Review Methodology — Novibet

The overall score is a weighted average of six sub-scores, each on a 0-10 scale. Cashier and payments in NZD carry 25% of the total. Bonus value, net of wagering, carries 20%. Casino game library, including live dealer, carries 20%. Sportsbook markets and odds carry 15%. Mobile app quality carries 10%. Customer support, licensing and complaints handling carry the remaining 10%. The weights are chosen to reflect what actually determines a New Zealand player's experience over a three-to-twelve-month horizon.

We test with real money, from freshly created accounts, on the same devices a typical reader would use. Deposits and withdrawals are put through in ranges that match the site's own limits, not in token amounts. Bonuses are wagered to completion — or to failure — so we know what the terms mean in practice. The app is installed and used for at least a week on both an iPhone and an Android device covering the mid-range of what New Zealand players own. Every test is logged with the date, the device, the account age and the outcome; the log is what the fact-checker reads before signing off on the score.

Withdrawal delays beyond the published SLA cost points immediately, and continue to cost points until the operator either speeds up or publishes a longer SLA. Bonus terms that materially differ from the offer headline — an advertised 100% match that turns out to be 40% on first deposit after fine print — trigger a permanent downward adjustment. App builds with crash rates above 2% on mid-range devices, live-chat wait times above ten minutes at evening peak and unresponsive KYC handling are the other three levers we watch.

Each sub-score is anchored to specific observable behaviour rather than to a feeling. A 9 in cashier means every payment method advertised for NZ works, withdrawal timings match the SLA within tolerance and no reader complaint is currently open. A 7 means one method underperforms or one SLA slips by 20%. A 5 means something structural is off — a method is silently disabled, an SLA is missed by more than half, or a specific bank rail is blocked without an explanation. Below 5, we stop reviewing and downgrade to a warning.

The full review is refreshed every six months, regardless of whether anything has visibly changed. Between refreshes, sub-scores can move based on the two-week cycle described on the team page — but only downwards without a full review, never upwards. Upgrading a sub-score requires the full test protocol. That rule stops the score drifting upwards on the back of thin evidence in the middle of a cycle.

The full criteria list lives in a public working document linked from our team page. It runs to roughly twenty pages and covers everything from the acceptable spread on prop bets to the size of the buttons on the mobile cashier. The criteria are versioned; when we materially change one, the version bumps and the previous version stays on file for readers who want to check whether an older score is still comparable to a newer one.