# The Shared-Credit Resume Line, Scored to Claim, Split, or Scope

*Score any team-built resume line on six dimensions, get a Claim, Split, or Scope call with its opening sentence, and know when the resume itself needs fixing.*

- Canonical URL: https://www.refolk.ai/candidates/guides/shared-credit-claim-split-scope
- Pillar: Interviewing
- Format: Framework
- Published: 2026-10-28
- Last reviewed: 2026-10-28
- Reading time: 22 min
- Keywords: I vs we in a job interview, resume says led but I was part of a team, what was your role in the project interview answer, how to explain team project contribution in interview, taking credit for team work interview, overstated resume line interview follow-up

## Key takeaways

- Every team-built resume line can be scored 0 to 2 on six dimensions (decision ownership, execution share, follow-up depth, verification alignment, scorer format, and employer emphasis), and the total out of 12 gives a Claim, Split, or Scope call.
- Standard HR verification confirms only title and dates, so "led", team size, and metrics can be contradicted only by a former manager or peer speaking informally, and that is exactly where overstated lines get caught.
- In Refolk's index, US profiles listing Behavioral Interviewing outnumber those listing Structured Interviewing 286x (13,743 to 48), which suggests many interviewers probe ad hoc rather than from a fixed probe list.
- A Scope opening that names the effort's size and your slice is not a hedge: it answers "What was your role?" before the interviewer asks it, and DDI tells interviewers to ask exactly that whenever they hear "we".
- If the verb on the page says "Led" but the scored call is Split or Scope, fix the resume line, not the spoken answer, because no framing in the room can repair a page that overstates.
- The self-promotion backlash research shows a larger penalty on hireability (d = -0.58) than on likability (d = -0.19) and concentrates it in explicit dominance, so it argues for naming concrete actions, not for avoiding "I" altogether.

This guide is for anyone preparing for interviews with a resume full of work they did as part of a team. It gives you a scoring sheet for each of those lines. You score six dimensions, and the total tells you whether to say "I" (Claim), separate the team's work from yours (Split), or open by naming how big the effort was and which slice was yours (Scope). Each call comes with an opening sentence. When the problem is the page rather than the answer, the sheet routes the line back to a resume edit.

## Why "say I, not we" fails as a line-by-line rule

"Say I, not we" tells you what pronoun employers want to hear. It does not tell you which lines have earned the pronoun. That depends on what you actually owned, who can check it, and who is doing the grading.

The blanket rule has real backing. Every employer and university source in my research points the same way. The coaching world pushes back, but with less to show for it.

| Source | Type | Instruction |
|---|---|---|
| Amazon | Employer | "I" statements |
| McKinsey | Employer | "I" (YOU, not team) |
| Yale OCS | University | Individual role |
| Kate Nasser | Coach | "I" and "we" |
| The Interview Guys | Commercial | 40/60 I:we |

Here is what the employer and university sources actually say:

- **Amazon** lists "Focus on 'I' not 'we'" as tip 4 of 11 in its recruiter tips. Its interview guide says that while describing collaborative efforts, candidates should "use 'I' statements to clearly articulate your personal contributions and decision-making process."
- **McKinsey's careers blog:** "We are interested in getting to know YOU, not your former team."
- **Yale's Office of Career Strategy:** "Though teamwork is valued, the interviewer is assessing your individual contributions."

The other side comes from coaches. Kate Nasser says "Let them see the 'I' and the 'we'." Circle City Consulting warns that without "we" stories, interviewers might think you are a lone wolf. The Interview Guys recommend a 40/60 ratio of "I" to "we" but publish no method behind it. I would not plan around that ratio.

### Why the rule exists

Interviewers don't trust "we". Alison Green of Ask a Manager says candidates "use 'we' to make their role sound bigger than it actually was," and that when asked about their role, "it often comes out that their role was fairly minor compared to what 'we' did." DDI's interviewer guidance says to "avoid accepting 'we' statements at face value. If a candidate talks about something 'we' did, ask them what their role was."

So whatever pronoun you use, the probe is coming. If you say "I", the interviewer tests whether it's true. If you say "we", they ask what your part was. The pronoun matters less than whether your opening sentence already answers the question that follows it. That is the job of this framework.

## How hard will the interviewer probe a "we" line?

Assume the interviewer will ask "What was your role?" at least once, and that the follow-ups may not come from any script. How deep interviewers usually go is not established publicly. My working target is two levels deep, and that is an assumption, not a sourced figure.

Structured interviewing, meaning every candidate gets the same questions and is scored against a predefined rubric, comes with planned probes. The US Office of Personnel Management's structured interview guide lays out eight development steps. Step 4 is "Develop the Probe Questions", and it tells interviewers to "make a list of acceptable probe questions for every competency interview question." Its action probes include "What was your role?"

The catch is that few interviewers describe their own method as structured. In Refolk's index of professional profiles:

| Segment | Profiles | Ratio |
|---|---|---|
| US, Behavioral Interviewing | 13,743 | baseline |
| UK, Behavioral Interviewing | 975 | US is 14.1x UK |
| US, Structured Interviewing | 48 | Behavioral is 286x Structured |

Counts are from Refolk's index. The ratios are derived from those counts.

**286x - US profiles listing Behavioral Interviewing versus Structured Interviewing**

In Refolk's index, 13,743 US profiles list Behavioral Interviewing and only 48 list Structured Interviewing.

In a 25-profile sample of the US Behavioral Interviewing set, 9 hold CEO or founder-type titles, which is 36%. Twenty-five profiles is a thin sample, so read it as directional. The picture it suggests is that many of the people interviewing you will be hiring managers and founders who probe based on what they hear, not from a fixed list.

What this means for scoring: you can't predict the exact follow-up, so prepare depth rather than memorising answers to expected questions. This is an inference from skill labels only. To check it for a specific employer, look at the public profiles of your scheduled interviewers and see which interviewing skills they list, if any.

## The six dimensions of a shared-credit line

Score every team-built line from 0 to 2 on six dimensions. The first four are **evidence**: what you did and whether it holds up. The last two are **context**: who is grading and what they reward. Evidence decides whether you can claim the line. Context decides how explicit you need to be.

| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Decision ownership | Recommended or executed | Owned a sub-decision | Made the call |
| Execution share | Minor slice | Named, separable piece | Most or a whole workstream |
| Follow-up depth | Stalls at first probe | One level | Two levels |
| Verification alignment | Record contradicts | Uncheckable or unsure | Manager would agree |
| Scorer format | Unstructured chat | Human rubric | AI transcript |
| Employer emphasis | Collaboration foregrounded | Standard behavioral | Publishes "I" rule |

### Decision ownership

This is whether the key decision behind the outcome was yours. It shows the interviewer judgement, which is usually the competency the question is really testing.

**What it looks like when it lies:** you recommended something and your manager approved it, but in memory it became "I decided". The test is the probe "What alternatives did you consider?" If you can't say who signed off, or the person who signed off made the actual choice, score it 0 or 1.

### Execution share

This is how much of the work you personally did. Use McKinsey's counterfactual: "think through what would have happened if you weren't involved." Write the answer as one sentence. If the honest answer is "it ships a week later", that is a 0. If it's "this piece doesn't exist", that is a 2.

**What it looks like when it lies:** you count hours instead of contribution, or you count work done by people you merely sat next to.

### Follow-up depth

This is how far you can go before you run out of specifics. Level one is your role and actions. Level two is the alternatives you rejected, why you rejected them, and how the result was measured. A line where you have the outcome but can't explain how you got there scores 0, however true it is.

**What it looks like when it lies:** you rehearsed the story so often that it sounds deep, but every detail is about what the team did.

### Verification alignment

This is whether the people and records that could check this line would back up your words. Standard verification confirms facts. A reference check is a more qualitative conversation.

| Resume claim | HR verification | Reference call |
|---|---|---|
| Title | Yes | Yes |
| Dates | Yes | Yes |
| Team size | No | Informal only |
| "Led" | No | Informal only |
| Metric/result | No | Informal only |

A neutral reference typically confirms only whether you worked there, your title, and your dates. Commercial employment-verification databases typically return employer name, tenure, job title, and gross income. No federal law requires a former employer to answer a verification request at all. So the claims most likely to be overstated, "led", team size, and the metric, are the ones only a former manager or peer can contradict, and only informally.

#### Who can check a resume line

1. **Interviewer probe** - Any claim, live, in every round
2. **Back-channel to a manager or peer** - "Led", team size, metrics, informally
3. **Formal reference call** - Title, dates, and a qualitative view
4. **HR verification** - Title and dates only

*The claims most likely to be overstated sit in the layers that formal verification never reaches.*

Score 2 if your former manager would describe the line in the same words and your title and dates match the record. Score 1 if it can't be checked or you aren't sure. Score 0 if anything checkable contradicts it: a title that differs from your offer letter, or a manager who would choose a different verb.

**What it looks like when it lies:** "uncheckable" feels safe. It isn't. You can't predict a back-channel call, and some interviewers know your old team.

### Scorer format

This is who grades the answer, and whether a "we" opening gets a second chance. A higher score means the format rewards explicit first-person action more and gives you less room to recover.

- **2, AI transcript-scored.** HireVue says that for AI-scored video assessments it "analyzes the transcript of what a candidate says" and scores each response against a BARS for each competency. A BARS, or behaviorally anchored rating scale, is a rubric where each score level is described by example behaviours. Secondary sources describe five levels from novice to expert, and a 1-to-5 rubric for skill questions in its AI Interviewer. Nobody is there to ask "what was your role?", so the transcript has to say it for you.
- **1, human panel with a rubric.** The probe will come, so a Split or Scope opening gets followed up instead of marked down.
- **0, unstructured conversation.** Context moves both ways, so a Scope opening costs you least here.

What happens to a mostly-"we" answer in an AI round is not established in HireVue's own documentation. Third-party sites claim that a story "that never names a decision you made scores low on leadership", and that a transcript of "we" sentences "shows no evidence of your own contribution". Both are unverified inferences. One secondary source also says the model judges meaning "regardless of the specific vocabulary used". If that is right, swapping "we" for "I" without naming concrete actions may not raise your score at all. Either way, naming specific actions is what helps.

### Employer emphasis

This is how much this employer says it cares about individual ownership. Score 2 if it publishes an "I" instruction, as Amazon and McKinsey do. Score 1 for a standard behavioral interview with no published guidance, which is Yale's description of the default. Score 0 if the posting and loop put collaboration first and the question asks directly about teamwork. Check the employer's careers pages and interview-prep material before you score this.

## What the score means: Claim, Split, or Scope

Add the six scores to get a total out of 12. Then apply one gate: the context dimensions can't push a line into Claim unless the evidence supports it. These cut-offs are my working model, built to stop context from inflating ownership. They are not a validated instrument, so adjust them if your own probing practice shows they are off.

| Call | Rule | Opening move | Verb that fits |
|---|---|---|---|
| Claim | Total 9+, evidence 6+, no evidence 0 | "I" from the first sentence | Led, owned, directed |
| Split | Everything else | "We" for context, "I" for your part | Coordinated, built, delivered |
| Scope | Total 4 or less, or ownership and execution both 0 | Size and slice first | Supported, contributed |

#### Decision ownership against execution share

Horizontal axis runs from Small execution share to Large execution share. Vertical axis runs from Recommended or followed to Made the decision.

| Quadrant | What it means |
| --- | --- |
| Decided, others built | Split - claim the call, credit the builders |
| Decided and built | Claim - say I from the first sentence |
| Followed, small slice | Scope - size the effort, then your slice |
| Built, someone else decided | Split - claim the build, name who decided |

*The two evidence dimensions that matter most set the default call before context adjusts it.*

### The opening sentence for each call

Your opening sentence carries the most weight in a team-work answer. It sets the pronoun, and it decides whether the first probe confirms your story or exposes it.

**Opening sentences for Claim, Split, and Scope**

```
CLAIM
I made the call to [decision] and built [piece] myself. The result was [measured outcome], and I'll walk you through how I got there.

SPLIT
We were a [team size]-person team working on [project]. My part was [specific piece], and the decision I owned was [decision], so I'll focus on that.

SCOPE
This was a [team size]-person effort over [period], led by [role of the lead]. My slice was [specific piece], and that's the part I'll take you through in detail.
```

*Fill the brackets from your scoring sheet; keep the opening to one or two sentences before the STAR detail.*

A Scope opening isn't a hedge. Interviewers are trained to ask about your role whenever they hear "we", so Scope answers the question before it's asked. From there, your actions are told in "I". Scope sets the size of the claim. It doesn't take you out of the story.

### The resume-fix flag

Some lines can't be fixed in the room. Flag a line for a resume edit when either of these is true:

- **The verb outranks the call.** Commercial verb guides say to "use 'led' when you directed the work, 'coordinated' when you organised its parts and 'supported' when you contributed under someone else's responsibility." "Led", "owned", or "directed" with decision ownership below 2, or with a Scope call, is a mismatch. So is "coordinated" with a Scope call.
- **Verification alignment scores 0.** If a record or a former manager would contradict the page, no spoken framing will save it.

> **Rule:** The page and the opening sentence must agree
>
> If the resume says "Led" and your opening sentence says "we", the interviewer notices before the second sentence. Fix the verb on the page, not the pronoun in the answer.

Resume edits are where most of the friction is, because each posting may need a different version of the same line. [Refolk](/candidates) writes your resume from your own history and tailors it to each posting. So when a line gets flagged, you change the verb once in the source and it carries through to every tailored version, instead of fixing it in several copies by hand.

### A worked example

Take the line "Led migration of the billing service to the new platform." Your manager picked the platform, so decision ownership is 1. You wrote the cutover plan and most of the migration code, so execution share is 2. You can explain the rollback design and how incidents were counted afterwards, so follow-up depth is 2. Your manager would say "drove the migration", not "led", so verification alignment is 1. That puts evidence at 6. The round is a human panel (1), at an employer with no published guidance (1). The total is 8, which means Split.

The opening: "We were a small platform team moving billing to a new platform; my part was the cutover plan and most of the migration code." The page says "Led" while the call is Split and decision ownership is under 2, so the flag fires. Change the verb to "Drove" or "Executed", then rehearse.

## Step-by-step: scoring every team-built line before the interview

Run this once per interview loop, and again whenever you tailor the resume for a new posting. It takes an evening for a typical resume. Most of that time goes into drafting stories, not scoring.

Sources disagree on where to start. Amazon's recruiters start from the job description, and McKinsey starts from the counterfactual. I start from the resume, because that is the page the interviewer has in front of them.

#### Score, call, and rehearse each team-built line

1. **List the team-built lines** - Flag every resume bullet whose outcome came from more than one person, in about 15 minutes. Done when no shared-outcome bullet is left unflagged.
2. **Write the counterfactual sentence** - For each line, write one sentence on what would have happened if you had not been involved, about 5 minutes per line. Done when every line has its 'without me' sentence.
3. **Check the verb against ownership** - Compare the verb on the page with the led, coordinated, supported convention. Any verb that outranks your real ownership gets a resume-edit flag.
4. **Map who can verify each line** - Tag each line as provable through title or dates, provable only through references, or not checkable, and note whether your former manager would use the same verb. About 10 minutes in total.
5. **Identify the scorer for each round** - Read your invitations and notices and label each round as AI transcript, human panel with a rubric, or unstructured conversation. Done when every round carries a label.
6. **Score, call, and draft the opening sentence** - Score the six dimensions, apply the evidence gate, and write a STAR story for each line that opens with its Claim, Split, or Scope sentence. Aim for eight or nine strong stories, 1 to 2 hours in total.
7. **Rehearse two levels deep with a peer** - Have a peer ask the action and result probes plus a second follow-up on each, for about 30 minutes. Done when you can answer two levels deep without changing the pronoun you opened with.

A few notes on the steps:

- **Step 5:** one source advises you to "check your invitation and notices for the process used for your role". If the invitation doesn't say, score scorer format as 1 and ask the recruiter.
- **Step 6:** STAR stands for Situation, Task, Action, Result. Put the Claim, Split, or Scope sentence in place of the usual Situation opener, or as its first line. Amazon suggests preparing "eight or nine strong examples", so you don't need a story for every flagged line. Choose the lines with the highest evidence scores.
- **Step 7:** the two-level target is my working assumption. If your peer keeps finding a third level you can't answer, lower that line's follow-up depth score and run the call again.

Use this sheet once per line:

**Shared-credit line scoring sheet**

```
LINE: [paste the resume line]
VERB ON PAGE: [Led / Owned / Directed / Coordinated / Supported / other]
WITHOUT ME: [one sentence: what happens if I was not involved]
SIGNED OFF BY: [name or role of the person who approved the decision]

EVIDENCE (0 to 2 each)
Decision ownership: _
Execution share: _
Follow-up depth: _
Verification alignment: _
Evidence subtotal: _ / 8

CONTEXT (0 to 2 each)
Scorer format (round: ______): _
Employer emphasis: _
TOTAL: _ / 12

CALL: Claim / Split / Scope
GATE CHECK: evidence 6+ and no evidence score of 0? yes / no
RESUME FIX: yes / no   Reason: [verb outranks call / record contradicts]
OPENING SENTENCE: [one or two sentences]
```

*One copy per team-built line; keep the evidence subtotal visible so context never lifts a weak line into Claim.*

## How this goes wrong: failure modes and false positives

Most shared-credit answers that fail do so in one of seven predictable ways. Each comes with a check you can run before the interview.

#### Where an overstated line gets caught

1. **Resume screen** - The verb on the page sets what you will be expected to back up
2. **Transcript or first probe** - An all-I answer scores well or survives one question
3. **Human review** - Someone watches the recording or probes for alternatives
4. **Later loop** - A deeper interviewer asks what you rejected and why
5. **Back-channel** - A former manager uses a different verb

*An inflated line can survive the first round and still fail at a later checkpoint you never saw.*

1. **Decision ownership gets inflated.** You say "I decided" when you only recommended. The probe "What alternatives did you consider?" exposes it within a minute. *Check:* can you name the person who signed off, and was the decision really yours?
2. **Execution share is overstated in an AI round.** An all-"I" transcript scores well, and then a human reviewing the recording, or a later interviewer, probes it. One source says HireVue "decides who a person watches first", so a high AI score makes human review of your answer more likely, not less. *Check:* would this answer survive a live follow-up? If not, the evidence gate should have kept it out of Claim.
3. **The verification gap creates false safety.** You assume "led" can't be checked because HR only confirms title. A back-channel call to your former manager is the false negative you never hear about. *Check:* would your manager use the same verb?
4. **Your resume title differs from the HR record.** One source notes that a mismatched title "sometimes reflects an informal promotion that never made it into official HR records." That is the one claim formal verification will contradict. *Check:* compare the title against your offer letter or latest official record. If they differ, score verification alignment 0 and either fix the title or be ready to explain it in one sentence.
5. **Over-correcting into "we".** Nasser reports leaders who "lose out because they never said 'I' in the interview." A candidate who has absorbed "don't take credit for team work" can turn every Claim line into a Scope opening and leave the interviewer with nothing to score. *Check:* if a line scores Claim, your opening sentence starts with "I".
6. **The gender backlash research gets misread.** See below.
7. **The verb doesn't match the spoken answer.** The page says "Led" and the answer says "we". *Check:* a mismatch between verb and call means fixing the resume, not the answer.

> **Watch out:** "Uncheckable" is not "safe"
>
> HR verification confirms title and dates, nothing more. The claims it can't reach, like "led", team size, and the metric, are exactly what a former manager confirms or contradicts in a call you will never hear about.

### What the self-promotion research does and does not say

Some candidates, particularly women, avoid "I" because they have heard that self-promotion backfires. The research is more specific than that.

- **Rudman (1998)** found that "both men and women who engaged in self-promotion were perceived as more competent. However, only men were perceived as more hirable."
- **The Williams and Tiedens (2016) meta-analysis** covered 71 studies of gender, dominance, and backlash. Dominance hurt women's likability relative to men's, but the overall effect was small (d = -0.19, k = 63). The effect on hireability was larger (d = -0.58, k = 20). Here, d is a standardised effect size and k is the number of effect sizes pooled.
- **Krings and colleagues** found that the backlash for high self-promotion depends on both gender and age.
- No peer-reviewed study I found tests "I" against "we" in interviews specifically. That is not established publicly.

The penalty is concentrated in explicit dominance and shows up more in the hiring decision than in how the conversation feels. It does not mean "I" is risky across the board. In practice: state your actions plainly, give credit to named teammates where it is due (which is what a Split opening does), and skip dominance displays like dismissing what the team did. Choosing Scope for a line that scores Claim doesn't protect you. It just leaves the interviewer nothing to score.

> A Scope opening answers the role question before it is asked, which is precision, not modesty.

## Keeping the scores current between postings

Scores go stale, because two of the six dimensions belong to the employer, not to you. Rescore the context dimensions for every new loop, and rescore the evidence dimensions whenever you change a line on the page.

What to re-check, and how:

- **Employer emphasis.** Reread the employer's careers pages and interview-prep material before each loop. Amazon's "I" guidance and McKinsey's "YOU, not your former team" are published pages that can change. Score what the page says now, not what you remember.
- **Scorer format.** AI scoring vendors revise their methods and documentation. HireVue, for example, publicly announced that it had removed facial analysis from its assessments. Before an AI round, read the vendor's current science or explainability page and check whether it now says how individual contribution is credited. Until it does, treat the "we"-penalty claims as unverified.
- **Verification alignment.** If you have talked to a former manager since you last scored a line, use what they said. One conversation can move a line from 1 to 2, or from 1 to 0.
- **The verb on the page.** Each tailored resume is a new set of claims. If you tailor per posting, the verb-to-call check has to run on the version you actually sent. Refolk keeps one source history behind every tailored version, so a verb you corrected once stays corrected in the next posting's resume.

#### Before the interview, every team-built line has

- [ ] A written one-sentence counterfactual describing what would have happened without you
- [ ] A named person who signed off on the decision behind the line
- [ ] Six dimension scores, with an evidence subtotal of 6 or more on every Claim line
- [ ] A verb on the page that does not outrank the Claim, Split, or Scope call
- [ ] A title that matches your offer letter or official HR record
- [ ] A labelled scorer for the round where this story will be told
- [ ] An opening sentence that starts with "I" for Claim, "we" then "I" for Split, or size and slice for Scope
- [ ] A rehearsal where a peer probed two levels deep and your pronoun held

> **Note:** Where the evidence is thin
>
> How deep interviewers usually probe, and how AI scorers treat "we", are both unknown publicly. Calibrate the two-level target and the scorer-format weights against your own rehearsals and the vendor's current documentation.

## Frequently asked questions

### Should I say I or we in a job interview?

Say "I" for the decisions and work you actually owned, and use "we" only to set context for the team around you. Employers like Amazon and McKinsey explicitly ask for "I" because they are assessing you, not your former team. But an all-"I" answer about work you only partly did will come apart under follow-up. Score each line on ownership and execution share, then open with the Claim, Split, or Scope sentence that fits.

### My resume says led but I was part of a team. What do I do?

Change the verb on the resume first, then prepare the spoken answer. Commercial verb guides reserve "led" for work you directed and suggest "coordinated" or "supported" otherwise. HR verification will not catch the mismatch because it confirms only title and dates, but a back-channel call to a former manager will. And if you say "we" in the room while the page says "Led", the interviewer hears the mismatch at once.

### How do I answer 'what was your role in the project' without sounding like I am hiding something?

Answer in one sentence that gives the size of the effort and your slice, then go straight to your own actions. For example: the team size, who led it, the specific piece you owned, and one decision you made. This is the Scope opening. It works because it answers the probe before it lands. Interviewers who hear "we" are trained to ask about your role anyway, so naming it first comes across as precise, not evasive.

### Do AI interview scorers penalise answers that use we?

That has not been established publicly. HireVue says it scores the transcript of what you say against a behaviorally anchored rubric for each competency, and one secondary source says its model judges meaning regardless of the specific words used. Third-party sites claim that "we"-heavy answers score low on individual contribution, but those claims are inferences, not vendor statements. The safe move is to name concrete actions you took, which helps under either reading.

### Is it risky for women to use I statements in interviews?

The research does not support avoiding "I" across the board. The Williams and Tiedens meta-analysis found the backlash concentrated in explicit dominance behaviour and in likability, with a larger effect on hireability downstream. Krings and colleagues found the backlash depends on both gender and age. No peer-reviewed study tests "I" against "we" in interviews specifically. State your actions plainly and skip dominance displays, like talking over or dismissing teammates.

---

*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/shared-credit-claim-split-scope*