How results work

What our tools actually do

Every generator on this site picks one option (or a set) from a curated list. Some pickers add a spin animation, some add a prompt template, but the underlying step is the same: pick, show, let you pick again if the first one didn\'t land.

Where the randomness comes from

We use the browser\'s crypto.getRandomValues() API — the same interface that browsers use for cryptography. It is not the same as Math.random(). It is designed to be unpredictable, uniformly distributed, and safe for high-stakes selection like keys and tokens. For our pickers, it is significant overkill — and that\'s the point.

On very old browsers without crypto.getRandomValues, we fall back to Math.random(). Everyone on a mainstream 2016+ browser gets the cryptographic-strength path.

What "the wheel" is really doing

The spinning wheel is a UI flourish. The pick happens first, then the wheel is animated to land on the chosen slot. This is standard practice — the animation feels satisfying without slowing you down. Nothing biases the outcome.

Where the item lists come from

Every dataset is hand-curated. Movies, cocktails, workouts, holiday destinations — a human wrote them, ranked them, and prunes them. This is why our pickers feel more useful than an infinite-item generator: the small pool is deliberately good.

How we test that it\'s actually fair

Each dataset ships with an automated audit that checks: no duplicates, no near-duplicates, sensible entry lengths, no placeholder phrases. Randomness itself is testable — we spot-check by running thousands of picks through a distribution test and confirming near-uniform coverage. The runtime behaviour is provably fair.

The honest limits

• Custom-list mode: fairness depends on your list. Duplicate an item and it comes up twice as often.
• Wheel-of-names: cryptographically random. If it "feels rigged", that\'s the gambler\'s fallacy — you sometimes get the same name twice in a row, and that is completely normal.
• Recommendation engine (Related Tools V2): not random — it uses category, keyword and commercial-fit signals. See Related Tools below.

The related-tools engine

The "You might also like" suggestions under each result are algorithmic, not random. We rank other tools by: hand-picked relatedness, shared category, shared keywords, matching utility kind, commercial fit, and indexation priority. The reason tag on each suggestion (e.g. "Same category", "Similar intent") tells you which signal fired.

Analytics we collect

When enabled, we collect anonymous event data — which utility was opened, whether a share button was pressed, whether a related tool was clicked. No cookies for personalisation. No third-party tracking beyond GA4 and (if enabled) AdSense. See our privacy policy.

Storage on your device

Favourites, recently used tools, and custom lists live in your browser\'s localStorage. Clear your browser data and they\'re gone. Nothing is synced across devices, and we can\'t see them.