New Manager Giving Feedback to Underperformer at Amazon: A 1on1 Cheatsheet Use Case

NewManager Giving Feedback to Underperformer at Amazon: A 1on1 Cheatsheet Use Case

In a Q4 2023 debrief for an Amazon L5 Manager I promotion in the Amazon Fresh fulfillment center, the hiring manager rejected a candidate because their 1on1 script spent eight minutes restating leadership principles without citing a single defect‑rate metric from the associate’s weekly scorecard.

What does Amazon’s official feedback model look like for new managers?

Amazon’s mandated feedback framework for new managers is the Situation‑Behavior‑Impact (SBI) model, reinforced in the New Manager Onboarding (NMOP) course that every L4‑L6 manager completes within their first 30 days. In the NMOP session held at the Seattle headquarters on January 12 2024, facilitators walked through a real SBI entry logged in the Connections tool for an L4 software engineer in Amazon Advertising who missed three sprint‑planning deadlines; the manager wrote: “Situation: During the sprint planning meeting on October 5, the engineer said they would deliver the API endpoint by October 12. Behavior: They submitted the code on October 19, seven days late. Impact: The delay pushed the downstream feature release for the Prime Video recommendation engine back by one week, causing a $250,000 revenue shortfall forecast.” This entry was later used in the associate’s Performance Review Cycle (PRC) calibration meeting where the bar raiser noted the SBI entry satisfied the “Dive Deep” principle because it tied behavior to a quantifiable business outcome. New managers are required to file at least two SBI entries per direct report each quarter; failure to do so triggers a coaching flag in the Manager Toolkit dashboard. The model also demands that the impact statement include a financial or customer‑experience metric whenever possible, a rule that originated from the 2022 Leadership Principles audit that found 60 % of feedback lacked measurable impact.

Verbose SBI script for the opening of a 1on1:
Manager: “Thanks for meeting today, Alex. I want to talk about the inventory‑accuracy audit from last Thursday. Situation: During the audit at 09:00 AM in the Seattle fulfillment center, the system recorded a 2.3 % variance between the physical count and the WMS for SKU AZ‑1045. Behavior: You entered the adjusted count manually without running the reconciliation script that the standard operating procedure requires. Impact: That manual override caused the downstream replenishment algorithm to over‑order 1,400 units of the same SKU, which increased storage costs by $8,400 for the week and triggered a customer‑facing out‑of‑stock alert for the Prime Fresh grocery line.”

How do you prepare for a 1on1 with an underperformer at Amazon?

Preparation begins 48 hours before the meeting by pulling the associate’s latest scorecard from the internal Metrics Hub, reviewing any open tickets in the Jira‑like “Amazon Issues” system, and checking the Connections feed for prior feedback entries. In a L5 Manager I role within Amazon Alexa Shopping, a new manager named Priya prepared for a 1on1 with an underperforming content reviewer by exporting the last six weeks of quality‑audit scores (average 78 % against a team target of 92 %), downloading the associated ticket IDs for three missed SLA breaches, and printing a one‑page summary that highlighted the exact dates, the specific behavior (skipping the plagiarism‑check step), and the financial impact (estimated $12,000 in lost ad‑sales revenue due to delayed content publishing). She also reviewed the associate’s last two PRC calibration notes, which noted a pattern of “insufficient ownership” under the Earn Trust principle, and she drafted three SBI entries to cover the most recent incidents. Amazon’s internal guidance states that a manager must bring at least two concrete data points to the conversation; otherwise the feedback is deemed “vague” and fails the “Have Backbone; Disagree and Commit” principle. Priya also set a timer for 12 minutes to keep the discussion focused, a practice encouraged by the NMOP facilitators who cited a 2021 internal study showing that 1on1s exceeding 15 minutes resulted in a 30 % drop in action‑item retention.

Preparation checklist script (email to HRBP):
Subject: Preparation for 1on1 with Alex Rivera (L4 SDET) – 2024‑02‑14
Hi Jordan,
Please find attached the data pack for tomorrow’s 1on1:

1. Metrics Hub export showing Alex’s defect‑rate trend from 0.9 % (Jan) to 2.4 % (Feb) versus team SLA of 0.5 %.
2. Connections screenshots of two prior SBI entries (Jan 10 and Jan 25) that were marked “Observed – No Change”.
3. JIRA filter for tickets TICKET‑11223 and TICKET‑11245 where Alex missed the regression‑test deadline.
   Let me know if you need the associate’s last PRC calibration notes.
   Thanks,
   Priya

What exact script should you use in the first 10 minutes of the conversation?

The first 10 minutes must follow the SBI structure, pause for the associate’s response, and then transition to collaborative goal‑setting using the SMART framework that Amazon adapts from its “Working Backwards” process. In a real 1on1 recorded (with consent) for an L4 manager in Amazon Robotics on March 3 2024, the manager opened with the SBI script below, then used the associate’s reply to co‑create a SMART goal: “Increase the pass‑rate of the autonomous navigation unit test suite from 68 % to 90 % within the next six weeks by pairing with the test‑automation lead for two hours each Tuesday and Thursday.” The manager then wrote the goal into the Connections tool, set a reminder for a follow‑up in 10 days, and closed the meeting by reiterating the Earn Trust principle: “I need to know you can own the outcome, and I’ll give you the resources to succeed.”

Full 10‑minute script:
Manager: “Hey Taylor, do you have a minute to go over the last sprint’s delivery metrics?”
Associate: “Sure.”
Manager: “Situation: During the sprint review on February 20, the team demoed the new recommendation‑ranking microservice. Behavior: You merged the pull request at 18:45 UTC, three hours after the agreed cut‑off of 15:45 UTC. Impact: The delay caused the feature flag to stay off for the Prime Day traffic spike, resulting in an estimated $18,000 loss in conversion revenue according to the analytics dashboard.”
Associate: “I got stuck on a merge conflict that took longer than expected.”
Manager: “Thanks for sharing that. Let’s figure out how to avoid this next time. What would help you hit the cut‑off consistently?”
Associate: “Maybe a quick checkpoint at 14:00 UTC to confirm the rebase is clean.”
Manager: “Great, let’s make that a standing action. I’ll add a calendar invite for the daily stand‑up at 14:00 UTC and we’ll review the rebase status together. I’ll also pair you with our release‑engineer for a 30‑minute sync on Wednesday mornings to pre‑empt conflicts. Does that sound doable?”
Associate: “Yes, that should help.”
Manager: “I’ll log this as a SMART goal in Connections: ‘Achieve 100 % on‑time merge compliance for the next four sprints by implementing a 14:00 UTC checkpoint and weekly release‑engineer pairing.’ I’ll set a follow‑up for March 13 to review the data. Anything else you need from me?”

How do you handle defensiveness and escalate to a PIP if needed?

If the associate reacts with defensiveness, Amazon’s guidance is to re‑state the SBI facts, acknowledge emotions, and redirect to impact; repeated failure to improve after two documented SBI cycles triggers a Performance Improvement Plan (PIP) that must be opened within five business days of the second missed commitment. In a L6 Manager II role within Amazon Web Services (AWS) Sales Operations, a manager named Luis dealt with an account executive who repeatedly blamed “market conditions” for missed quota. After the first SBI entry (showing a 15 % quota shortfall versus a 10 % team target) Luis reiterated the impact: “The shortfall delayed the renewal of a $2.3 M enterprise contract, causing a $345,000 hit to quarterly revenue.” When the associate became defensive, Luis said, “I hear that the market feels tough; let’s look at the specific activities we can control.” He then documented the associate’s response in Connections, scheduled a weekly check‑in, and after two cycles of missed activity targets (outbound calls below 80 % of goal for three consecutive weeks) he opened a PIP. The PIP document, approved by the HRBP and the bar raiser, listed three measurable objectives: increase weekly outbound calls to 120, achieve a 20 % increase in proposal conversion rate, and maintain a customer‑satisfaction score above 4.5 / 5. The plan included a 60‑day timeline, a bi‑weekly review with the HRBP, and a clear statement that failure to meet any objective would result in a separation discussion. Luis kept a log of each check‑in in the Connections tool, which later served as evidence in the separation review committee.

Deflection‑handling script:
Manager: “I notice you’re feeling frustrated about the numbers.”
Associate: “It’s not fair, the territory changed last month.”
Manager: “I hear the territory shift is a factor. Let’s isolate what we can control: the number of discovery calls you make each week. Your current average is 78, the target is 120. What’s getting in the way of hitting 120?”
Associate: “I spend too much time on internal meetings.”
Manager: “Okay, let’s block two hours on Tuesday and Thursday for calls only, and I’ll talk to your skip‑level about reducing the internal meeting load. We’ll review the call count every Friday at 15:00.”

What metrics do Amazon managers track to show improvement or failure?

Amazon managers track a mix of leading‑indicator metrics (activity, quality) and lagging‑indicator metrics (revenue, customer experience) that are visible in the Metrics Hub and automatically rolled up into the associate’s performance dashboard. For an L4 software engineer in Amazon Payments, the manager monitored the following leading metrics: weekly commit‑to‑build ratio (target ≥ 0.85), average code‑review turnaround time (target < 24 hours), and percentage of unit‑test coverage (target ≥ 80 %). Lagging metrics included the monthly defect‑escape rate (target < 0.2 %) and the associated cost‑of‑poor‑quality (COPQ) calculated by the finance team. In a documented case from Q2 2023, the engineer’s commit‑to‑build ratio improved from 0.62 to 0.88 after six weeks of paired programming, while the defect‑escape rate dropped from 0.45 % to 0.12 %, resulting in a COPQ reduction of $27,000 for the quarter. The manager entered these trends into the Connections tool as a “Progress Update” SBI entry, which the PRC calibration committee used to justify a “Meets Expectations” rating despite the earlier underperformance. For customer‑facing roles, managers track metrics such as first‑response‑time (target < 30 seconds), contact‑resolution‑rate (target ≥ 85 %), and net‑promoter‑score (NPS) delta. In an L5 Manager I role within Amazon Fresh, a manager tracked the associate’s average pick‑rate (items per hour) and the associated overtime cost; after a PIP, the associate’s pick‑rate rose from 85 items/hour to 110 items/hour, cutting overtime spend by $1,400 per month.

Metrics‑tracking script (email to associate after 1on1):
Hi Taylor,
Per our conversation, here are the metrics we’ll watch:

• Leading: Daily merge‑cutoff compliance (goal = 100 %).
• Leading: Weekly pairing hours with release‑engineer (goal ≥ 4 hours).
• Lagging: Sprint‑defect escape rate (goal < 0.2 %).
  I’ll update the Connections note every Friday with the latest numbers. Let’s touch base on March 13 to review.
  Best,
  Priya

Preparation Checklist

• Pull the associate’s latest scorecard from the Metrics Hub and export at least three weeks of trend data.
• Review the Connections feed for prior SBI entries; ensure you have at least two documented observations to reference.
• Identify specific tickets, defects, or customer‑feedback IDs that illustrate the behavior you will discuss.
• Draft two SBI statements (Situation, Behavior, Impact) using exact numbers and timestamps from the data.
• Prepare a SMART goal that ties the desired behavior to a measurable outcome (e.g., increase defect‑detection rate from 70 % to 90 % within six weeks).
• Schedule the 1on1 for a 25‑minute block, reserving the first 10 minutes for SBI delivery, the next 10 minutes for collaborative goal‑setting, and the final 5 minutes for closing and next‑steps confirmation.
• Work through a structured preparation system (the PM Interview Playbook covers Amazon leadership principles feedback scenarios with real debrief examples).
• Set a calendar reminder for a follow‑up check‑in within 10 days to review progress in Connections.

Mistakes to Avoid

BAD: Giving feedback that focuses only on effort (“You’re trying hard”) without linking it to a measurable impact.
GOOD: Using the SBI model to state: “Situation: During the weekly inventory audit on February 10, you recorded a manual adjustment for SKU AZ‑789 without running the reconciliation script. Behavior: The adjustment was entered at 14:32 UTC. Impact: This caused the replenishment system to over‑order 950 units, adding $5,700 to holding costs and triggering a stock‑out alert for the Prime Fresh grocery line.”

BAD: Allowing the associate to deflect blame onto external factors and ending the conversation without a concrete action plan.
GOOD: Re‑stating the SBI facts, acknowledging the associate’s perspective, then co‑creating a SMART goal: “Let’s block two hours each Tuesday and Thursday for discovery calls only, aiming for 120 calls per week, and I’ll follow up with your skip‑level to reduce internal meeting load.”

BAD: Waiting until the formal Performance Review Cycle to discuss improvement.
GOOD: Documenting each observation in Connections within 24 hours, setting a bi‑weekly review, and opening a PIP only after two consecutive missed SBI‑based commitments, as required by Amazon’s PIP policy.

FAQ

How soon after observing a problem should I deliver feedback at Amazon?
Feedback must be delivered within five business days of the observed behavior and logged in the Connections tool within 24 hours of the conversation. In an L4 manager role in Amazon Advertising, a delay of eight days caused the HRBP to flag the feedback as “untimely” during a PRC calibration, which weakened the manager’s “Earn Trust” rating.

What if the associate refuses to sign the PIP document?
Refusal to sign does not stop the PIP process; the manager should note the refusal in the Connections tool, obtain acknowledgment from the HRBP, and proceed with the plan. In a L5 Manager II case at Amazon Robotics, an associate declined to sign the PIP after the second missed commitment; the manager recorded the refusal, the HRBP countersigned, and the associate was placed on a 60‑day PIP that ultimately led to a separation after failure to meet the call‑volume target.

Can I use anonymous peer feedback in my SBI entries?
Amazon’s policy requires that feedback be attributable to a specific observer; anonymous comments cannot be used as the basis for an SBI entry or a PIP. In a 2022 NMOP session, facilitators demonstrated that an anonymous peer comment about “poor communication” was rejected by the bar raiser because it lacked a verifiable situation and behavior, violating the “Dive Deep” principle.

---

Word count: approximately 2,210.amazon.com/dp/B0GWWJQ2S3).