Teacher Evaluation
& Feedback Toolkit

It would be easy to treat teacher evaluation as a straightforward technical task — observe, rate, done. It isn't. It sits at the intersection of instructional quality, accountability, professional growth, employment decisions, and human relationships, and it's freighted with a history of approaches that either failed to differentiate anyone or overreached and backfired. Even defining 'good teaching' — the thing you're supposedly measuring — turns out to be contested. This toolkit takes the difficulty seriously rather than pretending it away, because the leaders who evaluate teachers well are the ones who understand what's genuinely hard about it — and who center evaluation on developing teachers rather than merely rating them.

In 2009, TNTP's landmark report 'The Widget Effect' — studying 12 districts and surveying ~15,000 teachers — found that more than 99% of teachers were rated 'satisfactory.' Evaluation happened rarely, against low standards, and produced almost no useful feedback and almost no consequences: excellence went unrecognized and poor performance went unaddressed. The report's memorable framing: if teachers are so important, why do we treat them like interchangeable widgets rather than professionals with real, meaningful differences? This 'indifference to performance,' it argued, disrespected teachers and gambled with students' futures. The report was extraordinarily influential — it directly shaped the Race to the Top reforms (§04) — and it remains the essential starting point: the problem it named (evaluation that neither differentiates nor develops) is the one good evaluation must solve. See TNTP.

Honesty about the reform era matters. In response to the Widget Effect, Race to the Top pushed states (with billions in federal grants) to adopt 'rigorous' evaluation systems incorporating student test-based value-added measures, and the Gates Foundation funded major research (the MET project). The systems became far more complex — multiple weighted components, test data, elaborate observation rubrics. And then the uncomfortable finding: rigorous research (including a major RAND study of the Gates-funded reforms) concluded these expensive, high-stakes reforms had little positive effect on student outcomes — and they generated significant backlash, with many states since rolling back the test-based components. The lesson isn't that evaluation doesn't matter; it's that accountability-by-measurement alone doesn't improve teaching. That reframing — from measuring teachers to developing them — shapes everything that follows.

The deepest tension in teacher evaluation is that it's asked to do two different jobs at once: accountability (judging performance against a standard, informing decisions about employment, pay, and tenure) and growth (giving teachers the feedback and support to improve). Both matter — but they can pull in opposite directions, because the same conversation can't easily be a supportive coaching dialogue and a high-stakes judgment (a teacher won't be vulnerable about their struggles with the person deciding their future, §13). The research points clearly one way: evaluation systems fail to foster improvement through accountability metrics alone, but multiple observations and detailed feedback can genuinely develop teachers — meaning evaluation works best as a developmental tool, not a screening device. Accountability still matters, but the wise emphasis is on growth.

Most evaluation systems are built on an observation framework — a detailed rubric describing dimensions of effective teaching. The most widely used is Charlotte Danielson's Framework for Teaching (four domains: planning and preparation, classroom environment, instruction, and professional responsibilities); Robert Marzano's model is another. At their best, frameworks provide a shared language for good teaching and clearer expectations. But two honest cautions: even Danielson herself has become critical of how her framework is used — misapplied as a rigid checklist, and used by evaluators who often 'don't have the skill to differentiate great teaching from merely good or mediocre.' And there's genuinely 'little consensus on how the profession should define good teaching.' A framework is a valuable tool — but it can't substitute for skilled, thoughtful judgment. See Danielson Group and Marzano Resources.

A core lesson of evaluation research (including the MET project): no single measure captures teaching well, so effective evaluation uses multiple measures. A single observation reflects one class, on one day, seen by one observer — far too thin a basis for a high-stakes judgment (and research found it's 'nearly impossible to differentiate performance using classroom observations alone'). Better systems combine multiple observations over time with student learning evidence (growth data, formative trends), professional-practice artifacts (lesson plans, assessments), and the teacher's own self-assessment and reflection. Combining measures produces a fairer, fuller picture, guards against the distortion any single measure introduces, and — crucially — provides richer information for growth. The point of multiple measures isn't more precise ranking; it's a truer, more useful understanding of teaching.

Since classroom observation is the one part of evaluation nearly every teacher experiences, its quality largely determines whether evaluation helps or harms. A strong observation process is a cycle, not a one-time drive-by: a pre-conference (understanding the lesson and the teacher's goals), the observation itself (gathering objective evidence of what happens — including what students are doing and learning, not just the teacher's performance), and a post-conference centered on reflection and growth (not just delivering a score). It should involve multiple observations over time (one class on one day is a thin sample), trained and calibrated observers (§12), and a focus on evidence over snap judgments. Make the observation a source of genuine, growth-oriented feedback (§10) — that's where its value lies.

Value-added measures (VAM) use statistics to estimate how much a teacher contributed to their students' test-score growth, factoring in expected gains. Proponents saw them as a more 'objective' counterweight to inflated observation ratings — and they were central to the reform era. But the problems are serious and well-documented: VAM relies on standardized tests (a narrow slice of learning), scores can be unstable and are only estimates, they can be unfair (especially to teachers of disadvantaged students), and they don't even apply to the many teachers in untested subjects and grades. Perhaps most tellingly for a growth-oriented system: VAM cannot identify specific areas for instructional improvement — it yields a number, not guidance a teacher can act on. Given the mixed evidence and the widespread rollback of test-based systems, treat VAM with real caution — as, at most, one limited measure among many, never the centerpiece.

Ask what actually improves teaching through evaluation, and the answer isn't the score — it's the feedback. Research is clear about what makes feedback developmental: teachers report that the most helpful aspects of evaluation are working collaboratively with their evaluator and receiving constructive, clear, and specific feedback — and that feedback lands better when it comes from an evaluator with genuine credibility and expertise. Feedback should be actionable (a teacher can do something with it), collaborative (a two-way conversation, not a verdict handed down), and forward-looking (aimed at the next step). Crucially, feedback works best when paired with coaching and support to act on it — a randomized study found that individualized coaching in response to observation feedback improved classroom practice. The goal of feedback is improvement, not judgment. See our Instructional Coaching toolkit.

The single most important shift in effective evaluation is from ranking to growth: the purpose is to help teaching get better, not merely to sort teachers or produce a rating. This means emphasizing the formative dimension (ongoing, low-stakes, developmental feedback throughout the year — 'the coaching that happens during practice') over the summative (the end-of-cycle rating, which is necessary for accountability but isn't where improvement happens). It means connecting what evaluation reveals to targeted professional development (if teachers consistently struggle with, say, formative assessment, provide learning there — not generic training). And it means treating teachers as professionals in a partnership, which turns evaluation 'from a moment of anxiety into' a genuine tool for growth. Rankings can inform decisions; only feedback and support actually improve teaching.

Here's a truth that surprises many: for all the (justified) worry about the variability of value-added scores, classroom observation is arguably less reliable in practice — one expert called it a 'statistical Wild West.' Different observers watching the same lesson often assign very different ratings; scores drift high (inflation) and inconsistent; and evaluator bias (the halo effect, favoring teachers similar to oneself, general leniency) quietly distorts judgments. The remedies: rigorously train and calibrate observers (norm them to the rubric with shared exemplars), check inter-rater reliability, and actively guard against your own biases. But one crucial caveat: an observer can get better at applying a rubric without getting better at helping a teacher improve — the gap between calibrated scoring and meaningful developmental feedback is often large. Pursue both reliable scoring and genuine feedback skill.

A principle that connects this toolkit to our Instructional Coaching toolkit: evaluation and coaching are fundamentally different roles. Evaluation judges performance for accountability; coaching develops practice through a safe, non-evaluative partnership. The problem with blending them is well-documented: a teacher will not honestly reveal their struggles or take real risks with the person who controls their employment — so when the evaluator is also the 'coach,' trust collapses and the developmental conversation becomes superficial. That's why, where possible, the two roles should be played by different people (the principal evaluates; a non-evaluative coach develops). This doesn't mean evaluation can't be growth-oriented — it should be (§11) — but even the most supportive evaluation isn't the same as coaching, and pretending otherwise shortchanges both. See our Instructional Coaching toolkit.

Evaluation done to teachers rarely develops them; evaluation done with them does. The most effective systems engage teachers as active participants through self-assessment (teachers reflecting honestly on their own practice against the framework), reflection, and goal-setting (teachers identifying, with support, what they want to work on). This isn't just pleasant — it's more effective: professionals who take ownership of their growth, reflect on their own teaching, and pursue goals they helped set improve more than those who simply receive a rating. Pairing teacher self-assessment and goals with observation and feedback creates a genuine partnership, treats teachers as the professionals they are, and puts the energy of the whole process where it belongs — on the teacher's own commitment to getting better.

When evaluation reveals genuine underperformance, the temptation is to avoid the hard conversation — to soften, to rate 'satisfactory,' to move on. But that avoidance is precisely the Widget Effect that failed everyone: it denies the teacher the honest feedback they need to improve, denies students an effective teacher, and is ultimately disrespectful. Difficult conversations, done well, combine honesty with support: be direct and respectful (clarity is a form of kindness — vagueness leaves people confused and stuck), ground everything in specific evidence rather than opinion, offer real support and resources to improve, establish a clear improvement plan with a reasonable timeline, and document fairly throughout. The goal is genuine improvement wherever possible. Facing real problems honestly and supportively is hard — and it's exactly what the job requires.

This is the part of evaluation no one enjoys but leadership can't dodge: occasionally, despite genuine support, feedback, and time, a teacher isn't effective and isn't going to become so — and students, who get only one shot at each grade, deserve better. Failing to act (the Widget Effect) protects an adult at children's expense. But acting fairly is non-negotiable: dismissal should be a last resort, reached only after providing real support and a genuine chance to improve (a fair improvement plan, resources, time), with thorough and consistent documentation, full due process, and strict adherence to district policies and contracts. Hold the tension between genuine care for the teacher as a person and your duty to students — both are real. This is among the hardest things a school leader does; do it with integrity, fairness, and resolve. See our School Leadership toolkit.

A key design tension: the higher the stakes attached to evaluation (pay, tenure, dismissal), the stronger the accountability pressure — but also the more it backfires for growth. Research shows that satisfaction with evaluation declines as it becomes heavily compliance-driven and high-stakes, and that when teachers perceive the process as unfair, they disengage from the feedback itself — triggering defensiveness rather than reflection, and making professional growth harder, not easier. High stakes also incentivize hiding problems rather than surfacing them. This is a big part of why the field has moved away from the high-stakes, test-based systems of the reform era. The wise approach keeps the developmental work as low-stakes as possible (so teachers can be honest and reflective), while reserving genuine accountability consequences for the relatively rare cases that truly require them (§15–§16).

Teacher evaluation tends to fail in two opposite directions, and both are traps. On one side, the Widget Effect: evaluation as a soft compliance ritual — inflated ratings, avoided hard conversations, no meaningful feedback, no growth, no accountability. On the other, the reform era's overreach: high-stakes, test-heavy, 'gotcha' evaluation that breeds defensiveness, gaming, and distrust while doing little to improve teaching. The other classic pitfalls cluster around these: relying on a single observation or measure, using uncalibrated and biased observers, over-relying on shaky VAM, and — subtly — producing ratings without development (the whole point). The path between the traps is the theme of this toolkit: credible, multiple-measure, growth-oriented evaluation that's honest about performance and genuinely helps teachers improve.

Leaders determine whether evaluation improves teaching or becomes a hollow ritual (or a feared weapon). Evaluating teachers well means: centering growth over ranking; giving specific, actionable, and credible feedback (which requires building your own skill at observing teaching and coaching — a skill many evaluators lack); using multiple measures rather than a single observation or test score; connecting evaluation to coaching and targeted support so feedback becomes improvement; keeping the developmental work as low-stakes as possible; calibrating your judgments and guarding against bias; and, when necessary, facing underperformance honestly and fairly. Above all, treat teachers as professionals and the whole process as a partnership for better teaching. Get this right, and evaluation becomes one of your most powerful tools for improving learning. See our School Leadership, Instructional Coaching, and Professional Growth toolkits.

For frameworks, the Danielson Group and Marzano Resources offer the leading observation models; TNTP's 'Widget Effect' is the essential (and reform-era research the essential critical) reading. Learning Forward, ASCD, and Edutopia offer practical strategies. Keep the through-line in view: center growth over ranking, use multiple measures, give credible developmental feedback, keep coaching distinct from evaluation, and be honest about performance. Pair this with our Instructional Coaching, Professional Growth & PD, and School Leadership toolkits. Start at K12academics.com.

Editable versions of these guides are available on request — see §26, Stay Connected.