In a recent edWebinar, education leaders came together to explore whether early-career teacher attrition and inconsistent field supervision aren't two separate problems, but instead, the same problem seen from two different budgets. Their conversation kept circling back to one idea: video-based instructional coaching, supported by AI, is what moves the needle on teacher retention and growth, from educator prep through the first years in the classroom.
Moderated by Terry Chvatal, Director of Vosaic, the session brought together Mikki Brooks, an Education Consultant at Educational Service Unit #3 in Nebraska with a background as a K-12 instructional coach and administrator, and Dr. Tara Kaczorowski, Professor and Executive Director of Education Programs at Daemen University, whose research centers on teacher preparation and video-enhanced reflection. Together, they unpacked how video-based reflection and AI are reshaping instructional coaching, both in K-12 districts and in teacher preparation programs.
When Observation and Feedback Systems Fail
Dr. Kaczorowski described the "bobblehead syndrome" of teacher candidates nodding along in a post-lesson debrief, only for everyone involved to later realize they remembered the lesson differently. Mikki recalled watching an observation model collapse into a compliance checklist: a scripted walkthrough, a software-driven rating, and a mandate to change practice to match a teaching partner's approach, with no real focus on growth.
The Real Cost of Weak Early-Career Support
Nationally, 1 in 8 teaching positions are unfilled or staffed by someone not fully certified for the role. Replacing a single teacher can cost a district $12,000 to $25,000. At current spending levels, most systems can only support one instructional coach for every 35 teachers.
Yet early-career support is often the first item to get cut. Dr. Kaczorowski and Mikki pointed to a familiar dynamic where coaching doesn't have the same direct, visible impact on students as staffing a classroom does, so it gets treated as expendable, even though weak onboarding support is a major driver of the teachers who don't stay.
Why the Traditional Observation Model Breaks Down
The presenters described the traditional cycle in four steps: schedule a visit, observe from memory, hold a conference days or weeks later, and file a rating. Drawing on Jim Knight's work, they noted that friction builds whenever a coach's perspective and a teacher's memory of the same lesson don't align.
Dr. Kaczorowski pointed to the layers of breakdown in that model. Supervisors either miss moments while they're occupied taking notes, or they wait until after the lesson and rely on memory that might not be accurate. Mikki added that feedback in this model tends to slide from fact into inference and opinion, meaning whatever the observer happened to be watching, and missing, shapes the entire conversation. As attendees noted during the session, an observation is only as reliable as that day's classroom temperature, not just whether the teacher or students were having an off day, but whether the observer was too.
From Conferences to Conversations
Both speakers have moved away from the language of "pre-conference" and "post-conference" entirely. Instead, they use planning conversations before a lesson and reflective conversations after it. This frames the focus on formative growth rather than a single evaluative snapshot.
Critically, the teacher or candidate helps set the agenda. Mikki mentioned asking teachers directly what they want observed and what would be meaningful for their own growth, rather than dictating what the coach is watching for. Dr. Kaczorowski echoes this by asking her candidates what they want their supervisor to watch before any video is even shared.
Video as the Great Equalizer
You can't watch yourself teach, at least not without video. Most educators are their own harshest critics, and reviewing footage of their own teaching consistently reveals they did better than they believed.
Following Jim Knight's principle that the recording belongs to the teacher and participation must be a choice, Mikki shared how she rolls video out with clear guardrails. Starting with forward-thinking teachers who already want the feedback, having administrators and coaches record and reflect on themselves first, and giving teachers a two-week window to keep their own footage before it's deleted. This reinforces that the tool is confidential and theirs.
Dr. Kaczorowski uses the same "show me" approach with teacher candidates. They record their own lesson, select the moments they want their supervisor to see, and reflect on those clips before any conversation happens. She calls it game film, which is footage used to teach candidates how to think about their own practice. It’s a good skill to bring with them into their first classroom.
Where AI Fits In
Both Dr. Kaczorowski and Mikki were clear that AI is a tool, not a replacement for coaching or judgment. In teacher prep, it's used to handle lower-cognitive-demand work, so people can spend more time on deeper reflection instead of scrubbing through footage.
AI can reliably quantify things like teacher talk time, wait time, and how often an educator built on a student's idea. What it can't do is tell you whether the lesson was worth teaching in the first place. One study cited during the session found that AI-scored observations correlate more closely with expert human raters than human raters correlate with each other, but Dr. Kaczorowski and Mikki were emphatic that the value still comes from the conversation the data prompts, not the data itself.
Bridging Teacher Prep and the First Years in the Classroom
A recurring theme was the disconnect between how teacher prep programs coach candidates and how early-career teachers are supported once they're hired. There are two systems addressing the same goal with little talking to each other. Dr. Kaczorowski sees video and reflective coaching as the bridge where candidates who are comfortable reflecting on their own footage are better equipped to keep doing it as full-time teachers, rather than waiting to be told what to fix.
The Data Backs It Up
Across roughly 60 studies, coaching shows a strong positive effect on instruction and student achievement. Smaller, targeted programs under 100 teachers see nearly double the impact of large-scale rollouts. One trial of 78 secondary teachers and about 2,200 students found real achievement gains, though they showed up the following year rather than immediately, underscoring that meaningful instructional change takes time to surface in outcomes.
Built with Privacy in Mind
Given how important video is to this model, Dr. Kaczorowski and Mikki closed the session by addressing privacy. Vosaic includes built-in face-blurring, holds relevant compliance certifications (i.e., FERPA, HIPAA, and COPPA), and integrates with existing LMS systems, with implementation typically taking about a day since it's fully cloud-based.
The combination of secured, teacher-owned video, AI that handles the busy work, and a coaching model built around conversation rather than evaluation is what allowed Dr. Kaczorowski and Mikki to connect two worlds that often operate in isolation: the classroom where a teacher is trained, and the classroom where they're expected to grow.
Ready to Close the Gap?
Whether you're supervising teacher candidates or supporting early career teachers already in the classroom, Vosaic offers a way to make reflection, not evaluation, the center of growth at every stage of the educator journey. Sign up for a free trial or submit a contact us form to learn more.
