Methodology

How we make our content

The failure mode of admissions-test websites is stating unsourced things with confidence: an invented cut-off score, a vague "studies show", a set of papers described as past papers that the exam board never released. Students cannot tell the difference until they sit the test and discover they prepared against the wrong thing. We treat this as an engineering problem, so this page describes mechanisms rather than intentions: where facts come from, what a number has to satisfy before it may appear in body text, and which claims we deliberately do not make.

Who we are Read the study guides

Quick answer

Where facts come from: Exam boards, university pages and their archived snapshots, and official Freedom of Information responses. Each one carries a link.
How numbers are checked: A deterministic build-time gate: if a number is not in the fact base, the build fails.
Original data: ESAT figures for 29 Cambridge colleges, from official responses FOI-2025-956 and FOI-2025-982.
What a question carries: A worked solution and a Chinese translation per question, sat on screen against the clock in the published format.
Independence: No affiliation with, authorisation from, or partnership with UAT-UK, Pearson, OCR or any university.

This page describes our own content process. It is not a statement of any exam board's rules.

In short

Every official fact links out to its source, and each page ends with a full source list you can check line by line.
Every number in the body must appear verbatim in that page's fact base. If it does not, the build fails and the page never ships.
Our ESAT figures for 29 Cambridge colleges come from the University's official Freedom of Information responses; anything we calculated on top is labelled as ours, never passed off as an official ranking.
Where a board has never released a past paper we say so, and label our own papers as original papers rather than dressing them up as past papers.

Part 1

How a fact gets onto a page

Writers do not start from a blank page. Before a batch of content pages begins, the official material for that topic is captured into a fact base: a JSON file in which each fact carries its official URL, archived snapshots included. Writers work only from that file. A fact that is not in it cannot be written in, and if someone writes it anyway the next stage catches it.

The cost of this rule is speed. The return is that a page still holds up three months later: when an official page changes we update the fact base and rebuild, and every page that depended on it moves together, with nobody having to remember which pages mentioned that number.

01Step 1

Capture the official material

Board pages, university course pages, official specifications and Freedom of Information responses are stored as individual entries, each with its URL. Where a page changes often we also store an archived snapshot, so the citation does not disappear when the other side redesigns.

02Step 2

Writers work only from the fact base

A writer receives the fact base and a list of red lines, not an empty document. The job is to explain what those facts mean for a candidate, not to supply further facts.

03Step 3

The deterministic gate

A build script runs page by page: the number whitelist, the source-URL whitelist, bilingual completeness, the Chinese em dash ban, SEO length bands, block-structure validity and the affiliation-claim ban. Fail any one of them and that page is not generated.

04Step 4

Adversarial human check

A gate only catches deterministic problems. A fact stated backwards, a lost qualifier, one college's figure described as the University's: those are only caught by a person checking cell by cell against the fact base, which is why this step is not optional.

Number whitelist

Body numbers must exist in the fact base

The script extracts every numeric literal from the body and matches it against the fact base. One extra number fails the build.

Source whitelist

Cited URLs must exist in the fact base

Extracted the same way. A writer cannot invent an address that is not in the fact base, and a page that tries will not ship.

Affiliation-claim ban

Negation-aware

It has to pass "not affiliated with Cambridge" while blocking "Cambridge partner", so the gate reads context rather than keywords.

Bilingual completeness

Missing either side blocks the build

Both versions must exist and be non-empty, so an English page cannot quietly sit half finished.

We are explicit that the gate does not catch semantic problems. A number whitelist proves a figure exists in the fact base; it cannot prove the figure was placed in the right row. That reconciliation is done by a person, not a script.

Part 2

Original data: ESAT scores at 29 Cambridge colleges

Cambridge publishes no ESAT score threshold, so a student asking "where does my score sit at this college" finds only hearsay. We obtained the college-level data through the official Freedom of Information route and built one page per college, 29 in all.

The caveats sit on every page: these are averages, not entry requirements; the response gives no applicant counts per college, so sample sizes are unknown and a handful of candidates can move a small college's average; and the mature colleges, which admit applicants aged 21 and over, are pooled separately because they are not comparable with the undergraduate colleges.

	Who produced it	How we label it
Per-college module averages, lowest and highest scores, for applicants and offer holders	The University's official FOI response	Transcribed line by line, with the response reference quoted
Pool rank, pool median, gap to the median, applicant-to-offer gap	Arithmetic we performed on the figures in that response	The page states we calculated it and that Cambridge publishes no such ranking
Cut-offs, pass marks, target scores	They do not exist	The page says so plainly, with no suggestive phrasing

Separating what we derived from what was supplied is the step most easily skipped on pages like these, and the one that matters most. A reader is entitled to know which figure came from Cambridge and which we calculated.

These figures come from responses the University of Cambridge made under freedom-of-information law. They do not imply any relationship between the University and this site, nor any endorsement of the readings we offer.

Part 3

What a question carries

The value of practice is not in finishing a paper, it is in whether you can work out what went wrong afterwards. So most of our cost sits not in the questions but in what comes after them.

A worked solution, not just the answer letter

Every question has a full solution. It has to explain why a step is the one you would reach for, rather than reciting the official answer backwards. The same wrong-answer card and explanation components are shared by the exam, practice and drilling modes, so a question gives the same depth of feedback wherever you meet it.

The Chinese translation is a subtitle, not a replacement

The stem is always in English, because the exam is in English. The translation is shown alongside it: the original never goes away, and figures always come from the original. The point is to get a Chinese-speaking candidate past the reading barrier, not to let them get used to sitting the paper in Chinese.

Sat on screen, against the clock, in the published format

Module count, question count and time limits follow the published format, and the interface is built to look like the on-screen test, with the timer, flagging and navigation between questions. Paper practice does not rehearse on-screen navigation or flagging, and both genuinely cost time on the day.

After a mistake you get a diagnosis, not another paper

Accuracy and timing are tracked by topic so the recurring problem areas surface, instead of simply pushing more questions at you. Volume is not the goal; spending the time where it is actually weak is.

Part 4

Four things we deliberately do not do

Most of a revision platform's credibility shows up in what it refuses to say. The four below are hard rules, and they do not bend because a phrase happens to have search volume.

We do not call our own papers past papers

For some tests the board has never released a past paper at all. Where that is the case we say so on the page, set out what official material does exist, and label the papers we write to the published specification as original papers. Plenty of people search for past papers, but calling an original paper a past paper leaves a student preparing against the wrong difficulty profile.

We claim no relationship with any official body

We are not affiliated with, authorised by, or partnered with UAT-UK, Pearson, OCR, the University of Cambridge, the University of Oxford, Imperial College London or any official test owner. That statement appears at the foot of every content page on the site in both languages, and a build-time gate watches for any page whose wording drifts loose.

We publish no unsourced cut-off scores

"What score is safe" is the most asked question and the easiest answer to invent. Where a board has published no threshold we say it has published none, then give the distribution data that does exist along with its limits, so the reader can judge, rather than inventing a number on their behalf.

We do not blur the billing terms

Access is bought per subject as a one-off payment: once unlocked, access itself does not expire, and updates are included for one year from the date of purchase, so anything new we publish in that subject during the year is included. After the update window ends, everything already unlocked keeps working, and the window can be renewed annually if you want to keep receiving new material. We avoid the phrase "lifetime access", because it implies new content is included for ever, which it is not.

Billing terms

What you are actually buying

One-off, per subject: £58.80 for one subject, £108.80 for any two. Not a subscription.
Access does not expire: Everything unlocked stays unlocked, including after the update window ends.
Updates for one year: New material in that subject is included for one year from purchase, renewable at £18 per subject per year.

We avoid the phrase "lifetime access": access itself does not expire, but new material is included for the update window, and describing that as lifetime would misstate what you get.

Go deeper

Sources

Official sources

Spotted something wrong on this page? Write to contact@frontierace.ai with the page address and the sentence in question.

FAQ

Frequently asked questions

Is your content written by AI?

We use tools in the process, but tools do not decide the facts. Facts are captured from official material into a fact base first, writing happens on top of that, a deterministic build-time gate checks numbers and sources page by page, and a person then reconciles the meaning against the fact base cell by cell. A more useful test of any page is whether it carries outbound sources you can check yourself, rather than who typed it.

What is your relationship with the exam boards?

There is none. Frontier Academy is an independent admissions-test tutor and FrontierVUE is the practice platform we built. We have no affiliation with, authorisation from or partnership with UAT-UK, Pearson, OCR, the University of Cambridge, the University of Oxford, Imperial College London or any official test owner. When we cite official material we link to it, so you can read the original yourself.

How did you obtain the Cambridge college data?

Through the official freedom-of-information route. The responses are FOI-2025-956, which gives the college-by-college module data, and FOI-2025-982, which gives a University-level comparison; both were issued by the University of Cambridge. The pages distinguish figures taken from the responses from figures we calculated on top of them, and the latter are always marked as ours, with a note that Cambridge publishes no rankings.

What if I find an error on a page?

Write to contact@frontierace.ai with the page address and the sentence you think is wrong, together with an official source if you have one. For factual issues we correct the fact base and rebuild, so every page that cited the same fact updates together and you do not end up with one page fixed and five others still carrying the old wording.

FrontierVUE is an independent practice platform. It is not affiliated with or endorsed by UAT-UK, Pearson, OCR, the University of Cambridge, Imperial College London, or any official admissions-test owner.

FrontierVUE 是独立的备考练习平台，与 UAT-UK、Pearson、OCR、剑桥大学、帝国理工学院或任何官方入学考试主办方均无隶属或背书关系。

Judge it for yourself

Every subject has a free paper you can sit end to end, with the same worked solutions and translations as the paid ones.

Start free See pricing