Overview of the product specs (not a work item itself). Written 2026-10-02 from three idea documents in ideas/ (a commercialization strategy, a feature roadmap, and a high school data pipeline), the existing specs, and new research. Each spec below is a separate work item on the roadmap.
Why
Today the site is a public explorer of federal data: every number is cited, nothing is personalized, and nothing is
remembered between visits except the compare list in localStorage. The idea documents push in one direction: turn
the explorer into a planning tool for one family (a student's list, their chances, what they would pay) and
pay for it with subscriptions, counselor accounts, and an API. Two things are foundational and were asked for
explicitly: login with households (a parent sees each child's work; a child never sees the parent's finances)
and usage telemetry so the site's use can be reported on.
The site's rules stay: federal data is the baseline, every figure is cited with its year, nothing is graded, and the public data stays free (Commercialization: Free, Plus, and Pro).
Themes and specs
Accounts and households
| Spec | Adds | Complexity |
|---|---|---|
| accounts.md | Login, households (guardians and students), who can see what, account deletion | Large |
| student-profile.md | A student's own numbers (GPA, scores, major, state, preferences) that every tool reads | Medium |
| saved-lists.md | Saved colleges with Reach / Target / Likely, status, notes, deadlines, sharing, export | Medium |
| follow-colleges.md | Follow colleges (lists follow automatically) and get one email per data release summarizing what changed, with years; a public "What changed" panel on profiles | Large |
| home-and-distance.md | A home address on the account, Explore's "Distance from home" filter with a nearest-first sort, and the distance to every college on a saved list, for students and guardians alike (built 2026-10-05) | Medium |
| household-hub.md | Add a parent or a student by role with only the details that role needs; everyone by name, invited people in the same roster with their link; the household page as the hub, one list per person with "updates" as a switch per college (built 2026-10-06 from the owner's review) | Large |
The application plan written on 2026-10-06 became the timeline stage of the planner on 2026-10-07 (The Planner: From the First List to the Deposit): six stages from the first list to the deposit, in their own section of the roadmap, with the award-letter analyzer as the last stage.
Planning tools
| Spec | Adds | Complexity |
|---|---|---|
| Chances and Fit: Where You Stand | "Where you stand" at each college from the student's numbers; a transparent rules-based classification, never a probability | Large |
| cost-by-income.md | Built 2026-10-10 (estimates gated until the pilot): cost as a curve over family income: published prices to $110K, a calibrated estimate above, where need-based aid ends for each college, and what merit aid can still do; replaces the average-first cost display | Large |
| Net Price Estimator: What This Family Would Pay | What this family would pay: federal Student Aid Index plus each college's aid pattern, as a range, with a 4-year projection | Large |
| Early Decision and Early Action: Does Applying Early Help? | Early decision and early action admit rates vs regular decision, from Common Data Sets, with the caveats | Medium |
| ../planner/offers.md | Real aid offers side by side, standardized the way the federal College Financing Plan does, with loans and 4-year totals made visible; moved into the planner on 2026-10-07 as its decisions-and-offers stage | Large |
High school context
| Spec | Adds | Complexity |
|---|---|---|
| high-school-data.md | A high school dataset (NCES, EDFacts, Civil Rights Data Collection, state report cards, school profiles) and high school pages | Extra large |
| Scattergrams: Outcomes From Your High School | Counselor-uploaded application outcomes from one high school, scrubbed and thresholded, drawn over the college's ranges | Large |
Platform, measurement, and business
| Spec | Adds | Complexity |
|---|---|---|
| telemetry.md | Product analytics with a typed event registry, privacy rules for minors, and usage reports (built 2026-10-07) | Medium |
| Commercialization: Free, Plus, and Pro | Free / Plus / Pro tiers, what each gates, Stripe billing, entitlements, the pricing page | Large |
| Counselor Portal: Quad for Counselors | Organization accounts for counselors and consultants: caseloads, reports, scattergram uploads | Extra large |
| Data API: The Dataset, History, and Lineage for Developers | A keyed public API over the dataset, history, and lineage, with a paid tier | Medium |
| Admin Console: People, Hidden Values, Data Health, and Operations | Planned 2026-10-12: the admin console: people and households (search, details, email, password reset, deactivate, delete, link and unlink), the hidden selectivity index and a student's hidden odds, each college's data health by when its data was received, and reruns and data approvals through a GitHub App | Large |
How the idea documents map
| Idea document | Where it went |
|---|---|
| Commercialization strategy: tiers, pricing, pricing page | Commercialization: Free, Plus, and Pro |
| Commercialization strategy: B2B counselors, data API | Counselor Portal: Quad for Counselors, Data API: The Dataset, History, and Lineage for Developers |
| Commercialization strategy: Niche-style high school integration | high-school-data.md (data), Scattergrams: Outcomes From Your High School (feeder and scattergram use) |
| Feature roadmap A: real cost and aid estimator | Net Price Estimator: What This Family Would Pay |
| Feature roadmap B: localized scattergrams | Scattergrams: Outcomes From Your High School |
| Feature roadmap C: award letter evaluator | ../planner/offers.md (the planner's last stage, since 2026-10-07) |
| Feature roadmap D: major-level ROI | Already planned: field-of-study.md (Scorecard Field of Study). The Pro tier gates its deeper views (Commercialization: Free, Plus, and Pro) |
| Feature roadmap E: ED/EA strategy engine | Early Decision and Early Action: Does Applying Early Help?, on top of cds-admissions.md |
| High school data spec: sources, pipeline, use cases A–C | high-school-data.md |
| High school data spec: use case D, counselor portal | Counselor Portal: Quad for Counselors |
Added here beyond the documents: households with parent-only finances (accounts.md), a student profile shared by every tool, Reach / Target / Likely classification with published rules (Chances and Fit: Where You Stand), telemetry, award-letter renewability and appeal support, following colleges with update emails (follow-colleges.md, added 2026-10-02), and the privacy rules for minors' data that run through all of it.
Build order
telemetry ──────────────────────────────────────────────────────────┐
accounts ─► student-profile ─► saved-lists ─► chances-and-fit │
│ │ │
│ ├─► commercialization ─► counselor-portal
│ │
├─► net-price-estimator ─► planner/offers (the award-letter analyzer, now the planner's last stage)
│
├─► follow-colleges (lists follow automatically; pays off once the scheduled data refresh runs)
│
├─► household-hub ─► the planner (planner/README.md: model, list, rounds, actions, timeline, applications, offers, parents)
│
high-school-data ─► scattergrams (also after saved-lists)
cds-admissions (data expansion) ─► early-decision-strategy
data-api (independent)
Telemetry first: it is small, and every later feature's success is measured with it. Accounts next: the planning tools are only worth building once a family can keep their work. Following colleges is the first thing to build on accounts: it needs no personal numbers, gives people a reason to sign up, and pays off as soon as the scheduled data refresh publishes releases on its own. Within the planning tools, the net price estimator comes before the award-letter analyzer because the analyzer reuses its cost-of-attendance, loan, and projection code. The high school dataset is the biggest item and can proceed in parallel; scattergrams need both it and accounts.
Shared rules for user data
These apply to every spec here and are enforced in the database (row-level security), not only in the UI.
- Nothing about a person leaves the server unless they're allowed to see it. Each table has an owner; guardians see their students' academic data; students never see guardians' financial inputs (accounts.md).
- Minors. Nobody under 13 may hold an account (the Supabase and PostHog terms both exclude under-13s, and COPPA applies). Students 13–17 may; a guardian link is optional. No behavioral advertising, no selling data, no session replay, no profiling beyond the features the user asked for (telemetry.md).
- User data lives only in Supabase (new migrations, applied to dev before prod), never in
data/*.json(supabase.md). The JSON dataset remains the public, reviewed source for college data. - Estimates are ranges with their inputs shown, in the same spirit as citations: "estimated from your AGI of $X and this college's 2023–24 net price by income". Never a bare number that looks like a quote from the college.
- Export and delete. Every account can download its data and delete itself; deleting a household's guardian never deletes a student's own data.