Built to replace the spreadsheet, not add another one

The problem

Peer evaluation is a routine part of running group work: students rate their groupmates at the end of a term, and the teacher folds that feedback into a grade or a conversation. In most classrooms, that routine still runs on a Google Form. Responses land in a spreadsheet one row per submission, and the teacher has to combine them by hand, build a pivot table to get one row per student, and mail-merge a results summary to send out, assuming they send results back at all.

None of that is hard, exactly. It's just slow, repetitive, and easy to get subtly wrong: a formula reference that doesn't update, a student whose name is spelled two different ways across responses, a low-response student whose "score" is really just one classmate's opinion presented as data.

The approach

Marka is built by the developer behind MarkIT, an attendance app for the same kind of classroom. Marka keeps the workflow a teacher already runs (build a rubric, open an evaluation window, collect peer feedback, hand back results) and replaces every manual step around it with a real data layer: bulk CSV roster import instead of a class list typed by hand, per-student access codes instead of a shared Google Form link anyone can share, and a one-button Consolidated Final Report instead of a pivot table rebuilt every term.

It also goes beyond simple averaging: Marka applies the rater/ratee statistical methodology behind CATME, Purdue's widely-used academic peer-evaluation framework, to automatically flag free-riders, inflated self-ratings, scoring cliques, and rater bias, for the teacher's eyes only, before grades go out rather than after.

The one rule that shaped the most design decisions: nothing reaches a student until it's ready. A result built from one or two evaluators isn't shown with a "limited data" caveat. It's withheld entirely, because at that sample size a supposedly anonymous score is close to identifiable. Public notes go through moderation before release. Every table that touches a teacher's data carries that teacher's owner ID, enforced at the schema level, so isolation between classes and between different teachers' accounts doesn't depend on every query remembering to filter correctly.

What's built in

• Teacher-authored rubrics, analytic or holistic, reusable across every class
• Anonymous peer scoring with optional self-evaluation comparison
• n ≥ 3 evaluator threshold before any result reaches a student
• CSV roster import with a validation preview and bulk access-code printing
• Live in-progress rollups, coverage view, and a completion matrix while a period is open
• CATME-style pattern detection: free-riders, self-rating overconfidence, collusion cliques, and rater bias, teacher-only
• Consolidated Final Report with an executive summary, cross-period trend view, and CSV export in one generation
• Note moderation with bulk-approve and auto-flagging
• No student email, ever: access codes only, per-tenant data isolation enforced at the schema level

Get in touch

For questions, feedback, or upgrade requests, email us at [email protected].

Marka is a small, focused product. We read every email.