About SummerTrack
SummerTrack started from a complaint: the good summer programs are findable, but only one at a time, by a counselor with thirty other students waiting. Everything else on offer is either a paid directory that ranks whoever pays, or a scrape nobody maintains. So this is a catalog kept by hand, wrapped in the workspace a counselor actually needs — a roster, per-student shortlists with the reasoning attached, and a way to send a family a plan.
Who maintains it
A small team. We build the product and we keep the catalog ourselves — it is not crowdsourced, it is not licensed wholesale from a directory, and there is no offshore data shop filling rows against a quota. When a record is wrong, the people who can fix it are the same people who wrote the software.
That is a real constraint, not modesty: it is why the catalog grows in deliberate passes rather than continuously, and why we would rather tell you a field is unknown than have somebody invent it to make a page look finished.
You can reach us at hello@summertrack.app. A counselor signed in can also use the feedback tab on the right-hand edge of any page, which comes straight to us.
How a program gets into the catalog
There are currently around 2,900 programs from roughly 1,500 distinct hosts. Each one goes through the same three steps.
- Find a candidate. Public directories and web search tell us a program exists. That is all we take from them — a name and a link.
- Read the program’s own page. The record is built from the program’s official website, not from the directory listing that pointed us at it. A directory entry is a floor, never an import: it can tell us a program exists, and it is not evidence of what that program costs this year.
- Reject the ones we could not actually read. A dead link, a parked domain, or a page that turns out to be a bot wall produces an empty record — and an empty record that inherited a directory’s facts would look complete and be a copy. So every candidate is scored on how much of it came from the program’s own page, and one that falls below the bar is thrown out rather than shipped. On the most recent large pass that rejected about one candidate in eleven, and every single rejection was that rule.
Records are entered against a fixed schema — the same twenty-seven fields for every program, with closed vocabularies for the ones you filter on. That is unglamorous and it is the whole reason a facet search works: “Massachusetts” is one bucket rather than three spellings of one.
How you know whether a record is current
Each program page shows the source it was built from and the date that source was last read. That date is stamped by the process that did the reading, and it is left off entirely when a read did not actually succeed — so a page never displays a check that did not happen. About one record in four additionally carries a caution marker, meaning its last re-read came back thin and its values are older than the date suggests.
There is a second and stricter status: human-verified, meaning a person sat down with the program’s own site and confirmed the record by hand. We have not made that pass yet. So every record in the catalog currently reads “Unverified”, and each one will keep reading that way until a person has actually checked it. We would rather the badge mean something than have it default to reassuring.
Check the program’s own site before a student relies on anything. Every record links straight to it. Deadlines move, costs change and sessions get cancelled, and no program is obliged to tell us — see the catalog terms.
How the programs are ordered
Browse opens in a recommended order rather than by whichever deadline happens to fall next. That order is, in this sequence: our own signal rating, whether the program selects its students, cost with free and low-cost first, and only then the deadline. Programs whose recorded deadline falls after the session has already started sit below the rest, because that is almost always a bad scrape rather than a real date. You can switch to deadline, rating or A–Z at any time, and none of those orders is capped.
No single organisation gets more than three cards in that view. A few operators list dozens of near-identical trips, and without this the first page was mostly one company. The rest are one click away, behind a link on the third card.
The signal rating is the one editorial judgment in the catalog, and it is deliberately a narrow one: it estimates what a program says about a student who was admitted to it, not how good or enjoyable it is. Ratings are assigned by hand against a written rubric, each with a one-line reason recorded alongside it, and no program can buy, request or appeal one. Today about one program in thirty carries a rating at all — everything else is simply unrated, which means nobody has reviewed it yet and never means we think poorly of it.
Why so many fields are blank
This is the editorial policy, and it is the one thing on this page we would most like a school to hold us to: when a source is silent, the field stays empty. No inferred cost band, no plausible-sounding admissions note, no guessed deadline, no “typically around”.
It costs us, visibly. Across the catalog today, about half of programs carry nothing on financial aid, close to four in ten list no application requirements, and more than a quarter have no deadline at all — because those programs do not publish them. A detail page for one of those is thinner than a competitor’s would be, and we have built the page to accommodate that rather than to disguise it.
The reason is simple. A counselor cannot tell a confident guess from a checked fact by looking at it, and a wrong deadline in a tool a student trusts is worse than no deadline at all — a blank prompts them to go and look. So the extraction is built to emit nothing when a page says nothing, and the empty result is allowed through to you.
One honest wrinkle, because it is the exception: when we re-read a program and its page is silent about a field we already had a value for, we keep the old value rather than clearing it. Losing a cost band the catalog already had helps nobody. The consequence is that a record can carry a recent verification date while one particular field is older than that date. Every one of those carried fields is recorded in our own review log, and it is why the verification date sits on the record rather than on each field.
What we do not do
- No paid placement. No program pays to be included, to rank higher, or to be recommended. We take no affiliate commission and no referral fee. Listing is not an endorsement, and we are not affiliated with any program in the catalog. What the default order does reflect is described just above.
- No selling data. No data brokers, no ad networks, no third-party analytics anywhere in the product. Student names and contact details are never used for marketing, and student records are never used to train AI models. The privacy policy is the long version.
- No AI writing the catalog from memory. A model is used to read a program’s page and pull structured fields out of it — with a text in front of it every time. It is never asked what it knows about a program, because that is precisely how a confident invention gets into a database.
Tell us when we are wrong
A record that is out of date is a bug, and we would rather hear about it from you than find it in six months. Email hello@summertrack.app with the program and what is wrong, or use the feedback tab inside the app. Program staff who want their own listing corrected should write to the same address — we will check it against your site and fix it.