How Shortlist sources its data
Shortlist is a research publication. Every data point on the site — the license, the state’s inspection and complaint record, tuition, schedules, what a program says it teaches — traces back to a public source or a provider’s own statement. This page documents what gets included, where each field comes from, how much of it exists today, how often it’s refreshed, and the standards behind every value.
The same sources, the same standards, applied equally to every provider — and no provider can pay to change what its listing says or where it appears.
Shortlist lists licensed childcare providers that operate in a covered metro area. A provider appears once it meets at least one of these conditions:
Every provider that clears one of these bars is shown, even when its profile is still sparse. A missing field is information Shortlist owes the parent, not a reason to hide a program that genuinely operates — so a listing with only a name and an address still appears, and fills in as sources are found.
Shortlist covers licensed, facility-based centers and home daycare programs in the metros it has researched — currently Seattle, Chicago, Denver, Austin, the Bay Area, Charlotte, and Kansas City: 12,567 listings as of 17 September 2026. In-home nannies, individual caregivers, and unlicensed care fall outside that scope. When a provider closes — confirmed by DNS, licensing status, or its own website — the listing is withdrawn.
Different fields come from different places. Shortlist follows a tiered source hierarchy — structured authoritative sources first, per-provider scraping only when nothing better exists.
Authoritative for licensing status, capacity, license number, inspection history, and state-administered quality ratings. One API call returns structured records for an entire state.
Kansas City spans two states, and Shortlist lists only the Missouri side — Jackson, Clay, Platte and Cass counties. There are no listings for Overland Park, Olathe, Lenexa, Shawnee, Kansas City, Kansas, or anywhere else in Johnson and Wyandotte counties. That is a hole in the site, and this is the reason for it.
Until the middle of 2026, Kansas child care licensing records were held by KDHE and were public records like every other state’s on this page. On 1 July 2026 licensing moved to the new Kansas Office of Early Childhood, and the same law added K.S.A. 75-7722(f), which provides that these records “shall be confidential and shall not be subject to the open records act.” Shortlist requested the Kansas roster and inspection history under the Kansas Open Records Act in July 2026 and again in August. The Office of Early Childhood declined; its final answer cited that provision. There is no bulk file left to ask for.
The information itself is still public. The same 2025 act requires the state to run an online search of licensed programs that includes citation history and substantiated findings (K.S.A. 75-7728), and Kansas does — one facility at a time, behind a CAPTCHA, for someone who already knows which program they are looking up. What is not available is any way to obtain the records as a set. Until Shortlist has a way to read that search at the scale a directory needs, Kansas programs stay off the site rather than appearing as a partial list. A parent comparing the Kansas side of the metro should know they are seeing nothing here, not that there is nothing there.
Not on the site today: no listing carries an accreditation value. The rule below is the one Shortlist will hold to when it does, and it is written here so a provider can see the standard in advance — a directory check is the only thing that will count, and a claim on a provider’s own website will not. Until that check has been run, Shortlist publishes nothing about accreditation either way. An absent accreditation on a listing means it was never checked, never that a program failed one.
Used to verify provider claims, never as a discovery source. If a provider claims accreditation but the directory doesn’t list them, Shortlist flags the discrepancy for review — it doesn’t silently clear the claim. Directory lag, name mismatches, and incomplete public scraping are all real reasons a legitimate accreditation may not appear.
Published parent handbooks are the single richest per-provider source — ratios, meal policies, sick policies, diapering, outdoor and screen policies often appear in one document. When a handbook is found on a provider’s website, it’s prioritized over individual subpage scraping.
Programs, tuition, philosophy, and enrollment status are read from provider websites — homepage plus targeted subpages (/programs, /tuition, /enrollment, /about, /careers). A field is filled only from an explicit statement: a printed tuition figure, a stated ratio, a named philosophy. Marketing language stays a lead until a concrete value backs it.
Not on the site today: no listing carries a teacher-pay figure. The rule below is what a figure will have to clear before one appears.
Teacher pay is sourced exclusively from real job postings with explicit dollar amounts — Indeed listings, provider careers pages, and IRS Form 990 filings for nonprofit operators. Salary estimation models, metro-area averages, and review-site comp ratings are not used.
When automated sources are exhausted and gaps remain, Shortlist emails providers directly to ask for two or three specific missing fields. There’s a three-email lifetime cap per provider, weekday-only sends, and no follow-up after a non-response — outreach is the last lever, not the first.
Providers can claim their listing through the Shortlist provider dashboard and submit corrections or additions directly. Self-submitted data goes to a review queue, not straight to the public record — it’s reconciled against existing sources before being applied.
Where each public field comes from, and how much of it exists. The third column is the count of listings carrying a value, live from the database as of 17 September 2026, out of 12,567 — a field can be well defined and still be empty, and a methodology that describes the rule without the count tells you only half of what is true.
| Field | Primary source | On the site |
|---|---|---|
| License status | State licensing database | 12,391 |
| Licensed capacity | State licensing database | 12,509 |
| Program type | State licensing database | 12,484 |
| Inspection & complaint records | State licensing database | 10,853 |
| License number | State licensing database, where the state publishes one | 7,164 |
| Location (latitude, longitude) | Geocoded from the licensed address | 5,506 |
| State quality rating | State QRIS (Early Achievers, Colorado Shines, NC Star, Texas Rising Star) | 4,085 |
| Non-profit or for-profit | The licensee on the state license, then the IRS Exempt Organization file — never the name, the domain, or the kind of building | 3,183 |
| What a program teaches | The program’s own words on its own site, quoted — never a label Shortlist assigns | 1,039 |
| Age range | Provider programs page | 997 |
| Named teaching staff | The program’s own staff or team page — a published name is required, never an aggregate claim | 830 |
| Schedule (hours per day, days per week) | Provider programs or enrollment page; both numeric values required | 668 |
| Meals | Parent handbook → FAQ | 542 |
| Year established | State licensing date → ProPublica 990 ruling date → provider website → Wayback Machine first crawl | 467 |
| Tuition | Provider tuition page; state-published rates if available | 451 |
| Staff tenure | Named teacher bios with specific years at the program — never inferred from “experienced staff” | 209 |
| Ratio (children per teacher) | Parent handbook or staff page, explicit only | 197 |
| Enrollment status | Provider enrollment page, with a date stamp on every update | 1 |
| Teacher pay | Indeed listings, provider careers pages, or IRS Form 990 filings — only real postings with explicit dollar amounts | none yet |
| Accreditation | Accrediting body directory; a provider website is a claim, never a verification | none yet |
Every populated field traces to one of the sources above. Where a field can’t be sourced, it stays empty and the record shows the gap. If a provider you know well looks sparse, it usually means Shortlist is still gathering information, not that the program is weak — a provider claim through the dashboard is the fastest way to fill it.
The thin rows are worth being blunt about, because a rule with no numbers behind it can read as a capability. Shortlist has read 2,308 provider websites looking for one; 197 of those programs published a ratio it could quote and 209 published how long a named teacher had been there. That is the yield, not a target. The thinnest rows are emptier still: no program’s careers page has produced a wage figure that clears the bar, accreditation has not been checked against a directory at all, and a single listing carries an enrollment status. Each stays blank until a real source fills it — the alternative is a state licensing minimum or a metro average standing in for a fact about one program, and a number a parent cannot act on is worse than a blank they can raise on a tour.
The counts above are read from the database, as of 17 September 2026. They are not typed and not cached — a rule the data has outgrown cannot sit here looking finished. The largest single reason a field is thin: only 2,514 of the 12,567 listings have a website on file at all, and 2,263 of those have been read — for four in five programs there is no page to read in the first place.
Where a state publishes the inspection and complaint record for its licensed providers, Shortlist shows it in full — this is the highest-stakes information on the site. Every complaint case and every routine monitoring finding on file appears on the provider’s listing, each linked to its source document so a parent can read the original. This coverage is live today in all seven metros — Seattle (Washington DCYF), Chicago (Illinois DCFS), Austin (Texas HHSC), the Bay Area (California CDSS), Denver (Colorado Department of Early Childhood), Charlotte (North Carolina DCDEE) and Kansas City (Missouri DESE Office of Childhood) — each labeled with the state agency that keeps the record. That is 242,914 records across 10,853 listings.
A listing showing no findings is not the same as a program with a clean record, and Shortlist keeps the answers apart rather than collapsing them. A listing says whether the state holds a record and reported no violations, whether the state does not list that program at all, or whether Shortlist has not read it yet. Only the first of those is a clean record, and it is the only one written as one.
Complaints come first and are always shown verbatim — a complaint is the clearest signal a parent can act on, so it is shown as the state wrote it and never held back. Each complaint case carries the dates it was received and, once closed, resolved, along with the specific licensing rule cited where the record names one; each monitoring finding carries the date of the visit. Shortlist reports what the record says and links the source — it does not layer its own verdict on top of the state’s.
No state publishes this as a clean data feed, so each one is read a different way. Washington posts per-provider PDF documents, so Shortlist scans each licensed Seattle provider’s DCYF portal record on a recurring schedule and re-reads the underlying documents whenever that record changes; Texas and California expose a per-facility lookup; North Carolina lists visits on the facility page, including the visits where the state cited nothing; Illinois publishes a compliance report PDF per provider; Missouri opens each visit from its facility portal one record at a time; and Colorado publishes no machine-readable record at all, so Denver’s findings come from an annual records request to the state. A newly filed complaint is the highest-priority output of every one of those reads.
The same treatment extends to a new metro as its state records become linkable. If a record looks wrong, email [email protected]; findings are corrected against the state’s own source, never quietly removed on request.
Whether someone takes profit out of a program is a fact about who holds its license, and every state that publishes a licensee publishes it differently. Shortlist reads that field, applies the same four rules to it in the order below, and refuses rather than guesses. 3,183 of 12,567 listings carry an answer; 4,185 have had their licensee read, so the difference between those two numbers is the programs where the record was read and still did not settle it.
Two things the page will not do. It does not read Inc. as for-profit — most registered charities are incorporated, so incorporation says the license is held by an organization and says nothing about profit. And it does not read a church, a foundation or a university as either: a church is tax-exempt automatically and need never appear in the IRS file, so its absence there is not evidence. Both of those stay blank until a register settles them.
A program’s own answer outranks all four rules. A director who states it through the provider dashboard sets it, and the weekly read leaves it alone from then on.
Where a state publishes no licensee, no rule can run. Texas and Illinois publish an owner for no provider at all, which is 6,654 listings that will stay blank until that changes — not a backlog, a ceiling. Two more states publish an owner but no field naming a person, so an individually-owned home there is read and left unresolved rather than inferred from the fact that it is a home: where a regulator does publish an owner of record, roughly a quarter of family child care homes turn out to be owned by an organization.
Provider tuition is published in wildly different formats — weekly, monthly, annual, per-semester, hourly, in installments. To make providers comparable, every price is normalized to a monthly figure:
If days-per-week varies, the lowest option becomes price low and the highest becomes price high. If the timeframe of a published price is genuinely ambiguous after applying these rules, the price is flagged for human review and not auto-committed.
Co-op caveat. Co-op tuition reflects the dollar fee only. Required parent volunteer hours are a real cost — lost work time, schedule constraints — and Shortlist discloses the volunteer requirement separately rather than burying it in the headline price.
Different fields go stale at different rates, and every value on a listing carries the month it was last checked, so a parent can see the age of the specific number they are reading rather than a site-wide “last updated” date. Refreshing is a manual run — roughly weekly, not a cron job and not a live feed — and it does not reach every metro equally. What actually runs:
| What | How it refreshes |
|---|---|
| License status and closures | Every run, for Seattle, Austin, the Bay Area, Denver, Charlotte and Kansas City — the state file is re-read and a provider the state marks inactive is withdrawn. Chicago’s licensing re-ingest is a separate tool and is not yet part of the run. |
| Inspection and complaint records | Every run, for Seattle, Austin, the Bay Area and Charlotte — each provider’s state record is re-scanned and the underlying documents re-read where it changed. Chicago is refreshed by a separate tool; Denver is an annual records request to the state, because Colorado publishes no machine-readable record. |
| Provider websites | A stored website is re-verified after 90 days, then re-captured and re-read. A re-read that comes back empty does not wipe what is already sourced — it is flagged for a person instead. |
| Tuition, schedule, policies and the rest | Refreshed whenever a provider’s site is re-read, and whenever a provider submits an update. There is no automatic expiry on these — the checked date on the value is the honest signal, and an old date means old. |
The data on the site is a snapshot, not a live feed. State agencies publish on their own schedule, providers update their own pages on their own schedule, and Shortlist’s pulls happen on the cadence above. Before making any decision based on Shortlist data, parents should verify pricing, availability, and program details directly with the provider.
A methodology is as much about how findings are framed as how they’re gathered. Shortlist holds to a few principles:
Shortlist presents each provider’s structural facts — ratios, group size, credentials, accreditation — as evidence for a parent to weigh, and leaves the judgment of fit to the family who tours. The research supports this: the von Suchodoletz et al. 2023 meta-analysis of 185 studies finds that the structural factors a public site can measure are only weak predictors of child outcomes, so Shortlist treats them as evidence rather than a single quality score. (A/B/C/D letter grades were part of the site through the first half of 2026 and were removed entirely, code included, on 5 June 2026 for this reason.)
Every provider is included on the same terms — operating in a covered area and meeting licensing requirements — and a city index is a plain list of every one of them, not a ranking. The same sources and the same standards apply to each, and placement is never for sale.
Every value comes from a real source: a state record, a provider’s own page, a job posting, a directory. When a field can’t be sourced, Shortlist leaves it blank and the record shows the gap. A blank means the research is still open — it is always a real value or nothing, never a modeled or predicted number standing in for one.
Shortlist uses first-party analytics to count page views and understand which features parents actually use. Visitor data stays first-party. See the privacy policy for details.
If something on the site looks wrong, email [email protected]. Corrections backed by a public source (a license record, a provider’s own website, an accreditation directory) move quickly. Providers can also claim their listing and submit edits through the provider dashboard; submissions go to a review queue and are reconciled against existing sources before being applied.
Last updated: August 20, 2026