Design a "best value" score — and stress-test it
Every "best phones under a budget" list hides a formula. Build yours in the open: define a composite value score, apply it to 10 real products, let AI attack your weights, and see if your winner survives the stress test.
Posted by The AI Internships
The work brief
- 01Pick a category you’d genuinely research (phones, earbuds, laptops, bicycles) and a budget cap.
- 02Collect real specs + current prices for 10 products into a CSV from real listings.
- 03Design your score: factors, normalisation, weights — justify each weight in writing, then ask AI to argue against your choices and fix what it genuinely catches.
- 04Stress-test: change the weights two different reasonable ways. Does the winner change? Report what that says about your metric.
What you’ll produce
Submission standard
Submit your metric design, the 10-product CSV, the ranked result, and the sensitivity analysis. Include your AI workflow — tools, best prompts, and what you changed. Attach dated screenshots of at least three real listings and add a source-link column so your prices are verifiably current.
Your score: factors, normalisation, weights — and why each weight
Written responseRequiredThe 10 products as CSV: specs, current price, computed score
10 products (specs, current price, computed score) with a source-link column per product; prices must be real and current, matching your listing screenshots.
CSV dataRequiredThe ranking and your winner — with the one-line pitch you’d give a buyer
Written responseRequiredThe stress test: two alternate weightings, whether the winner survived, what that means
Written responseRequired
You’ll complete these inside your private workspace.
What you must submit as proof
This brief requires evidence an AI can’t fabricate.
Dated screenshots of at least 3 of the real listings (product page showing model, price and date)
Show real store/manufacturer pages with visible prices and dates that match rows in your CSV.
Image uploadRequired
Submissions without this evidence cannot be submitted.
Protect other people in your proof. Blur faces, names, phone numbers and email addresses before you upload, and refer to anyone you worked with by role or number ("Listener 1", "the stall owner"). Your proof is only ever used to check your work — it is never published, never appears on your certificate, and is never shown in your public portfolio.
How your work is evaluated
The passing benchmark is 70/100.
Metric design
22%Factors and weights are justified; normalisation is correct.
Real product data
11%Specs and prices are real, current, and consistently collected.
Sensitivity honesty
22%The stress test is genuine and its implications are faced.
Real, current prices
44%The dated listing screenshots and per-row source links must show the specs/prices are real and current, matching the CSV. Invented or unsourced prices are a fail regardless of how elegant the score is.
How we grade your AI usage
Using AI is the point — it’s the skill this certificate proves. You’ll answer three short questions about how you used it: what you asked, what was wrong with its first answer, and what you changed. Specific, honest answers score high. “I pasted the brief and submitted the answer” scores near zero.