Document 02  ·  Advisory

Competitorguidelines.

Nothing here is binding — the rules are. This is what tends to win, and what the two masterclasses go into properly.

01

Finding an idea.

Round one is marked on creativity, feasibility and profitability. Most entries are strong on one and thin on the other two — the ones that win are decent at all three.

  • It has to be yours. You cannot copy an existing company, and AI cannot generate the core idea. Both are disqualifications, not deductions.
  • Start from a problem you have actually seen. Something annoying at school, at home, on the bus. Specific beats grand every time.
  • Name the customer. “Everyone” is not a customer. Who exactly, how many of them, and how would you reach them?
  • Check nobody already does it. If they do, that is fine — but you need to say what you would do differently, and why.
  • Sanity‑check the money early. If you cannot see who pays, the profitability marks are gone before you start writing.

02

Researching it.

Judges can tell within a paragraph whether you have spoken to anyone. The tell is specificity: real interviews produce odd, concrete details that nobody invents.

Ask about the past, not the future

  • “Walk me through the last time you had to do this” beats “Would you use an app that…”. People are honest about last Tuesday and unreliable about hypotheticals.
  • Ask what they tried already, what it cost them, and why they stopped. A problem nobody has attempted to solve is usually not a problem.
  • Ask how often it happens and what it costs when it does. Frequency times cost is where a business hides.
  • Shut up. The useful material comes after the pause you were about to fill.

Practicalities

  • Cite your data. The rules require it, and an uncited number is worth less than no number.
  • Be careful with AI research. It is allowed, but it is often wrong or biased, and incorrect information will be penalised. Check anything it gives you against a real source.
  • Ask people outside your friend group. Classmates and family are the easiest to reach and the least useful to hear from.
  • Eight to twelve conversations is enough to find a pattern. Two is an anecdote.

A hard test

After each interview, write one sentence: what did I learn that would have surprised me yesterday? If you cannot answer for three interviews running, you are pitching, not researching.

03

Writing the proposal.

A Word document, 1,000 words plus up to 20%, emailed once as an attachment. That is not much room, so every paragraph has to earn its place.

  • The problem

    One paragraph. A specific person in a specific situation, and what goes wrong for them.

  • The idea

    What you are proposing, in plain language. If a reader cannot repeat it back after one read, rewrite it.

  • The customer

    Who exactly, roughly how many, and how you would reach them.

  • The money

    Price, cost to serve, and the arithmetic between them. This is the profitability mark.

  • Why it works

    Feasibility: what would have to be true, and why it could actually be built and run.

  • Competition

    Who solves this today — including “people put up with it”.

  • Sources

    Cite your data. Include your AI declaration here: which tools, what for, which parts.

Common ways proposals lose marks

  • A market size quoted from a headline with no path from it to your first hundred customers.
  • Numbers with no source — the rules require citations.
  • An idea that turns out to be an existing company with a new name.
  • Jargon standing in for a mechanism: “an AI‑powered platform” without saying what happens when someone taps the button.
  • Going over 1,200 words.

04

The pitch.

Five minutes to present, five minutes of questions, in the PAC. The pitch is marked out of 50 and weighted above the idea itself — round one already established the idea is good; this round is about selling it.

Where the marks are

  • Delivery and stage presence — 15. The single biggest category. Projection, eye contact, energy, command of the room, body language. This is practice, not talent.
  • Clarity and structure — 10. A logical flow and one clear message.
  • Quality of the business idea — 10. Convince them it has investment potential.
  • Q&A response — 10. Appropriate, convincing, relevant.
  • Visual aids and timing — 5. Helpful diagrams, professional deck, and no rushing or dragging.

Practicalities

  • Every team member must present on the stage. Split it up in advance.
  • Slides are optional and go in as a PDF two days before. They are locked after that — no edits on the night.
  • You can bring a physical example of your product, within the set size, shown to the organisers beforehand.
  • Take the 10‑minute practice slot in the PAC on the day. Rehearse to the clock: five minutes is shorter than you think.

Surviving the Q&A

  • Answer the question that was asked, first, in one sentence. Then elaborate.
  • “We don’t know yet — here’s how we’d find out” scores well. A confident invention scores badly and tends to unravel.
  • Know every number on your slides and where it came from. Expect to be asked about exactly the one you are least sure of.
  • Do not argue with a judge. Take the point, say what you would do about it.
  • Rehearse the questions you are dreading. They will be asked.

05

Masterclasses.

Two of them, both 45 minutes in the Lecture Theatre during school: thirty minutes presenting, then fifteen going round the room answering questions. Popcorn provided. Sign‑up only, and dates are still being confirmed.

Masterclass 1

Building the idea

Open to everyone who has entered. How to find an idea worth entering and turn it into a proposal that scores.

Masterclass 2

Delivering the pitch

For the ten teams through to the final. How to build and deliver five minutes that land.

06

Before you send it.

Run this the day before, not the hour before. You can only email your proposal once.

  • A Word document, text‑based.
  • 1,200 words or fewer (1,000 plus the 20% allowance). Check the word count.
  • The idea is yours — not an existing company, not another team’s.
  • Every number has a citation.
  • AI declared: which tools, what for, and which parts of the work they assisted with.
  • At least 75% written by you, without AI.
  • You could explain any part of it out loud if a judge asked.
  • Emailed as an attachment, once, to jiescalantephilips@perse.co.uk, before the deadline.

Still have a question

Ask early. It never counts against you.

Questions asked before a deadline get an answer and often a rule clarification for everyone. Questions asked after one get sympathy.