SDSKUba Divedeep local deal intelligence
More
Public beta
Menu

OUR METHODOLOGY

A big percentage is not enough.

Every candidate deal is evaluated across product identity, realistic comparison price, location, freshness, evidence and purchase risk.

1. Match the exact product

We normalize retailer identifiers such as SKU, model, UPC, ASIN and TCIN, then check variants and pack sizes. Similar-looking products are not treated as identical.

2. Establish a realistic street price

Manufacturer list price alone can exaggerate a discount. The comparison price should reflect recent, credible selling prices for the same product and configuration.

3. Measure local relevance

Store-specific markdowns are tied to a location and search radius. A reported price in another region is a lead—not proof of a nearby deal.

4. Separate signals from facts

Public chatter can identify an early lead. It remains separate from verified offers until identifiers, time, location and an approved retailer or in-store observation corroborate it.

5. Apply freshness and risk gates

Older observations, uncertain inventory, mismatched variants and checkout failures reduce confidence. Expired or contradicted records are removed from verified results.

6. Rank value before compensation

Deal scores may consider discount quality, distance, evidence confidence, availability and risk. Affiliate commission is excluded from the ranking formula.