Written by: Mark Hull, Co-Founder and CEO, Exceeds AI | Last updated: September 3, 2026
Key Takeaways
- Vague goals like “improve reliability” create misalignment and unprovable outcomes because they lack specific, measurable criteria.
- The GROW coaching framework (Goal, Reality, Options, Will) gives engineering leads a repeatable way to move from vague intentions to concrete, committed action plans.
- 24 copy-paste coaching prompts organized by GROW stage help engineering leads run goal-setting conversations that end with measurable outcomes.
- Clear goals require objective measurement, and traditional analytics struggle to track AI-assisted work accurately.
- Exceeds AI adds the missing measurement layer with code-level analytics that track AI versus human contribution, rework rates, and goal progress, so you can verify impact with data. Book a demo today.
The GROW Coaching Model for Modern Engineering Teams
The GROW model structures a coaching conversation into four sequential stages: Goal, Reality, Options, and Will. Each stage has a distinct purpose. Together they move an engineer from a vague intention to a concrete, committed action plan.
- Goal: What does the engineer want to achieve? The goal belongs to them.
- Reality: What is the current state? What data exists? What is blocking progress?
- Options: What approaches have not been tried? What else is possible beyond the obvious?
- Will: What is the specific commitment? When will the first action happen?
This sequence matches how effective engineering work runs in the AI era. Christoph Nakazawa’s “Modern Engineering Values” recommends that when working with AI coding agents, engineers start with a clear goal, gather context from specified sources, make a proposal, and ask clarifying questions before execution. That sequence mirrors GROW. Clear goal first, then context, then a proposal, then action produces reliable outcomes whether you coach a direct report or prompt an AI agent.
See how Exceeds AI supports GROW-style goal tracking. Book a demo.
24 Coaching Prompts for Engineering Team Leads by GROW Stage
Goal-Setting Coaching Prompts
- “What does success look like in three months? Describe it as if it has already happened.”
- “How will you know this goal is achieved? What metric will move?”
- “What is the smallest version of this goal that would still feel like a win?”
- “How does this goal connect to our team’s quarterly priorities?”
- “What would make this goal feel meaningful to you personally?”
- “If we fast-forward to the end of the quarter, what specifically did you accomplish?”
Reality Prompts
- “What is currently blocking you from achieving this?”
- “What data do you have about the current state? What metrics are you tracking?”
- “What has changed since we last discussed this?”
- “What have you already tried, and what happened?”
- “What resources or support do you need from me?”
- “What is the honest gap between where you are and where you want to be?”
Options Prompts
- “What are some approaches you have not tried yet?”
- “If you had unlimited resources, how would you solve this?”
- “What would you do if you could not fail?”
- “Who else in the org has solved a similar problem? What did they do?”
- “What is the most unconventional option available to you?”
- “If you had to solve this in 48 hours, what would you do first?”
Will Prompts
- “What is your first step, and when will you do it?”
- “What could derail you, and how will you prevent it?”
- “How will you hold yourself accountable? What checkpoints will you set?”
- “Who needs to know about this goal, and when will you tell them?”
- “On a scale of 1–10, how committed are you to this? What would make it a 10?”
- “What will you have accomplished by our next 1:1?”
From Vague Goals to Measurable Outcomes
The prompts above work because they force specificity. They turn a loose intention into a measurable goal through a GROW conversation.
Vague: “Improve reliability.”
Measurable: “Reduce checkout-service error rate from 1.8% to below 0.8% by end of Q4.”
Vague: “Improve code quality.”
Measurable: “Reduce rework rate on AI-assisted PRs by 20% within two sprints.”
Vague: “Be more proactive.”
Measurable: “Lead the migration of our CI pipeline to GitHub Actions by March 15, with zero unplanned downtime.”
Each transformation comes from a coaching conversation using GROW prompts. The engineer builds the goal instead of receiving it. That shift increases ownership and follow-through. As Charity Majors argues, the knowledge in engineers’ heads stays unavailable to AI until they encode it into the system. Coaching prompts support that encoding. They surface what the engineer already knows about the problem and translate it into a form that you can track and act on.
How to Measure Goal Progress with AI-Heavy Work
Clear goals need measurable outcomes. Tracking “rework rate” or “error rate” objectively requires more than a spreadsheet and a quick gut check at the next 1:1.
Traditional developer analytics platforms track metadata such as PR cycle times, commit volumes, and review latency. These tools usually cannot distinguish AI-generated code from human-written code. That limitation matters because many 2026 engineering goals involve AI-assisted work. When a goal focuses on “reducing rework on AI-assisted PRs,” the goal becomes unmeasurable if your tooling cannot identify which PRs used AI.
Exceeds AI fills that gap as the measurement layer. Exceeds AI provides code-level analytics that track the metrics that matter for goal-setting: cycle time, rework rates, incident rates, and AI versus human contribution. It works down to the commit and PR level across every AI tool your team uses, including Cursor, Claude Code, Codex, GitHub Copilot, and Windsurf.
Exceeds Ink, the provenance layer that powers the platform, captures line-level authorship across those tools. When an engineer sets a goal like “reduce rework on AI-assisted PRs,” Exceeds shows which patterns drive rework and which interaction modes (plan, ask, agent, edit) correlate with higher quality outcomes. Coaching Surfaces turns that data into prescriptive guidance. The next 1:1 coaching conversation then rests on evidence instead of impressions.
See how Exceeds AI measures AI-assisted goals. Book a demo.
The 1:1 Prompt Card for Fast Goal Conversations
These six questions cover the full GROW arc for any goal-setting conversation. Keep them in your 1:1 doc template so you can move from idea to measurable goal quickly.
- What does success look like, and how will you measure it?
- What is the current reality, and what is blocking you?
- What options have you not yet explored?
- What is your first step, and when will you take it?
- How will you track progress between now and our next conversation?
- What support do you need from me to make this happen?
Frequently Asked Questions
What are the three key goals of coaching?
Coaching aims to help engineers take ownership of their goals, gain clarity on what they want to achieve, and grow their skills and capabilities. Ownership comes first. Engineers care more about goals they help create, so coaching prompts ask the engineer to define success instead of receiving a goal from above.
That ownership, in turn, makes open-ended questions effective at pushing vague ideas into concrete targets. Clear targets reduce the ambiguity that causes misalignment at the end of a quarter. Strong prompts also connect daily tasks to long-term career skills, so the engineer sees the goal as meaningful for their development and the team’s roadmap.
What is a good goal for a team lead?
A good goal for a team lead is specific, measurable, and aligned with both team and organizational priorities. “Reduce the average PR review cycle time from 48 hours to 24 hours by end of quarter” beats “improve review speed” because it names a starting point, a target, and a deadline.
Effective goals tie directly to outcomes the lead can influence and track with objective data. Goals that include a measurement method, such as a metric, a threshold, or a date, create a stronger coaching conversation at the next check-in than goals that rely on a subjective judgment call.
What are some examples of SMART goals for engineers?
SMART goals for engineers are specific, measurable, achievable, relevant, and time-bound. Strong examples include: “Reduce API latency from 300ms to under 150ms by end of Q4,” “Increase test coverage for the payments module from 65% to 85% within two sprints,” and “Cut CI pipeline build time from 12 minutes to under 5 minutes by March 1.”
Each example names a current state, a target state, and a deadline. That structure makes it straightforward to assess progress at any point in the quarter without relying on subjective self-reporting.
Conclusion: Turning Coaching Conversations into Provable Outcomes
Vague goals sit at the root of misaligned engineering teams and unprovable AI ROI. The GROW framework, applied through structured coaching prompts for engineering team leads, turns that ambiguity into measurable outcomes by making the engineer the author of their own goals.
Clear goals still need clear measurement. As Charity Majors writes, AI demands more engineering discipline and that discipline depends on observable, evidence-based progress instead of subjective status updates. Exceeds AI closes the loop with code-level analytics that show whether goals are being met at the commit and PR level across every AI tool your team uses. When an engineer commits to reducing rework on AI-assisted PRs, Exceeds shows you and the engineer exactly whether that change is happening.
Track whether your team’s goals are working in practice. Book a demo.