check:* written confirmation dated after the offer, naming one person.
2. **Supplied "independent" source (criterion 3).** The former report was suggested by the manager or the recruiter, so the employer picked them. *Check:* you found the person yourself through your public-profile count or your own network.
3. **Stale source (criterion 4).** The source left more than three years ago and describes a manager who may no longer exist. *Check:* the source's exit date, written in the file.
4. **Attrition undercount (criterion 6).** About 20% of established users retro-edit past titles, and private profiles stay invisible, so the count reads low. *Check:* count only reporting lines that two sources confirm, and treat the total as a floor, never a clean bill of health.
5. **Title mismatch (criterion 6).** In Refolk's index, 12% of a 25-profile Engineering Manager sample held titles other than Engineering Manager. *Check:* match on org-chart evidence, not titles.
6. **Generic praise (criterion 5).** "Really supportive" with no instance attached. Because references default to positive, an answer that comes back mixed is the meaningful signal. *Check:* if an answer has no situation, action, and outcome, it scores zero on that dimension.
7. **Review-site extremes.** Voluntary reviews over-represent strong opinions. Research using Glassdoor data found incentivized ratings less extreme than voluntary ones, and Brandes, Godes and Mayzlin classified 67 percent of reviews in their sample as extreme. How sites attribute a review to a specific team or manager is not established. *Check:* reviews corroborate a pattern already in your calls. They never count as a source.
8. **Exposed current employee (criterion 8).** A current report talks to you and the conversation becomes known. *Check:* prefer people who have already left, and never cold-contact current reports.

> **Watch out:** Current reports carry the risk, not you
>
> The NLRB protects employees sharing information about working conditions with co-workers, but one person griping on their own is not "concerted activity". A current report who briefs you is therefore not clearly protected. Talk to people who have left.

### What the legal position does and does not cover

In the US, the NLRB says employees have the right to share information about pay, benefits, and working conditions with co-workers. The NLRA applies to non-supervisory employees, and a single person complaining is not concerted activity. It is unverified whether non-solicitation clauses reach a candidate who is simply asking questions. If a source raises that worry, do not argue the point. Thank them and move on to someone who has already left.

## How do you grade the file: Pass, Hold, or Fail?

Pass needs all eight criteria met and no negative pattern described by two independent sources. Fail needs one of three named triggers. Everything else is a Hold, and every Hold comes with a named fix. Missing evidence never produces a Fail by itself. Only evidence you actually have can do that.

### The verdict rule

- **Fail** if any of these is true:
  - The employer declines in writing to name the manager of record.
  - The employer declines a post-offer conversation with the manager after you asked directly.
  - Two or more independent counted sources each describe the same negative behavior with a specific instance.
- **Pass** if all eight criteria are met and no Fail trigger applies.
- **Hold** in every other case.

#### How much each kind of evidence weighs

1. **Written employer record** - Manager of record and deadline, dated after the offer
2. **Direct contact** - The one-to-one and team meeting with the manager of record
3. **Independent recent sources** - Self-found former reports and peers within three years
4. **Public-profile attrition count** - A dated floor, explained by sources
5. **Review sites** - Corroboration only, never counted as a source

*Evidence carries less weight the further down this stack it sits, and the bottom layer can support a verdict but never decide one.*

### The Hold path

Each unmet criterion maps to one action. If that action cannot finish before the deadline, ask for 48 to 72 hours and name the action as your reason.

- **Criterion 1 unmet:** Email the recruiter asking for the manager's name in writing. About one day.
- **Criterion 2 unmet:** Ask the manager directly for 30 minutes and ask the recruiter for an extension tied to that meeting.
- **Criterion 3 or 4 unmet:** Find one more self-found former report from your attrition list who left within three years. One 15 to 30 minute call.
- **Criterion 5 unmet:** Send a source a short follow-up asking for one instance on the missing dimension.
- **Criterion 6 unmet:** Ask one source to confirm reporting lines for your unconfirmed entries, then get reasons for the confirmed ones.
- **Criterion 7 unmet:** Email the manager one question about planned changes to their role or reporting line.
- **Criterion 8 breached:** You cannot repair this. Stop contacting anyone connected to that team and grade on what you already have. The verdict cannot be Pass.

If the deadline arrives and the file is still on Hold, the verdict stays Hold. Signing is still your choice, but write it down as a Hold you accepted, not a Pass.

#### Before you write the verdict

- [ ] The file contains a written message, dated after the offer, naming one manager of record.
- [ ] The one-to-one with that manager has taken place and every question on the list was asked.
- [ ] At least three counted sources are listed, each with an exit date within three years.
- [ ] At least two sources are independent and at least one is a former direct report you found yourself.
- [ ] Every source was asked the same questions, and each of the five dimensions has at least two specific instances.
- [ ] The attrition list is dated, each entry is tagged confirmed or unconfirmed, and every confirmed departure has a reason.
- [ ] The manager's start date in role is recorded and the seat-stability question was asked.
- [ ] No review is counted as a source.
- [ ] No current direct report was contacted cold and no source was quoted to the employer.
- [ ] The three Fail triggers were checked and the result is written next to the verdict.

## Keeping the standard current, and what to do after the verdict

Re-check the mechanisms behind the thresholds, not just the values, and treat the verdict as the start of a decision rather than the decision itself. If a team adopts this as policy, version the thresholds so files graded months apart stay comparable.

What to re-check, and how:

- **Extension norms.** The 48 to 72 hour figure comes from practitioners, not a survey. Ask your recruiter what is normal and record their answer.
- **Profile edit behavior.** The retro-edit rate comes from one working paper. If a newer study changes it, the direction of the bias (counts err low) is what matters for grading.
- **Review-site attribution.** It is not established how sites tie reviews to a team. Check the site's own help pages before relying on any filter by team.
- **Source thresholds.** If you raise the count toward AVRA's seven for senior roles, write the new number into the standard before you grade, not after.

After the verdict:

- **Pass:** Sign, and keep the file. The five dimension answers make a useful baseline for your first months.
- **Hold accepted at deadline:** Sign knowing which criterion is open, and put that question to the manager in your first week.
- **Fail:** Decline, or ask whether a different reporting line is possible. That is a negotiation in its own right. If you go back to searching, Refolk tailors your resume and cover letter to each new posting from the history you already have, so a declined offer does not mean starting over.

## Frequently asked questions

### Is it OK to ask to meet the team before accepting an offer?

Yes, and the period after the offer is the best time to ask. Ask the Headhunter notes that an employer is more likely to consent to these meetings once it has committed to a job offer, and Ask a Manager calls it reasonable to want to meet more than the handful of people who interviewed you. Ask the hiring manager directly instead of going through human resources, and ask early enough to leave room before the deadline.

### How do I backchannel a future manager without the company finding out?

Talk mainly to people who have already left. Find them yourself through public profiles, promise them you will not attribute anything they say, and never quote a source back to the employer. Do not cold-contact the manager's current direct reports. They carry the risk, because one employee griping on their own is not clearly protected concerted activity under US labor law. Keep anyone at your current employer out of it completely.

### What questions should I ask former employees about a manager?

Ask for instances, not opinions. Ask every source the same questions on five dimensions: feedback, credit, conflict and escalation, workload and boundaries, and growth and exits. Phrase each one as "tell me about the last time" so the answer has to describe what happened, what the manager did, and how it turned out. Generic praise scores zero. A mixed answer is usually a bad sign, because references default to positive.

### What are red flags in a new manager before accepting a job?

Under this standard there are three hard red flags. The employer will not name your manager in writing. The manager will not meet you after the offer when you ask directly. Or two independent sources each describe the same negative behavior with a specific example. A cluster of confirmed departures that nobody can explain, or a manager whose own role is about to change, is a reason to Hold and ask more questions.

### How much extra time can I ask for to finish manager diligence?

Practitioner sources describe 48 to 72 hours as a commonly granted extension. If you need a full week, tie the request to a specific reason, such as a meeting with the manager that is already scheduled. No survey measures how often employers grant extensions. Ask early, ideally well before the deadline, and get the new date in writing. Exploding offers can carry deadlines of 24 to 72 hours, so request the extension on the same day.

### Should I trust Glassdoor reviews about a specific manager?

Use them only to back up what your sources tell you. Research using Glassdoor data found that voluntary ratings are more extreme than incentivized ones, because people with moderate views tend not to post. It is not established how review sites attribute a review to a particular team or manager. A review can support a pattern your sources already describe, but it cannot be one of your sources and cannot trigger a verdict alone.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/future-manager-diligence-standard*