Research that wins — from Google to ground truth
Information is free; insight is scarce. Build the stack that produces insight.
Every team in the competition has the same Google. The decks that survive screening are the ones whose claims trace to sources the judges trust — and whose one or two original data points no other team has.
TL;DR · Key Takeaways
Key takeaways
- Work the stack: sponsor filings → industry reports → government data → primary research; cite a tier for every claim.
- Tier-4 primary research (survey, store visits, alumni calls) is the cheapest differentiation available — ship the survey by day 3.
- Nothing enters the deck until it completes fact → so-what → now-what; statistics are not insight.
The research stack, bottom to top
The 5-day research sprint
Day 1 — frame before you search
From your issue tree, write the 8–12 questions the deck must answer. Research without questions is doom-scrolling with a spreadsheet open.
Days 2–3 — secondary sweep (tiers 1–3)
Divide tiers among the team. Everything goes into one shared sheet: claim, number, source, link, year. A claim without a source row does not exist.
Day 3 — launch primary in parallel
Surveys need 3–4 days to collect responses, so the Google Form ships on day 3, not day 8. Recruit respondents through batch groups, family WhatsApp, and the target segment — and report n honestly.
Days 4–5 — field and expert pass
Store visits with photos (photos go in the deck — instant credibility), 2–3 alumni calls (15 minutes each; people love being asked), mystery-shop the sponsor's own product and its rivals.
Day 5 — synthesis gate
Each researcher presents findings as fact → so-what → now-what triplets. Anything that can't complete the triplet is cut, however interesting.
Sizing the market without faking it
Almost every PS needs a market size or an impact estimate, and this is exactly the guesstimate skill the rest of MECE drills. Build it bottom-up (units × frequency × price), sanity-check it top-down against a published figure, and show both in the deck — the cross-check is worth more marks than the number. The guesstimates section is your training ground; competition decks are where it cashes out.
Citation hygiene
Footnote every number on the slide where it appears (source, year). Two failure modes get teams quietly binned: numbers with no source, and numbers from a rival's sponsored blog. When sources conflict, show the range and say which you used and why — judges read that as maturity, not weakness.
AI tools, used like a professional
LLMs are excellent at summarising annual reports, drafting survey questions, and listing sources to check — and terrible as a primary source. Rule: AI may point, only tiers 1–4 may be cited. Competitions increasingly ask teams to declare AI use; declare it plainly.