How TAFtrack works, end to end

TAFtrack holds scans of pupil writing and lets teachers tag evidence directly on the image against the teacher assessment framework "pupil can" statements. It then rolls that up into coverage — how many tags, across how many pieces, for each statement.

Coverage is arithmetic. Judgement is human. The app will tell you that a pupil has three pieces of evidence for a statement and two gaps in a strand. It will never turn that into a standard. Awarding working towards, expected or greater depth is a separate, deliberate record that you write in your own words.

TAFtrack supports formative assessment, moderation and internal tracking. It is not a statutory assessment system and does not submit anything to a local authority or the DfE.

1. 1.Ask your senior lead for the school join code — a short code such as WHSPS.
2. 2.Go to the sign-in page, choose to create an account, and enter your name, email, a password and the join code.
3. 3.Your account is created against that school and you can sign in immediately; there is no confirmation email to wait for.
4. 4.Your senior lead then allocates you to the classes you teach. Until that happens you will see the school's pupils listed but cannot open their records.

Teacher

Uploads work, tags evidence, records assessments and rationales for pupils in classes they are allocated to. Sees other pupils only as names in a list.

Moderator

Reads across the school for moderation, adds moderation notes and sets moderation status on assessments.

Senior lead

Everything a moderator can do, plus staff management, class and pupil administration, retention settings and school-wide reports.

Designated safeguarding lead (DSL)

Sees and triages safeguarding concerns raised anywhere in the school, and the access log.

Global admin

Cross-school control: schools, roles, AI model settings and the framework criteria editor.

Pupils are records, not users. Children never have accounts and cannot sign in.

A class belongs to a school and has a key stage. Any number of teachers can be allocated to the same class, so job-shares and team-taught classes both work — every allocated teacher can upload, tag and complete assessments for those pupils.

Allocation is enforced in the database, not just hidden in the interface. A teacher who is not allocated to a pupil's class can see the pupil's name in a school list but cannot open the portfolio, the scans, the tags or the assessments.

• Move a pupil between classes from the pupil record — use this when a child changes class or moves up a year. History follows the pupil, not the class.
• Allocate teachers to a class from the class page or the Staff area, depending on your role.
• Pupil groups (pupil premium, SEND, EAL, gender, and other markers your school chooses to record) are edited from the pupil record and drive the group filters in Analysis.

Only record group markers your school actually needs for assessment analysis. Everything held here is children's personal data.

1. 1.Open the pupil record and choose to add a piece of writing.
2. 2.Give it a title, a genre and a date, and set the independent writing flag — this is required, because independence changes what the evidence is worth.
3. 3.Add one or more images. A piece can have several pages; they stay together as one piece.
4. 4.Save, then open the piece to start tagging.

Phone capture: on the pupil record, generate a phone capture link. It shows a QR code you scan with a phone camera. The phone opens a stripped-back capture screen already scoped to that pupil, so photographs land in the right record with no typing and no emailing images to yourself. You can start a new piece or append pages to an existing one.

• Images are downscaled in the browser before upload, so a full-resolution phone photograph does not stall on school Wi-Fi.
• Scans are stored in a private bucket and only ever served through short-lived signed links.
• An upload can be deleted from the piece screen; this removes the image files and the tags attached to them.

1. 1.Open a piece. The scan appears with zoom and pan, and the statement checklist for the class's key stage sits beside it.
2. 2.Drag a box around the part of the writing you want to evidence.
3. 3.Attach one or more statements to that box — a single sentence can evidence several statements at once.
4. 4.Set the polarity for each statement on the box, and add a comment if the evidence needs explaining.

Shows

Clear, secure evidence of the statement in this region.

Partial

Present but inconsistent, or attempted without control.

Not shown

An opportunity arose here and the feature is absent.

Contradicts

Evidence against the statement — errors that count against consistency.

Boxes are stored as percentage coordinates, so a tag stays on the right words if the piece is re-scanned at a different size. Handwriting is not box-tagged: it is a holistic rating recorded once per piece, because handwriting is judged across a whole piece rather than at one point on the page.

The coverage panel beside the script updates as you tag, and is kept visually separate from anything to do with judgement.

Where a school leaves it enabled, an uploaded scan can be read once by the AI model to propose candidate evidence: a region, a suggested statement, a suggested polarity and a one-line reason.

• Suggestions appear as dashed ghost boxes, deliberately different from confirmed tags.
• Accept, edit the box, change the polarity, or dismiss. Only accepted suggestions become evidence, attributed to you, with the fact that it started as an AI suggestion recorded for audit.
• Every accept or dismiss can be undone from the toast that appears straight afterwards.
• Running the suggestion pass twice will not duplicate work — overlapping candidates are suppressed.
• Handwriting and spelling-consistency statements are excluded from suggestion: they are holistic and the model would be guessing.
• Any transcription produced is stored as low-trust text used for search only. It is never shown as the thing being assessed.

If the AI provider is rate-limited, blocks the request, or rejects the configured model, you will see a visible "suggestions unavailable" state. Tagging continues to work normally by hand. A school can switch AI reading off entirely in Safeguarding, and a global admin sets which model is used in the Admin console.

A pupil is assessed many times as they move through the school, so assessments hang off the pupil record rather than replacing one another. Each assessment records the academic year, the term, an optional label, the standard awarded, a written rationale, any acknowledged gaps, who decided and when.

1. 1.Open the pupil record and choose to record a new assessment, or open an existing one to continue it.
2. 2.Pick the academic year and term. A second assessment in the same term needs a distinct label so the two can be told apart.
3. 3.Choose the standard, including below working towards where that is the honest answer.
4. 4.Type the rationale. Saving is blocked without a standard and a rationale, and that rule is enforced in the database, not just in the form.
5. 5.Set the moderation status: draft, internally moderated, or externally moderated.

An assessment can be deleted, but only after typing the pupil's full name to confirm. Where a pupil has no assessment yet for a key stage, the delete control is visible but disabled and explains why.

The pupil portfolio report is the moderation document: coverage by strand and standard, the cropped image regions for each tag with the teacher's comment, the handwriting rating, and then the teacher's judgement and rationale as a clearly separate final section. It is designed to print.

Gap report

Statements with no evidence yet, by pupil and by class — the fastest route to what to teach and collect next.

Strand heatmap

Coverage across pupils by strand, so thin areas stand out at cohort level.

Hardest statements

Statements where evidence is scarcest across the cohort.

Moderation readiness

Evidence density per pupil, showing who is ready to take to a moderation meeting.

Cohort outcomes

Distribution of standards awarded so far, filterable by group.

Independent writing

How much of the evidence base is flagged independent, which is what moderation will ask about.

• Reports use each pupil's most recent assessment by default, while the full history stays intact underneath.
• Filter by class, key stage, academic year and pupil group.
• Every report prints cleanly and exports to CSV for your own tracking system.

Children's writing is unusually revealing — a story can carry a disclosure. Any member of staff can flag something they have noticed, from the pupil record or the piece they are tagging.

• A concern is visible to the designated safeguarding lead, senior leadership and the person who raised it. Nobody else.
• Concerns cannot be edited away or deleted. Corrections are added underneath; acknowledgement, escalation and closure are all recorded with a name and a time.
• Every time a pupil record is opened, a concern is read, or data is exported, an entry is written to an access log that cannot be edited from inside the app.
• Retention defaults to 36 months — a key stage plus a moderation cycle — and is set per school. Where automatic deletion is on, scans and tags are removed at expiry while the judgement and its rationale are kept as the assessment record.
• A school can export everything held about one child as a single file for a subject access request, and can erase a child's record entirely. Only the fact of the erasure and its reason are kept.
• AI reading of scans can be switched off entirely per school in Safeguarding.

This is a record, not a referral. Urgent matters must still be raised in person with your designated safeguarding lead through your school's own procedures.

Schools

Create schools, edit names and see the join code staff use to register.

Staff

One combined list of staff with search and filtering, role assignment, and class allocation.

AI

Choose the model used for reading scans and the settings around it. If a key does not have access to the chosen model, the app automatically falls back to the first supported one.

Framework

The KS1 and KS2 "pupil can" statements, grouped by standard and strand. Statements are versioned: clone the current version to make changes rather than editing history in place, so old assessments still mean what they meant.

Criteria are keyed by key stage, not year group. A class carries its key stage, and the tagging checklist filters to match.

Join code not recognised

Codes are matched exactly. Check for a stray space and confirm the code with your senior lead in the Admin console, where it is shown against the school.

I can see a pupil but cannot open the record

You are not allocated to that pupil's class. Ask a senior lead to add you to the class; access appears immediately afterwards.

Upload fails or stalls

Usually the network. Try fewer pages at a time, or use phone capture, which downscales images before sending. Very large PDFs and unusual formats are not supported — use JPEG or PNG.

AI suggestions unavailable

Either the school has AI reading switched off, the provider is rate-limiting, or the configured model is not available to the key. A global admin can change the model in Admin. Tagging by hand is unaffected.

A report looks empty

Check the class, key stage and academic year filters, and confirm the pupils have at least one saved assessment — reports use the latest assessment per pupil.

Cannot save an assessment

A standard and a written rationale are both required, and a second assessment in the same term needs a distinct label.

TAF

Teacher assessment framework — the national framework of "pupil can" statements used for writing at the end of a key stage.

Pupil can statement

A single assessable claim, such as writing that uses paragraphs to organise ideas.

Strand

A grouping of related statements, for example composition, punctuation or spelling.

Standard

Working towards (WTS), working at the expected standard (EXS), or working at greater depth (GDS). Below WTS is also recordable.

Key stage

KS1 covers Years 1 and 2; KS2 covers Years 3 to 6. Statement sets differ between them.

Independent writing

Writing produced without adult prompting, modelling or scaffolding beyond what the framework allows. Flagged on every piece.

Coverage

The count of tagged evidence per statement across a pupil's portfolio. A prompt for judgement, not a judgement.

Moderation

Checking judgements against colleagues, internally or with other schools, so standards mean the same thing everywhere.