How Tony Shopping works
Last updated: 2026-05-25
Why this page exists
Tony Shopping is an AI meta-shopping agent built to help you make thoughtful purchase decisions. In line with the US FTC endorsement guidelines, the EU Digital Services Act's disclosure requirements, and Korea's KFTC review guidelines, this page lays out how Tony makes money, what data it touches, and what drives every recommendation. Questions? Use the contact channel at the bottom of this page.
Affiliate links and commissions
Every "Buy" button on Tony Shopping is an affiliate-marketing link. When you follow one and complete a purchase, that merchant pays Tony a small referral commission. You pay no extra and your price does not change. Active programs today: Amazon Associates, Coupang Partners, eBay Partner Network, AliExpress Affiliate, Shopee Affiliate, Lazada Affiliate, and Naver Shopping. Whether a listing earns Tony a commission has zero influence on how it is ranked — Tony's recommendation algorithm scores only on price, shipping, reviews, and authenticity.
Where prices come from
Prices are pulled in real time from each merchant's public API or via a meta-search aggregator (SerpAPI). FX rates refresh once a day from a public rate provider, and converted prices are for reference. The final price you pay — including discounts, shipping, taxes, customs, and the FX rate at checkout — is whatever the merchant shows on their checkout page. Tony cannot be held responsible for differences. If you spot a stale or wrong price, tell us via the contact channel; we update adapters fast.
How Tony Score is calculated
Tony Score is a 0–100 weighted blend of four signals: (1) Similarity 25% — how close the listing matches your query, (2) Price edge 35% — how it compares to the rest of the cohort, (3) Review trust 15% — rating, review volume, and bot-pattern penalty, (4) Authenticity 25% — official-seller status, past complaint history, seller reputation. The "priority" chips on the compare page tilt these weights toward value, fast shipping, or authenticity. Every weight is documented and never changes based on ad spend.
Recommendations vs. advertising
The headline "Pick this" verdict on each search (the VerdictCard) is produced purely by the algorithm above. No advertising revenue influences it. If Tony ever introduces explicit ad placements, they will carry an "AD" badge above the card and remain visually distinct from algorithmic picks. LLM-flavoured copy (the compare-page narrative and review summaries) is generated by Anthropic Claude or OpenAI; the data sent to those models is strictly limited to what is already shown on the page.
Data handling and privacy
By default Tony stores all activity (search history, shortlist, click log, priority learning) in your browser's LocalStorage. When you are signed out, no activity leaves your device. Sign in and Tony syncs your shortlist and clicks to a Supabase database so the experience travels across devices. Tony never sells or shares your activity with third parties and never exposes it to ad networks. Account deletion requests are handled via the contact channel. See the privacy policy for the full breakdown.
Contact
For pricing or authenticity errors, questions about the algorithm, or affiliate-program inquiries, email support@tonyshopping.io. We respond within 3 business days.