Educator Growth & Evaluation: User Manual

Part of the Leadership Hub. For how this tool sits beside the rest of the suite, see For school leaders in the AlloFlow teacher guide. That chapter is the suite overview; this is the operating manual for one tool.

17 sections about 25 minutes end to end updated August 2026 prints to about 20 pages

Where to start

Evaluators and principals: read sections 1 through 7. That is everything you need to run a cycle.

Educators being evaluated: section 8 is written for you, and section 10 shows the summary you will receive.

District IT and administrators: section 9 is the deployment, and sections 11 and 12 cover privacy, data location, and backups.

1. Choosing Your Version

Both versions run the same evaluation workspace. The difference is where records live and who can see them.

On-device workspace compared with the district portal
	On-device workspace	District portal
Who it is for	A single evaluator organizing their own walkthroughs, observations, and trends. Good for getting started today.	A district that wants shared two-way records, where teachers see and acknowledge their own records under their district sign-in.
Setup needed	None. Open it and go.	A one-time Apps Script deployment by a district administrator, with IT permission.
Where data lives	Only on your device. Nothing is uploaded.	In the district's own Google Workspace: a private Drive folder owned by the district deployment account.
Teacher access	None. Role switching previews the teacher view on your device; it is not a sign-in.	Real. Each teacher signs in with their district account and sees only their own published records.
Drive sharing	Not available. Export a growth snapshot file instead.	Released summaries are created as Google Docs and shared to the teacher's Drive, with an opened receipt.
Official record	No. It is a working notebook, not the personnel record.	Closer, but the district's authorized personnel system remains the official record unless the district decides otherwise.

A good path: start in the on-device workspace to learn the tool and sharpen your walkthrough practice. When your district is ready for shared records, connect the portal. Your habits transfer directly, because the screens are the same.

2. Quick Start

Three ways in, all the same tool. Inside AlloFlow, open Educator Tools → Leadership Hub → Educator Evaluation — that is the route the Leadership Hub documents and the one to teach colleagues. It is also still reachable from AlloFlow's Project Settings with Open Educator Evaluation. Or go straight to alloflow-cdn.pages.dev/educator-evaluation in any modern browser, which is the link to bookmark or hand to a colleague. Once the tool is open, Manual in its header returns you to this page from any tab.

First launch asks how you want to begin. Both choices stay entirely on your device.

Choose a starting point. Explore simulated data opens a fictional roster with example statuses, ratings, and workflow history, which is the fastest way to see how the cycle works. Start with a blank workspace preloads nothing.
Set up the workspace. On the Setup tab, enter your organization, building, academic year, and evaluator name, then choose your evaluation framework (section 6).
Add educators. On the Staff tab, add the educators you evaluate along with their assignment type, which drives the correct scoring composition.
Start observing. Log walkthroughs as you do them, and assign formal observations when the cycle calls for them. Trends build themselves.

Everything saves automatically on your device as you work. There is no account and no server in this version. Before you rely on it for a full year of records, read section 12 on backups.

3. A Tour of the Workspace

Evaluators see eight tabs. The Overview is the daily home base: what is coming due, how far the roster has progressed, and how the selected educator's final evaluation is composed.

Evaluation overview tab showing a Coming due card with counts for past due, due within 14 days, and due in 15 to 30 days; a Teachers evaluated donut reading 2 of 8 finalized; and a Weight in final evaluation donut showing Observation and Practice 70 percent, Building Level Data 10 percent, Teacher-Specific Data 10 percent, and LEA Selected Measure 10 percent. — The Overview tab. The weighting donut always shows the composition that applies to the selected educator's assignment type.

Overview: cycle progress, what is coming due, and the weighting for the selected educator.
Trends: growth over time and de-identified peer context (section 7).
Staff: the roster, assignments, and cycle status.
Walkthroughs: short, frequent, low-stakes visit notes (section 4).
Formal observations: the full ten-step cycle (section 5).
SPM / SLO: student performance measures and student learning objectives, where your framework uses them.
Reports & audit: the audit timeline, exports, and handoff (section 12).
Setup: workspace settings, framework choice, official references, the portal connection guide, and the Share by QR card.

A Manual link sits in the header on every tab, so this document is always one click away from wherever you are working.

A footer runs along the bottom of every screen as a standing reminder of the design commitments: no AI scoring, evidence and judgments stay separate, and published records are append-only.

What the tool deliberately does not do

It does not score anyone. Every rating is entered by a person, and the software performs arithmetic only.
It does not reproduce licensed rubric text. Component names organize evidence; the performance-level descriptors stay with your licensed materials.
It does not rank educators against each other. Cohort context is de-identified and suppressed for small groups.
It does not replace your district's personnel system, and it does not decide anyone's employment.

4. Walkthroughs: Evidence and Interpretation

A walkthrough is a middle-of-lesson visit that captures factual evidence. It does not replace a comprehensive observation and it never auto-scores a rubric.

Walkthrough observations tab. The left column lists visit records with badges reading Private draft and Published. The right column shows a published walkthrough for Teacher 03 with separate sections for Directly witnessed evidence and Interpretation or feedback, component tags 2B and 3D, and a conversation area noting that published comments are appended and cannot alter the original evidence. — Evidence and interpretation are recorded in separate fields, and drafts stay private until you publish.

Drafts are private. A visit marked Private draft is yours alone. Nothing reaches the educator's record until you publish it.
Publishing freezes the snapshot. A published walkthrough becomes a locked record. Later thoughts are appended as comments and cannot alter the original evidence.
Two separate fields, on purpose. Directly witnessed evidence is what you saw and heard. Interpretation and feedback is what you make of it. Keeping them apart is what makes a walkthrough defensible and useful to the educator.
Interpretations pay off later. They are mined into the released summary, so a strength you noticed in October is still visible in June.
Component tags (2B, 3D, and so on) organize evidence by framework component without reproducing licensed rubric text.

5. The Formal Observation Cycle

Formal observations walk a ten-step tracker in order, and the tool will not let steps be skipped silently. The tracker across the top of the record always shows where you are.

Formal comprehensive observations tab showing a record for Teacher 03 badged Step 8 of 10, with a progress tracker whose ten points read Assigned, Prework, Pre-conference, Observation, Evidence review, Reflection, Post-conference, Ratings, Acknowledged, and Finalized. The first seven points are complete, Ratings is the current step, and the panel beside it reports that the evaluator signed on August 13 and the record is awaiting teacher acknowledgment, which does not indicate agreement. An evidence map lists Domain 1 Planning and Preparation at 20 percent, Domain 2 Classroom Environment at 30 percent, and Domain 3 Instruction at 30 percent. — The ten-step tracker partway through a cycle. Completed steps are filled in, the current step is highlighted, and later steps stay dim until they are reached.

Assigned. The observation is created for an educator and stamped with the framework snapshot in force.
Prework. The educator submits pre-observation reflection.
Pre-conference. You meet and record it.
Observation. You observe and collect evidence.
Evidence review. Evidence is published so the educator can see exactly what was recorded.
Reflection. The educator responds to what was observed.
Post-conference. You discuss it together.
Ratings. You enter human ratings with a written rationale for each domain. Nothing is scored by AI at any point.
Acknowledged. You sign, and the educator acknowledges. Acknowledgment records that they saw the record, not that they agree with it.
Finalized. The record locks. Later context appears as appended comments.

Alongside the tracker, the evidence map organizes evidence by domain and shows each domain's weight within Observation and Practice. Component names appear, but rubric-level performance descriptors are not reproduced, because that text is licensed.

The educator statement

At any point before finalization, the educator can attach a statement in their own words. It is theirs: no evaluator can edit it, it appears verbatim under "In your own words," and it leads the released summary, ahead of any ratings. This mirrors the contractual right to attach a memorandum to a personnel record. Once the record is finalized, the statement is frozen with it.

6. Framework Profiles

Choose the profile on the Setup tab. Every record permanently remembers which framework and weights it was scored under, so changing profiles mid-year never rewrites old scores.

Setup tab headed Setup, sources, and sharing, with a link to this user manual. The Workspace setup card holds organization, building, academic year, evaluator name and initials, and an Evaluation framework dropdown set to Pennsylvania Act 13 with Danielson 2021, above a note that full performance-level rubric text is not bundled. An Official references card lists four links including the Pennsylvania Department of Education and the Danielson Group, above an amber note explaining that this workspace uses assignment-aware 70/10/10/10 rather than the older 50 percent observation model. — Workspace setup and the framework picker, with the official references for whichever profile is active.

Pennsylvania Act 13 (Danielson 2021)

Assignment-aware composition: 70/10/10/10 for classroom teachers, 80 percent Observation and Practice where Building Level Data is unavailable, and 100 percent Observation and Practice for temporary classroom teachers.

Portland, Maine (PEPG guidebook)

Uses the published guidebook's four performance levels (Unsatisfactory, Novice/Needs Improvement, Proficient, Excellent), all 22 Danielson components, a categorical decision matrix rather than a numeric average for the summative practice rating, and an evidence expectation of at least nine pieces. Confirm the details against your current district guidebook, because the district plan governs.

How the Portland matrix reaches a summative practice rating

Portland's summative practice rating is categorical, not an average. The tool checks four rules in order and stops at the first one that fits:

Decision rules, applied top to bottom
Result	When
Unsatisfactory	Any domain is rated Unsatisfactory.
Excellent	Two or more domains are Excellent and none is below Proficient.
Novice/Needs Improvement	Three or more domains are at Novice/Needs Improvement.
Proficient	Everything else: no more than two domains below Proficient, and none Unsatisfactory.

Because the rules run in order, a single low domain outranks several high ones. That is the intended behaviour of a matrix and the main way it differs from an average:

Worked examples, using the four domain ratings in order
Domain ratings	Result	Why
0, 3, 3, 3	Unsatisfactory	One Unsatisfactory domain decides it, even beside three Excellent domains. An average would have read 2.25.
3, 3, 2, 2	Excellent	Two Excellent domains and nothing below Proficient.
3, 3, 1, 2	Proficient	Two Excellent domains, but one sits below Proficient, so the Excellent rule does not apply.
1, 1, 1, 2	Novice/Needs Improvement	Three domains at Novice/Needs Improvement.
2, 2, 2, 2	Proficient	The ordinary case.

No summative practice rating is produced until all four domains are rated. A partially rated record stays blank rather than guessing.

Maine PEPG (district plan governs)

Maine systems are local. Your district plan, built with a steering committee that has a teacher majority and revised by consensus, defines the rubric, rating levels, category weights, and process. Enter your plan's Professional Practice and Student Learning and Growth split. Since the 2019 amendments, student learning and growth measures are a district choice rather than a state mandate. This workspace mirrors your plan; it never substitutes for it.

7. Trends and Why Old Scores Never Change

The Trends tab shows finalized ratings and workflow activity over time. Evidence text, comments, and rationales are never aggregated into a number.

Teacher trends and cohort context tab. An amber notice reads that this is a privacy-aware aggregate and not FERPA certification, that cohort values appear only when at least ten eligible peers contribute, that small groups are suppressed, and that results are descriptive and must not be the sole basis for personnel decisions. Below are trend filters for educator, metric, and a date range, then a longitudinal snapshot for Teacher 03 counting published walkthroughs, finalized formal observations, and released cycle snapshots. — Trends, with the cohort-suppression rule stated on the screen where the comparison is made.

Small cohorts are suppressed. Peer context appears only when at least ten eligible peers contribute, so no one can be identified by subtraction in a small building.
Comparisons are descriptive. The screen says plainly that these results must not be the sole basis for personnel decisions.
Prior cycles are reported separately from the current year, and released cycle snapshots are immutable.

The framework snapshot

Every record is stamped with the framework and weights that were in force when it was created. If your district changes frameworks, or you switch profiles to compare, the tool recalculates nothing retroactively: a record scored under Pennsylvania Act 13 keeps its Act 13 math forever, and a Maine record keeps its Maine math. This is why a trend line stays trustworthy across a policy change, and it is also why two records in the same list can be scored differently on purpose.

8. For Educators Being Evaluated

In the district portal, an educator signing in with their district account gets their own set of tabs, scoped to them alone. Colleagues' records are never visible.

Teacher view showing tabs My evaluation, My trends, My evidence, Formal observation, SPM / SLO, Timeline, and About. The My evaluation tab shows Teacher 03 badged Awaiting teacher, a weighting donut reading Observation and Practice 70 percent with three 10 percent measures, and a card titled Your statement for the record whose text box is open for writing, explaining that the statement appears verbatim in the released summary and that no one else can edit it. — The educator's own view. The statement box is theirs to write, and no evaluator can edit it.

My evaluation: how the final evaluation is weighted, current status, and the statement card.
My trends: your own growth over time.
My evidence: published walkthroughs and evidence, where you can add comments to the conversation.
Formal observation: where you are in the ten-step cycle, with prework and reflection.
SPM / SLO: your student performance measures and learning objectives.
Timeline: the dated record of what happened and when.
About: the framework in use and the official references behind it.

The district portal signed in as an educator. The banner reads District Google account with the educator's own address and Educator access. Only the seven educator tabs appear, the page shows that educator's own weighting donut badged In progress, and the statement box is open for writing. No roster and no colleagues' records are visible. — The same portal signed in as the educator. They see their own record and nothing else, with no roster and no way to reach a colleague's file.

Three things worth knowing as an educator. You see the same evidence your evaluator sees, as soon as it is published. Acknowledging a record means you have seen it, not that you agree with it. And if you disagree, or if there is context the ratings do not capture, write a statement: it leads the document, word for word, and no one can edit it.

You can download your own workflow summary and growth snapshot at any time. The teacher view cannot export the whole workspace or see organization-wide audit events.

9. Setting Up the District Portal

Get IT permission first. The portal must live in a district-owned Google account, never a personal one, and district IT and leadership should approve it like any system that touches personnel records, including LEA authorization, a privacy and security review, and approval of any rating forms.

The portal is a separate deployment from the AlloFlow Class Mailbox. The mailbox is deliberately open to anyone with the link, because homework is low-stakes. The evaluation portal is the opposite: district-domain accounts only, and it fails closed without one.

A district administrator copies apps_script/educator_evaluation/ (Code.gs, Portal.html, appsscript.json) into a district-owned Apps Script project.
They run the one-time setupEvaluationRepository with the district domain, bootstrap admin, educators, members, and evaluator assignments. The package README shows the exact call.
They deploy as a Web app with Execute as: Me and Who has access: users in your domain, then verify with verifyDeploymentIdentity().
Each user pastes the deployment's /macros/s/…/exec link into AlloFlow's Project Settings, or bookmarks it directly. Opening it signs them in with their district Google account. The server, not the link, decides their role and which records they see.

The district portal in an evaluator's browser. A green banner reads District Google account with the signed-in address, Evaluator access, and Saved to district repository, beside buttons for Refresh, Email educator a portal notice, and Share released summary to educator's Drive. The eight evaluator tabs and the overview cards look the same as the on-device workspace, and there is no role switch. — The connected portal. The signed-in identity replaces the role switch, because the server decides what each person can see.

After connecting, the Setup health panel confirms the deployment is configured correctly, checking the domain lock, repository files, and assignments before anyone relies on it. It also re-hashes the audit log and reports whether any entry has been edited, deleted, or reordered since it was written, so an administrator can confirm the record is intact without opening the script editor.

In the district portal, when an evaluator releases a finalized evaluation:

The summary is generated as a Google Doc in a private "Released evaluations" folder owned by the district deployment account.
The teacher's district account is added as a viewer of that one document, so it appears in their Drive under "Shared with me."
Email notifications are content-free, saying only that there is portal activity. No ratings or details travel by email.
When the teacher opens their summary, an opened receipt is recorded, so both parties can see the evaluation was actually received.

Sharing is a deliberate evaluator action at release time, never a silent background job. In the on-device workspace, use Export growth snapshot instead: it produces a formative, ratings-free file you can hand to an educator directly.

What the document actually says

Below is a real summary produced by the tool, abridged, using fictional names. Notice the order: the educator's own words come first, strengths come before growth areas, and the arithmetic is explained rather than asserted.

Educator Effectiveness Summary — 2026-27

Prepared for Teacher One on 2026-08-13 by Principal Rivera.

This document is a plain-language summary of your finalized evaluation. It is shared view-only with you and your evaluation team — no one else. The district portal remains the official record and holds every observation, note, timestamp, and revision behind this summary; nothing here is hidden from you there.

In your own words

This year I rebuilt my small-group reading block so every student conferences with me at least twice a month. Attendance in my first period was uneven through the winter, and I want that context on the record alongside my ratings.

Written by you in the portal (2026-08-13); no one edited it.

Your strengths

Planning and Preparation — rated Distinguished (how the lesson and its goals were designed)
Plans name the standard, the misconception to watch for, and the check for understanding.

Classroom Environment — rated Distinguished (the respect, routines, and culture students experience)
Transitions are routine and student-run; the room lost under a minute across three transitions.

[ two further domains follow, each with its evaluator's written rationale ]

Walkthrough observation (2026-08-09)
The discussion-norms anchor chart is doing real work; students referenced it without prompting. Worth sharing at a team meeting.

Your overall rating, in plain language

Overall score: 2.69 out of 3, which is the "Distinguished" performance band. Bands are fixed statewide cut points: 2.50 and above is Distinguished, 1.50–2.49 Proficient, 0.50–1.49 Needs Improvement, below 0.50 Failing.

Component	Weight	What it measures
Observation & Practice	70%	Your observed practice across the four domains below.
Building Level Data	10%	Your building's performance data for the year.
Teacher-Specific Data	10%	The measures selected for your role and assignment.
LEA Selected Measure / SPM	10%	The measures selected for your role and assignment.

Your final score is the weighted average of these components — each score is multiplied by its weight and the results are added. No component is hidden and no other factor enters the calculation.

[ a domain-by-domain table follows, giving each rating in plain language ]

Growth focus

No component of your finalized evaluation was rated below Proficient.

Transparency and your rights

This summary was assembled only from the finalized records in the district portal: 1 finalized formal observation and your locked student performance measures. Every rating shown here was assigned by a person and carries that person's written rationale in the portal; the software performs arithmetic only.

You can read every underlying record, timestamp, and revision in the portal at any time.
You acknowledged the observation before finalization; acknowledgment records that you received it, not that you agree.
You can add a written response through the portal dialogue, and it becomes part of the record.
Finalized records are immutable — nothing in this summary can be edited after release without a new, visible record.

Questions about this evaluation go first to your evaluator or to Sample School District leadership. This copy is shared view-only to your district account; if any detail here disagrees with the portal, the portal record governs.

The wording adapts to your framework profile: a Maine or Portland workspace names that plan's rating levels instead of the statewide Pennsylvania bands.

11. Privacy, Records, and Boundaries

On-device workspace: nothing is uploaded, ever. Records live in your signed-in browser profile on this device, so anyone who can unlock that profile can open them. Use a device you control.
AlloFlow adds no encryption of its own. On-device records are protected by your device sign-in and whatever the device provides, which on a managed district laptop usually means the user profile is encrypted at rest. That is a question for IT, not an assumption to make.
Retention applies here too. Evaluation records are subject to your district's retention and discoverability rules wherever they live, including on your own laptop.
District portal: all data stays inside the district's own Google Workspace. The server code makes no external network calls, so nothing leaves the district tenant, which is what lets it operate under the district's existing Google Workspace for Education agreement.
Repository files are private by verified default. The deployment checks that its Drive folder and files are private and refuses to proceed if that cannot be confirmed.
Evaluations are personnel records. Treat them with the confidentiality your state law and collective bargaining agreement require. FERPA governs student records and enters here only if observation evidence names identifiable students, so keep evidence focused on educator practice.
Role switching is a preview, not authentication. Only the portal, with district sign-in, gives educators real authenticated access.
The official record: unless your district decides otherwise, official summative ratings and anything requiring two-party records belong in the district's authorized personnel system.
No AI scoring. Ratings are entered by people. The tool organizes evidence; it does not judge it.

12. Backing Up and Moving Devices

The on-device workspace lives in your browser's storage. That is what keeps it private, and it is also its main risk: clearing your browser data will erase it, and it does not follow you to another computer. The Reports & audit tab is where you protect against that.

Reports and audit tab headed Audit, reports, and handoff. The left column shows an audit timeline listing dated events including assigned, prework submitted, observed, evidence published, reflection submitted, and signed, each naming the actor and role. The right column holds an Export and transfer card with buttons for Export workspace JSON, Export status CSV, Workflow summary HTML, and Growth snapshot, an Import another device export section, and a warning that the district-authorized system remains the official summative rating record. — The audit timeline records who did what and when. The export card beside it is your backup and transfer route.

Export workspace JSON is your backup. Take one at the end of any working session you would not want to redo, and keep it in district-authorized storage.
Import another device export moves a workspace to a new computer. Import replaces the current workspace after validation, so export first if the destination has anything you want to keep.
Export status CSV and Workflow summary HTML are for reporting and handoff rather than backup.
Growth snapshot is the formative, ratings-free file you can share with an educator directly.

Exports can contain confidential personnel information. Store and transmit them only through district-authorized systems, never personal email or cloud storage.

In the district portal this concern disappears: records live server-side in the district's Drive, and the portal keeps its own tamper-evident audit log. Each audit row is hashed together with the hash of the row before it, so an administrator can run verifyAuditChain() at any time and find out whether any row has been edited, deleted, or reordered since it was written.

13. Using It on a Phone

Walkthroughs happen while you are standing in a doorway, so the workspace reflows to a single column on a phone. The tab strip scrolls sideways, cards stack, and buttons keep a large touch target.

The workspace at phone width showing the Walkthroughs tab in a single column: a horizontally scrolling tab strip, a large Start walkthrough button, and stacked visit records badged Private draft and Published, each showing the educator, date, duration, and note. — The Walkthroughs tab on a phone. Everything stacks into one column and nothing scrolls sideways except the tab strip.

Add to your home screen so the workspace opens in one tap, the same as any bookmarked web page.
The workspace is per device. A walkthrough logged on your phone stays on your phone unless you are signed in to the district portal, which syncs server-side. On-device workspaces do not sync between your phone and your laptop.
Draft on the phone, publish at your desk if you prefer. Drafts stay private until you publish them.

14. Accessibility

The workspace is built to the same standard as the rest of AlloFlow and is checked automatically on every change.

Keyboard throughout. Tabs, forms, dialogs, and record controls are reachable and operable by keyboard. The first-launch dialog holds focus until you choose, so keyboard and screen-reader users cannot land behind it.
Screen-reader structure. The tab strip and panels use a proper tab and tabpanel relationship, so each panel announces the tab it belongs to.
Touch targets are sized for a phone in a hallway between visits: primary buttons, inputs, and the role switch are at least 44 pixels tall, and no control falls below the 24 pixel minimum that WCAG sets. Both floors are checked automatically at desktop and phone widths.
Automated checks. Every screen in both roles is scanned against WCAG 2.1 AA rules as part of the test suite, and the tool ships with zero violations. This manual is scanned the same way.
Dark and light. Both this manual and the workspace follow your system's colour scheme.

Language

The evaluation workspace is currently available in English only, while the rest of AlloFlow is translated into many languages. Evaluation records are personnel records, and which language they are written in is a district policy question rather than a display setting, so translation is waiting on that decision rather than on the technology. If your district needs another language of record, that is worth raising before you adopt the portal.

Automated checks catch a great deal but not everything. If you hit something that does not work with your assistive technology, that is a bug worth reporting, not a limit you should work around.

15. Sharing the Tool by QR

The Setup tab includes a Share by QR card with a scannable code and a copy-link button.

Share by QR card on the Setup tab, showing a QR code beside the address https://alloflow-cdn.pages.dev/educator-evaluation and a Copy link button, with the caption that anyone can scan this to open their own private on-device workspace and that your data is not shared by the code. — The Share by QR card in the on-device workspace, pointing to the published page rather than to anything on your machine.

On-device version: the code points to the published workspace page, so a colleague who scans it gets their own private workspace. Your records are not in the code and are not shared by it.
District portal: the code points to your district's portal address. Each person's district sign-in still decides what they can see, so the code is safe to print on a staff-meeting handout.

16. Sending an Evaluation to an Educator by Email

The QR card above shares the tool. This section is about sharing one educator's evaluation with that educator, and getting their written response back, without any district server.

On Reports & audit, with an educator selected, choose Educator packet (send to educator). You get a single HTML file containing only that educator's records: their own walkthroughs, observations and student-measure records, and nothing that reveals which other educators exist. A checkbox lets you send codes instead of names.

Email that file to the educator as an attachment. They can simply open it and read their evaluation in a browser, with nothing installed and no account. If they want to respond, the same file contains a form: a statement in their own words, a reflection on each observation, and a box confirming they have read it. Clicking the button downloads a small response file, which they email back to you.

Import that response with the same Import control you use for a workspace file. It merges into the educator's record and the timeline notes where it came from.

What a returned file can and cannot change. Only the educator's own words are accepted: their statement, their reflections, and their read-receipt. Ratings, evidence and evaluator comments are ignored even if the file has been edited by hand, and the import tells you how many fields it dropped. If you changed a record after issuing the packet, the import says so rather than quietly overwriting what the educator replied to.

The receipt is not a signature. Ticking the box records that the educator read the evaluation. It is not the signature your district's evaluation form asks for, and it does not replace the conference where you and the educator both sign.

17. Using Your District's Own Rubric

The tool ships with three scoring profiles (Pennsylvania Act 13, Portland ME PEPG, and Maine PEPG). If your district uses a different instrument altogether, you can supply it.

On the About tab, under Rubric, Download current rubric gives you the active one as a JSON file to use as a starting point. Edit the domains, their components, weights, colours and rating-band labels, then use Load a custom rubric. Restore the built-in rubric puts it back.

A rubric is rejected rather than half-applied if any domain is missing an id or a label, or if two domains share an id, because a partly-valid rubric would mis-score everyone rated against it. A custom rubric is given a version tag so that records can say which rubric produced them.

Changing a rubric mid-cycle. Ratings are stored against domain ids. If a new rubric does not include a domain that already carries a rating, the tool names the affected educators and asks before applying it. Nothing is deleted: restoring the previous rubric brings those ratings back.

18. Evidence Checks and the Optional AI Second Read

When you assign ratings, the tool compares them against the evidence you tagged and reports what it finds: a domain rated with no evidence tagged to it, a rating below proficient resting on a single piece, domains with no evidence at all, and how your total compares with what the plan expects. Portland's guidebook, for instance, looks for at least nine pieces across a cycle.

This is counting, not judgment. It runs on the device, needs no AI, and never leaves. Favourable ratings are not questioned for thin evidence, and unrated domains are left alone. The point is narrow and practical: a rating that rests on little documented evidence is the one most likely to be overturned, so it is worth seeing before the record is finalised rather than afterwards.

The optional AI second read. Off by default. A district that permits it can enable AI reflection on the About tab. An evaluator can then ask a model whether the evidence they wrote supports the ratings they assigned, and what other readings that same evidence allows, including ones favourable to the educator.

Three boundaries apply and they are deliberate. The model is asked about the documentation, never about the educator, and is explicitly instructed not to assign, suggest or imply a rating. Its answer is advisory: it is shown to the evaluator and is never written into the record, so nothing a model produces becomes part of a personnel file. And enabling it does send that educator's evidence notes and ratings to your configured AI provider, which is why it is opt-in, per workspace, and why the educator's name is not included in what is sent. If your district or contract does not permit AI in evaluation, leave it off and everything else works unchanged.

Section 16 sends the packet as an email attachment, which needs nothing from anyone. This section is the optional upgrade: the same packet delivered through your Google Drive, so the share can be withdrawn on a date and every evaluation you send lands in one organised place.

It needs a small Apps Script that you deploy in your own district account. It is not the district portal: it has no roster, no roles, and no repository, and only you can open it. The setup steps are in apps_script/educator_evaluation_share/README.md.

Once deployed, you paste an exported packet into it along with the educator's email. It files the packet under AlloFlow Evaluations / academic year / educator, shares it with that educator as a commenter, and can end that access on a date you choose. A listing view shows everything filed for the year, so at the end of a cycle you move or copy one folder to wherever your district keeps evaluation records.

What it gives you that email cannot. An emailed attachment cannot be recalled. A Drive share can be withdrawn, and can expire on its own.

What to check before relying on expiry. Google only expires viewer and commenter access, only on some Workspace editions, and only within about a year. The helper has a check button that tells you whether expiry is available to your account before you depend on it, and after each share it says plainly whether the expiry was applied or whether the file is shared with no timer running.

What it does not do. It sends no email of its own, it cannot tell one person from another beyond you, and the folder lives in your account, so treat it as a working store and hand it over at the end of a cycle. Whether you may deploy Apps Script at all, and whether personnel records may be shared this way, are district policy questions rather than technical ones.

20. Glossary

Walkthrough: A short, frequent, low-stakes classroom visit that captures evidence. It does not score a rubric.
Formal observation: The full ten-step cycle with prework, conferences, evidence, reflection, ratings, acknowledgment, and finalization.
Observation and Practice (O&P): The classroom-practice portion of a final evaluation, as distinct from data-based measures.
SPM: Student Performance Measure. A district-selected measure that contributes a defined share of the final evaluation.
SLO: Student Learning Objective. A goal set for a group of students, used where the framework calls for it.
PEPG: Performance Evaluation and Professional Growth, the Maine term for a district's educator evaluation system.
Framework snapshot: The tag stamped on every record identifying the framework and weights used when it was created, which is what keeps historical scores stable.
Band: The label a score falls into, such as Proficient. In this tool a band is triage for planning, not a verdict about a person.
Released summary: The final strengths-first document shared with the educator, opening with their own statement.
Acknowledgment: A record that the educator has seen a document. It does not signal agreement.
Cohort suppression: Withholding peer comparisons until enough peers contribute, so no individual can be identified from an aggregate.

21. Troubleshooting

"This page must be opened from the district Apps Script web-app URL": the portal bundle was opened outside its deployment. Use the /exec link from your district administrator.
"The secure workspace could not be opened": this is the portal refusing to show anything, which is the design rather than a fault. The line underneath names the reason: an account that is not a member, a repository that was never set up, or the page opened outside its Apps Script address. Access is granted by the district, so the fix is with the administrator who deployed the portal, not on the page.
The portal looks out of date: the administrator should re-paste the current Code.gs and Portal.html and create a new deployment version. Apps Script serves the old version until a new one is deployed.
Setup health shows a warning: follow its specific message. It checks the domain lock, repository files, and assignments in order.
A finalized record needs a correction: finalized records are immutable by design. Add an appended comment, record a new entry, or have the educator attach a statement.
Scores differ between old and new records: that is the framework snapshot working as intended (section 7). Each record keeps the framework and weights it was scored under.
Peer context is missing on the Trends tab: fewer than ten eligible peers contributed, so the comparison is suppressed to protect identities.
The QR code shows a different address than you expected: outside a web address, such as inside the desktop app, the card falls back to the published page on purpose, because a local file path cannot be opened by anyone else.
Your workspace looks empty on another device or browser: the on-device workspace does not follow you, and clearing browser data erases it. See section 12 for exports and transfers.