Reading a match

A match answers two questions: how well does this professor's work line up with yours, and what specifically would you say to them. It appears on the professor's Match tab once their research has been verified.

4.1 The headline#

At the top you get:

• A score out of 100 — overall fit against your profile.
• A verdict — one of four:

• A confidence figure, also out of 100. This is how sure the scoring is, and it's separate from the score itself. A 78 with confidence 40 means "possibly a strong fit, but the record is thin" — go read the sources before you act on it.
• Which version of your profile it was scored against. If you've substantially rewritten your research interests since, an old match is measuring against an old you.
• Concerns, when there are any. Read them.

4.2 The dimension breakdown#

Under the headline, fit is broken into the things that actually make a professor a good or bad advisor for you:

The shape of this matters more than the total. An 80 built on topic and methodology fit is a strong lead. An 80 built on "other strategic fit" while topic fit sits at 30 means you like the university, not the advisor.

4.3 The angles — the part you'll actually use#

Below the scores sit at least three connection angles. Each one is a different, specific way of opening a conversation, and each is built the same way:

• Professor evidence — the thing of theirs it rests on, from the verified record.
• Applicant evidence — the thing of yours it rests on, from your profile and CV.
• Connection — why those two belong in the same sentence.
• Potential direction — where the work could go if you joined them. This is often the most quotable line in your email.
• Risks or gaps — what's weak about the angle. Read this before you commit.
• Strength and confidence, each out of 100, plus how many pieces of evidence support it.

Angles are deliberately different from one another rather than three phrasings of the same observation. One might rest on a shared method, another on a paper of theirs your project extends, another on a direction they've hinted at and you're already working in.

Picking one#

Click Use this angle on the one you want. It gets marked, and both email drafts are written from it. Change your mind and pick another at any time — then regenerate the drafts.

How to choose:

• Prefer the specific over the flattering. "Your 2025 paper's segmentation approach" beats "your impressive body of work".
• Prefer what you can defend for twenty minutes. If they reply and ask a question, you have to be able to hold up your end.
• Read the risks line. An angle resting on a five-year-old paper is a weak opener, however good the connection.

4.4 "Why here" — the case, in three sizes#

An angle tells you what to open an email with. It does not answer the question every application eventually asks in writing: why this place? The Why panel on a professor's page builds that case at three levels, because applications ask at three levels and the honest answer is different at each:

• Why this university — reputation, funding, the shape of the programme.
• Why this department — where it sits in the field, who else is there, what it is known for.
• Why this professor — the specific work, and what you would contribute to it.

Each is built only from what has been researched and published for that record, so a thin record produces a thin case rather than a confident invention. If the panel says there is no case built yet, that is the roster telling you it does not know enough about the place — not a failure of the generator.

Two things travel with the case and are the reason to read it rather than paste it:

• What you can say — claims the record actually supports.
• What not to overclaim — the ones it does not, stated plainly, so you don't reach for them under deadline pressure.

Copy all three puts the set on your clipboard for a statement of purpose or an application form. Treat it as sourced raw material, not finished prose — it is written from your records, but the sentences that land are the ones you rewrite in your own words.

4.5 Contribute, Questions and Prepare — the rest of the packet#

A match tells you whether to write. These three tabs tell you what to say, and what to have ready for everything after the first email. A research run fills them in at the same time as the match, so firing research once produces the whole packet rather than the first third of it.

Every point comes in two versions#

This is the part that matters, and the reason these tabs are safe to use.

Each point is written twice. The first version is written to you and is allowed to be blunt — "their evaluation is English-only". Underneath it, in a shaded box, is the same point as it could actually be said to them: "would it be worth checking whether the gain holds on a non-English collection?" Only the second version has a Copy button.

Some points have no second version. That is deliberate: the point is real, it is worth knowing, and there is no way to say it to a professor without implying they missed something. Those stay in your head. The tab says so rather than softening the wording until it looks sayable.

Sections that are empty on purpose#

An empty section says which of two things happened. Not written yet means no run has looked. Deliberately empty, with a reason, means a run looked and found nothing small and genuinely useful to say — and that is the correct output. Seven full sections on a thinly-researched professor is the signature of writing to fill space, which is exactly the thing that makes an applicant look careless.

What gets refused before it ever reaches you#

The material is checked as it is written, and phrasing aimed at the professor is refused outright when it says they got something wrong, instructs them, flatters them, or claims a result. "Your evaluation fails to cover other languages" does not get stored; "the evaluation is on English data — would a second language be worth trying?" does. Your own blunt version is kept either way, marked with what it was flagged for, because bluntness in your notes is useful and bluntness in an email is not.

Reading it before you use it#

New material shows a Not read yet badge, and Review → Collaboration material lists everything waiting so a run's output is not missed. Keep and Dismiss are acknowledgements, not a gate — unlike research findings, this material is usable the moment it is written. Dismissing keeps it rather than deleting it, so a later run can see the section was rejected once.

Written against an older profile means your profile has changed since the material was written. It is still there and still readable; it just has not been rewritten against who you are now.

4.6 Working through your matches#

The practical loop, once you have a few batches behind you:

1. Professors → High match. This is your shortlist.
2. Open each one, read the dimensions, and check the concerns.
3. Pick an angle. If nothing convinces you, that's a legitimate answer — close the tab and move on. A weak email costs you the professor permanently.
4. Read Contribute and Questions before drafting — the safe phrasings there are what an email should be built from.
5. Move to the Email tab and draft.

The dashboard keeps this loop visible: Contact next lists your strongest matches you haven't written to, and Ready to contact counts them.

What's next#

You have a professor worth writing to and an angle to write from: outreach →