Why Your Professional Development Isn’t Changing Classroom Practice (and What Will)
Districts spend an average of $18,000 per teacher per year on professional development. Across the 50 largest school districts, that adds up to an estimated $8 billion annually. According to TNTP’s landmark study “The Mirage,” only three out of 10 teachers improve substantially over a two-to-three-year period. Two out of 10 actually get worse. The researchers found “no evidence that any particular kind or amount of professional development consistently helps teachers improve.”
The problem is not that we don’t know what effective professional development looks like. Researchers have been publishing that list for decades. The problem is that the most important element on the list, sustained coaching during implementation, is the one most districts skip because they cannot staff it. When PD stops at the workshop and never reaches the classroom, knowledge does not become practice.
We Know What Works. We’ve Known for Decades.
In 2017, Darling-Hammond, Hyler, and Gardner published a comprehensive review through the Learning Policy Institute that examined 35 rigorous studies linking professional development to changes in teaching and student outcomes. They identified seven features of effective PD: it is content-focused, incorporates active learning, supports collaboration, uses models of effective practice, provides coaching and expert support, offers feedback and reflection, and is sustained over time.
This is not new information. Bruce Joyce and Beverly Showers established the research base for most of these features in the 1980s. Their work on training transfer found five components that contribute to teacher learning: presentation of theory, demonstration of practice, practice in a simulated setting, structured feedback, and coaching.
The critical finding from Joyce and Showers’ research, replicated many times since, is what happens to implementation rates depending on which components are included. Workshop-style PD that covers theory and demonstration but stops there produces implementation rates below 20 percent. When sustained coaching is added, implementation jumps to 80 to 90 percent.
That gap, from less than 20 percent to more than 80 percent, is not explained by better content or more engaging presentations. It is explained by what happens after teachers leave the training room and try to use what they learned with actual students.
The Implementation Gap Is the Whole Problem
TNTP’s “The Mirage” studied three large school districts and one charter network over two years. The study’s title refers to the finding that improvements in teaching from professional development are largely an illusion. They appear to happen in small pockets but do not show up consistently across teachers, PD types, or time periods.
The study is often cited as evidence that PD “does not work.” A more precise reading is that PD as currently delivered does not work, because it focuses on the part that is easy to deliver (workshops, presentations, conferences) and skips the part that actually changes practice (coaching during implementation).
This distinction matters. When Kraft, Blazar, and Hogan conducted their 2018 meta-analysis of 60 causal studies on teacher coaching, published in the Review of Educational Research, they found pooled effect sizes of 0.49 standard deviations on instructional quality and 0.18 standard deviations on student achievement. Coaching works. The evidence is strong. The challenge is delivering it.
A more recent meta-analysis by Sims, Fletcher-Wood, and colleagues, published in the Review of Educational Research in 2025, examined 104 randomized controlled trials and proposed the I/M/T/P framework. Their finding: professional development programs that incorporate mechanisms across all four functions, instilling insight, motivating change, developing techniques, and embedding practice, show approximately three times the student-level effect of programs that do not. The “embed practice” step is where coaching lives. It is also where most PD programs run out of budget and bandwidth.
Why Districts Cannot Deliver Coaching at Scale
The math is straightforward. A full-time instructional coach can typically maintain a caseload of 15 to 25 teachers, depending on the model and the depth of coaching. A school with 50 teachers needs two to three full-time coaches just to provide regular cycles. A district with 500 teachers would need 20 to 30 coaches. At typical coach salaries, that is $1.5 to $3 million in annual staffing costs for coaching alone.
Most districts have a fraction of that coaching capacity. So they make the rational decision: invest in PD events (workshops, conferences, professional learning days) that can reach all teachers at once, and hope that teachers will implement what they learned on their own.
Joyce and Showers’ research predicts exactly what happens next. Without coaching support during implementation, fewer than one in five teachers will transfer the new skill into their regular practice. The training was not wasted. Teachers learned something. They just cannot turn that learning into a sustained change in how they teach, because the support structure drops away the moment they walk back into their classroom.
This is the pattern TNTP documented at $18,000 per teacher per year. Not bad PD. Incomplete PD.
What Implementation Support Actually Looks Like
When coaching does happen, it follows a recognizable pattern. A teacher tries a new instructional strategy. A coach observes or reviews the attempt. The coach asks questions that prompt the teacher to reflect on what happened: What did you notice about student responses? What would you change next time? The teacher identifies one thing to adjust. The cycle repeats.
This cycle is what closes the gap between knowing and doing. As we explored in “Why Teachers Need a Trainer, Not a Fitness Tracker,” coaching is the most effective and most expensive form of professional development a district can offer. Each coaching conversation requires preparation, observation, reflection, and follow-up. Multiplied across a full caseload, the staffing math is what stops most districts from getting coaching to every teacher every week.
This is where AI Coach by Edthena changes the math. AI Coach delivers a structured coaching cycle to every teacher in the district, on demand. A teacher records or reflects on a lesson, engages in a guided coaching conversation, identifies a specific area to work on, and develops an action step. The cycle mirrors what a skilled coaching conversation does. It asks questions, prompts reflection, and focuses on one change at a time.
AI Coach extends a district’s coaching capacity from a small caseload to every teacher. That shift is what makes the implementation support Joyce and Showers identified, the difference between 20 percent and 90 percent transfer, deliverable across the full staff rather than rationed to a handful of teachers per coach.
From Knowledge to Practice: What Districts Should Change
If your district’s PD budget is heavily weighted toward workshops, conferences, and training days, the research says you are investing in the part of the pipeline that produces knowledge but not the part that produces practice change. That does not mean you should stop those activities. It means you should pair them with implementation support.
Four shifts to consider:
- Audit your PD budget by function. How much goes to content delivery (workshops, presenters, conferences) versus implementation support (coaching, observation, feedback cycles)? If the ratio is heavily skewed toward content delivery, the research predicts low transfer.
- Add a coaching layer to your highest-priority PD initiatives. You do not need to coach on everything. Pick the one or two instructional strategies your district is investing in this year and ensure teachers have structured support while they implement them. The Joyce and Showers research says this is the difference between 20 percent and 90 percent adoption.
- Extend coaching capacity with AI Coach. Platforms like AI Coach deliver structured reflection and practice cycles to every teacher every week. The 48 weeks between two scheduled coaching cycles do not have to be an implementation dead zone.
- Measure implementation, not just participation. PD systems that track attendance and satisfaction surveys are measuring the wrong things. Track how many teachers are actively using the new strategy in their classrooms three months after the training. That is the number that predicts student outcomes.
Darling-Hammond’s seven features are the right list. The gap is not in knowing the list. The gap is in delivering feature number five, coaching and expert support, at a scale that reaches every teacher. Closing that gap is what turns professional development from an annual compliance ritual into something that actually changes how students experience school.
Ready to extend your coaching capacity? Learn more about AI Coach and how it helps teachers turn PD into practice.
Frequently Asked Questions
How much do districts spend on teacher professional development?
According to TNTP’s “The Mirage” study, districts spend an average of $18,000 per teacher per year on professional development. Across the 50 largest U.S. school districts, the estimated total is $8 billion annually.
What percentage of teachers improve from professional development?
TNTP found that only 3 out of 10 teachers improved substantially over a two-to-three-year period, while 2 out of 10 saw their performance decline. The researchers found no evidence that any particular type or amount of PD consistently produced improvement.
What makes professional development effective?
The Learning Policy Institute identified seven features: content focus, active learning, collaboration, models of effective practice, coaching and expert support, feedback and reflection, and sustained duration. Of these, coaching has the largest documented effect on implementation. Joyce and Showers’ research found that adding coaching to training increases implementation from under 20 percent to 80 to 90 percent.
Does instructional coaching improve student outcomes?
Yes. Kraft, Blazar, and Hogan’s 2018 meta-analysis of 60 causal studies found that coaching programs improve instructional quality by 0.49 standard deviations and student achievement by 0.18 standard deviations. A 2025 meta-analysis by Sims et al. of 104 RCTs confirmed that PD programs incorporating implementation support show approximately three times the student-level effect of programs that do not.
