Back to blog
Buying Guide
4 min read

How to Choose Teacher Observation Software in 2026

Updated

Score tools on workflow fit, rubric fidelity, follow-up speed and procurement readiness — in that order. Includes a weighted scorecard for demos.

In short

Choose on workflow fit first, rubric fidelity second, follow-up speed third, procurement readiness fourth. Feature count is close to irrelevant — the tools that fail in districts fail on adoption, not on missing capabilities.

Choose observation software on whether your observers will actually use it, not on how many features it has. Rank candidates on workflow fit, rubric fidelity, follow-up speed, and procurement readiness. Anything else is a tiebreaker.

Nearly every tool in this category can capture notes and produce a score. The ones that fail in districts do not fail because a feature was missing. They fail because the workflow did not match how observers actually work, so people quietly went back to a notebook.

What should you evaluate first?

Whether one observer can complete an entire observation — open the form, capture evidence, score, write next steps, and send — in a single sitting without switching tools.

Ask for that specific demo. Not a feature tour: one continuous run-through of a real observation, start to finish, in real time. Watch for the moment the salesperson says "and then you would export that" or "that part happens in your SIS." Every one of those is a handoff, and handoffs are where observations stall.

Count the clicks and count the tool switches. A tool that takes 40 clicks to complete an observation will be abandoned in November regardless of how good the analytics dashboard looks in March.

How much does rubric support really matter?

Enough to disqualify a tool. If observers have to mentally translate your district's language into the vendor's, inter-rater reliability drops and teachers lose trust in the scores.

Be precise in this conversation. Frameworks like Danielson and the Marzano Focused model are published intellectual property, so most vendors ship structurally similar templates in generic language rather than licensed framework text. That is usually fine. What is not fine is a fixed structure you cannot change.

The question is not "do you support Danielson?" — it is "can I reproduce my rubric exactly, with my domain names, my criteria, and my scoring levels, without rewriting it?" Ask them to build one of your domains during the demo.

What is the scorecard for comparing tools?

Weight workflow fit at 30%, rubric fidelity at 25%, follow-up speed at 20%, procurement readiness at 15%, and reporting at 10%. Score each candidate 1–5 and multiply.

CriterionWeightScore 5 looks likeScore 1 looks like
Workflow fit30%One observer completes a full observation in one sitting, on the device they carryEvidence in one place, scoring in another, feedback in email
Rubric fidelity25%Your exact domains, criteria and scoring levels, configured by youFixed vendor rubric; your language has to be translated
Follow-up speed20%Teacher receives written feedback the same day, from inside the toolWrite-up is a separate document, manually sent
Procurement readiness15%FERPA alignment, a data protection addendum, role-based access, clear data ownershipVague answers; addendum has to be drafted from scratch
Reporting10%Patterns across a grade or school that inform PD planningPer-observation PDFs and nothing aggregate
Weighted scorecard. Take it into demos and score live — retrospective scoring drifts toward whoever presented most recently.

How should you handle pricing models?

Model the total at your real headcount, not the headline number. Per-teacher and per-observer pricing scale very differently for a district than for a single school.

  • Per-teacher pricing punishes growth and makes district-wide rollout expensive — model it at full headcount, not pilot size
  • Per-observer pricing is cheaper up front but discourages adding coaches and department heads as observers, which is usually the behaviour you want to encourage
  • Flat school or district pricing is easiest to budget and does not penalise adding users. Voxento's schools plan includes unlimited users for this reason
  • Always ask what happens at renewal, and whether the quoted rate is introductory
  • Ask about implementation and training fees separately — they are frequently not in the headline price

What do people forget until it is too late?

Procurement and data protection. A tool that clears every instructional requirement can still die in legal review in March.

  • FERPA alignment, and whether a school data protection addendum already exists or has to be negotiated
  • Who owns the observation data, and what you get back if you leave
  • Role-based access — can a department head see only their own team's observations?
  • Retention: how long records persist, and whether that matches your evaluation agreement
  • Whether the tool touches anything union-negotiated in your evaluation process

Bring these to the first conversation, not the last. Vendors who answer them crisply have been through district procurement before; vendors who improvise have not, and you will spend months being their education.

For the procurement conversation specifically, the companion piece on questions to ask before buying observation software has the full list. If you have not settled on a framework yet, decide that first — the comparison of Danielson, Marzano and CLASS covers it.

Frequently asked questions

How long should a software evaluation take?
Plan for one term. Shortlist three tools, run one continuous end-to-end demo each, then pilot the leader with two or three observers for a few weeks before committing.
Should we pilot with our most enthusiastic observers?
No. Pilot with a sceptical, time-poor assistant principal. Enthusiastic early adopters will make almost any tool look workable, which tells you nothing about adoption.
Is AI-assisted feedback worth paying for?
It can speed up drafting, but it does not fix the underlying problem in most districts, which is turnaround time caused by handoffs between tools. Fix the workflow first; treat AI drafting as a bonus rather than a differentiator.
What if we already have an evaluation system we cannot replace?
Then you are buying an observation layer, not a replacement. Be explicit about that in demos and ask specifically how records move between the two — that handoff is where these deployments usually break.

Written by

Muhammad AminCo-founder, Voxento

I co-founded Voxento and build the platform. I work directly with the schools and training teams running observations and AI roleplay on it, which is where most of what I write here comes from.

Related reading

Get started

Put these ideas into practice with Voxento.

See how observations, evaluations, coaching, and walkthroughs work in one workspace.

Made in USA

Hosted in USA, GDPR compliant.

White Label Solution

Fully customizable platform with your branding and requirements.

Just start

Ready to use in minutes. No prior knowledge required.

Real support

Personal, fast and with real people.