Common App: Deadlines, Requirements, and Application Trends (Requirements Grid, Explore pages, research reports)

Status: planned. Specified 2026-10-06 from a survey of commonapp.org (every number below was measured that day; nothing is built). Waits on Common App's permission: its terms of use forbid the pull as written, so the Permission section comes first and the build starts only after a written yes. Reads alongside cds-application-logistics.md and cds-test-scores-and-policy.md (the same facts from each college's own Common Data Set, for the few colleges with one on file), feeds the application tracker in saved-lists.md, the early-rounds view in Early Decision and Early Action: Does Applying Early Help?, the aliases table in aliases.md, and the cycle watch idea.

Question it answers

When are this college's deadlines, and does it have Early Decision II? What does it cost to apply, and will it take a fee waiver? Do I need an essay beyond the Common App one, a writing supplement, two teacher recommendations, a mid-year report? Does it want test scores this cycle? And, for the trends pages: how is this application season going nationally, and in my state, long before federal data says?

Why

The site has these facts for almost nobody. Deadlines and early rounds come only from the Common Data Set records of the round-3 agent, which today cover a handful of colleges; the test policy comes from IPEDS, which describes the cycle before last; and nothing anywhere says whether a college wants an essay, a supplement, or recommendations. Common App publishes all of it, for every member, for the cycle that is open now, in one dated document. About half of the site's colleges are members.

Source

Four things on commonapp.org, in order of value. None is an API; the first two are public files, the third is a set of PDFs, the fourth an embedded Tableau Public workbook.

1. The Requirements Grid (per member, the current cycle)

https://content.commonapp.org/Files/ReqGrid.pdf, titled "2026-27 Requirements Grid", stamped "Updated: 09-25-2026" on every page, 55 pages, one row per member that takes first-year applications (about 1,128 rows). Columns, with the grid's own notes (page 55):

The grid is a layout PDF, not a table: a column parse by the header words' x-positions (pdfplumber) recovered 1,080 rows in the survey, with the Writing column mis-assigned (its "Y" lands in a neighbor; the fix is a narrower column window). Names, not ids: 1,060 of the 1,080 parsed names match an Explore page title exactly; the rest differ by a hyphen or a campus suffix ("University of Michigan- Dearborn", "Marian University - Indiana") and need the alias table or a hand map. What the parsed rows said, as a feel for the data (approximate, from the probe parse):

2. The Explore pages (per member, the join key and small facts)

https://www.commonapp.org/explore/<slug> is a Gatsby site, so each college also exists as https://www.commonapp.org/page-data/explore/<slug>/page-data.json, and https://www.commonapp.org/sitemap-0.xml lists all 1,174 slugs. The record has field_su_ipeds (the IPEDS unit id: the join to the site's unit_id), field_member_id, the address with latitude and longitude, admissions phone and email, first-year and transfer admissions and financial-aid URLs, a DACA aid page (183 members), a transfer-guarantee page (83), social links, field_su_alternate_names (860 members; "VU, Vandy", "K-State, KSU", "Dub Dub U"), need, merit, and international aid flags (field_mp_fac_*), the first-year flags (field_su_ff_fy_*: personal essay, test optional, fee waiver, self-reported scores, recommendations required), MSI filters, setting, control, and size band. A single static query, /page-data/sq/d/2437817458.json (9.4 MB), holds every member's record except the IPEDS id, so the join needs one request per member. The pages also carry marketing text, a student quote, photos, and a logo, none of which this spec takes (see Not taken).

3. The reports (national and state aggregates, months ahead of federal data)

The reports index is itself a static query, /page-data/sq/d/146375732.json (61 entries with title, date, type, and PDF link). The series that matter:

The key-findings pages are text ("1,527,328 distinct first-year applicants applied to 1,146 member institutions, an increase of 2%"; "7.06 applications per applicant"; "Mississippi was the fastest-growing state, 28%"). Every figure and every appendix table is an image inside the PDF, so state-level numbers need the vision-model route the college-reported agent already uses for flattened PDFs, with the usual checks.

4. The state dashboard (Tableau Public)

"State-level application trends and insights" embeds CommonApp2024StateReport/StateReports from the Tableau Public profile common.app.research, covering the 2023–24 season for all 50 states, DC, the armed forces, and territories. The workbook's metadata says data access is allowed, so its underlying rows can be downloaded as a crosstab; one season only, so far.

Things that exist but are not reachable from a script

• Direct Admissions participating colleges (234 for 2026–27): the list is on a Salesforce support page that renders only in a browser. A per-college flag needs a browser session once a year, or the list from Common App.
• Essay prompts per college ("writing requirements"): the same kind of page. The grid's essay and writing columns cover the yes/no.

Permission

Common App's terms of use (last updated 2025-10-24) define "the Solution" to include www.commonapp.org and the content on it, grant a "personal, revocable, limited… non-commercial" license, forbid "data mining", "page-scrape", "bots", and derivative works, and forbid linking "to any page of or content on the Solution other than the URL located at http://www.commonapp.org". Its robots file allows ordinary crawlers but disallows AI-training agents. The grid and the reports are public downloads meant for counselors and the press, but they are licensed under the same terms, and the site has paid tiers planned (Commercialization: Free, Plus, and Pro).

So the first unit of work is a letter, not code. The ask, in one page:

1. What we would use: the Requirements Grid's columns, the Explore record's IPEDS id, alternate names, and aid flags, headline figures from the reports, and the state workbook's rows.
2. How it would appear: per-college facts on that college's page and in filters, each cited "Common App, Requirements Grid, updated {date}" in the ⓘ popover with a link to commonapp.org; national lines on the trends pages cited to the report by name and date; never bulk re-export of their fields through the data API.
3. What we would not do: touch the application platform or any signed-in page; copy marketing text, quotes, photos, or logos; scrape more than one request per member per refresh.
4. What we offer: attribution on every value, a link from each college's card to the college's own application page and to commonapp.org, removal of anything on request, and a copy of the parsed grid back to them if useful.
5. Who to ask, in order: the media and research inquiry form (/contact-media, "reporter or researcher looking to speak to someone at Common App"), copying Emma Steele (esteele@commonapp.org, the press contact printed on every research report); the Data Analytics and Research team, whose vice president is Mark Freeman (co-author on the reports since 2021) and whose corresponding author on the 2025–26 End-of-Season report is Rodney Hughes; the partner form (/contact-partner, "education services provider or organization interested in partnering"). The research Call for Proposals address (researchproposals@commonapp.org) is for academic studies on their de-identified warehouse, not for this.

Until the answer: the site keeps citing report headline figures the way it cites any published report (a number, the report's name and date), builds nothing against the grid or the Explore pages, and keeps the survey's parsers out of the repo. If the answer is no, deadlines and requirements come the slow way: the college-reported agent reads each college's own admissions page (college-reported-data.md), where the same facts are published by the college itself.

Store

A new source common-app in lib/fields.ts (label "Common App Requirements Grid", vintage = the grid's cycle and "Updated" stamp, from meta.json like every source), and a block per member:

Rules that the lineage guards enforce as usual: every displayed value cites apply.* through citeField, so the popover reads "Common App Requirements Grid, 2026-27 cycle, updated September 25, 2026"; no page types a cycle or a date; the cycle label comes from apply.common_app.cycle.

Newest wins (quiet sources rule): where a CDS record and the grid both give a deadline, fee, or test policy for the same cycle, the value with the later as-of date shows (the grid's stamp against the CDS edition's publication date), the other in the popover. The IPEDS test policy stays the federal value in admissions.federal_tests, as the CDS build already arranged. The grid's codes map to the site's vocabulary:

Keep history? Yes, by cycle: every grid edition is stored (data/common-app/<cycle>-<updated>.json), and the diff between editions is exactly the cycle watch change table (test policy, early rounds, deadlines, fee and waiver). One edition a year stays after the cycle closes; the in-season refreshes overwrite each other once the next edition exists.

National and state figures

A reviewed file, data/trends/common-app.json, with one row per season per measure (applicants, applications, applications per applicant, members, growth by fee-waiver eligibility and first-generation status, share reporting a test score, applicants by state), each with the report name, date, page, and a quote, typed from the key-findings text or extracted from the appendix table images by the agent's vision route with a second-pass check. Registered as source common-app-reports with the report date as its vintage.

Display

• Profile, Admissions page: "What it takes to apply", one card: the rounds offered with their dates ("Early Decision November 1 · ED II January 1 · Regular January 1"), fee and whether a fee waiver is taken, the essay, supplement, and transcript lines, recommendations ("two teacher evaluations, a counselor recommendation, and a mid-year report"), and the test policy with tests_used when it is not "See website". The card names the cycle from lineage and shows "Rolling" as a word, not a date. After a cycle ends and before the next grid, the card says "last cycle" in the same place, from the same field.
• Explore: filters "No application fee", "Takes the Common App fee waiver", "Offers Early Decision" (and ED II), "Offers Early Action", "No essay beyond the Common App", "Test-optional this cycle" (grid N or I); a "Common App" chip is not shown (quiet sources: the fact is the deadline, not the vendor).
• Saved lists and the application tracker (saved-lists.md): deadline columns and the requirements checklist per college, which the tracker today leaves for the student to type.
• Early decision strategy (Early Decision and Early Action: Does Applying Early Help?): "offered, and when" for every member, so the ED view is not limited to colleges with a CDS on file.
• Aliases: "Vandy", "K-State", "Dub Dub U" in search, weighted like Wikidata's.
• Trends hub and state pages: a "This season on the Common App" block (applicants, applications per applicant, growth for fee-waiver-eligible and first-generation applicants, the share reporting a score) and, on each state's page, that state's applicant count and growth, each cited to the report; the series by season once two End-of-Season reports are on file.

Sync

npm run sync-common-app: download the grid (keep the PDF under data/common-app/), parse by column window, join names to IPEDS ids through the Explore records (one request per member, cached by slug, refreshed yearly), write the block, and print every unmatched name for the hand map. In season the grid is refreshed monthly (its stamp changes; the sync is a no-op when the stamp is unchanged); the Explore join yearly, in August when membership changes. The reports file is hand-reviewed per release, like data/overrides.json.

Checks

• Parse: the row count within 2% of the grid's own count of members (the sitemap's first-year members minus the international ones the site doesn't carry); every date within the cycle's window (September of the stamp year to August of the next); fees within $0–$250; test policy one of the five codes; a row whose name matches no Explore title and no alias goes to the unmatched list, never to a guess.
• Join: IPEDS ids found on Explore must exist in data/schools.json or be logged as not carried (community colleges, the 66 non-US members); a title match that lands on a different id than last cycle is a change for review.
• Lineage: npm run check:lineage finds every apply.* field; the hard-coded-year guard applies to the card.
• Fixture tests: the parser on three saved grid pages (a wrapped name, a "See website" cell, a Coordinate college); the code map; the newest-wins choice against a CDS record with an earlier and a later date.
• A test asserts the site never links to an Explore page or any commonapp.org URL other than the homepage.

Not taken

Marketing text, student quotes, photos, logos, and the "Why apply" copy on the Explore pages (the colleges' and Common App's words); admissions staff emails (stored for a future contact card only if the owner wants it, never published without the college's own page saying the same); the essay prompts page and the Direct Admissions list (browser-only, see above); anything from the application platform or a signed-in page.

Complexity

Medium: one PDF parser with fixtures, one join with a hand map, a block of fields and a source, one profile card, six filters, two product hooks, and a reviewed figures file. The permission step is the long pole and sets the start date.

Open questions

1. Ask or wait? Recommendation: ask now, with the one-page letter above; the answer sets whether anything else here is built. If the owner would rather not ask, the agent route for deadlines is the fallback and this spec is parked.
2. Explore join or grid only? The grid alone gives names; the join needs 1,174 small requests a year. If Common App would rather send a member list with IPEDS ids, take that instead.
3. Show the international fee? The grid has it for every member; the site's international visitors are a small share. Recommendation: in the popover, not the card.
4. Direct Admissions flag. Worth a yearly browser pass to mark the 234 members? Recommendation: ask Common App for the list in the same letter; a flag with no date is worse than none.