<aside> 🔒

Recreated for this portfolio. The original coaching materials belong to my former employer, a home care company, so this version shows the same method without confidential details. Informed by my Cognitive Coaching training (Foundation and Advanced); structure and wording are my own.

</aside>

The problem

Most manager coaching programs end at the workshop. Managers leave with a model and a handout, and nobody checks whether the conversations they have on Tuesday look anything like what they practiced. The program gets rated on attendance and satisfaction, which measures the training, not the behavior.

I trained 6 of our 7 managers to move from directive management to coaching, and coached 15+ coordinators alongside them. I needed to know whether they were coaching after I stopped watching, not whether they liked the session.

The design: coaching as a demonstrated skill

Stage What happens What it proves
1. Co-coach Group workshop on the model, then the manager coaches coordinators alongside me. I model, then they lead while I watch and debrief. They can run the conversation with support
2. Solo, recorded The manager coaches without me. Every coaching call is recorded. They can run it when no one is in the room
3. Scored over time Periodic AI scoring of transcripts against a four-part rubric, with my own review layered on top. The behavior holds, or drifts, after training ends

Where this sits on the Kirkpatrick model

Level What it asks How this program answered it
1. Reaction Did they like it? Not the goal. Satisfaction doesn't tell you whether anyone coaches on Tuesday.
2. Learning Can they do it? Stage 1: managers coached alongside me before going solo.
3. Behavior Are they doing it on the job? Stages 2 and 3: recorded calls, scored against the rubric over time. This is the core of the program.
4. Results Did the organization change? Fewer escalations reached me once the manager layer existed, and managers ran performance plans and caregiver issues themselves. Observed, not measured with a number.

The rubric

Four observable behaviors. Each one is something you can point to in a transcript, not an impression.

Behavior Not yet Developing Coaching
Talk ratio Manager does most of the talking Roughly even Coordinator does most of the talking
Asking vs telling Manager supplies the answer Asks, then answers their own question Asks, waits, and lets the coordinator reach it
Paraphrasing and listening Moves straight to advice Paraphrases occasionally Paraphrases before asking, and the coordinator corrects or confirms it
Coordinator-owned next step Manager assigns the next step Next step is agreed but manager-shaped Coordinator names the next step in their own words

A scored call, in 20 seconds

An illustrative excerpt of a reflecting conversation, scored the way the program scored real calls.

Speaker What was said What the rubric sees
Manager "How do you think yesterday's coverage call went?" Asking, not telling
Coordinator "Honestly, not great. I called three caregivers, nobody could take it, so I panicked and called you."
Manager "So you tried three people, came up empty, and felt stuck. Is that right?" Paraphrase before asking
Coordinator "Yeah. I didn't know who else to try."
Manager "What would have helped you know who to try next?" Asking, not telling
Coordinator "Maybe checking the floating caregiver list first, instead of calling who I usually call." Coordinator finds the fix
Manager "What will you do next time?" Asking, not telling
Coordinator "It's fine. I'll just check the list." Next step named, but listen to how
Behavior AI score My review
Talk ratio Coaching Coaching
Asking vs telling Coaching Coaching
Paraphrasing and listening Coaching Coaching
Coordinator-owned next step Coaching Developing. The words say she owns it; the tone says she's deflated. The manager moved on instead of checking in.

<aside> 💡

This is the gap I kept finding: AI scores the words. It misses the emotional cue underneath them. That's why the human review stayed in the loop.

</aside>