Same Identifier, Different Offer: A Scoped Evidence Model for Product and Offer Comparability in B2B Research

Product identifiers retrieve candidate records but do not, on their own, establish interchangeable buying offers. This paper combines a purposive synthesis of identifier rules, merchant-feed specifications and entity-resolution literature with a reproducible, synthetic specification exercise. Its contribution is a scoped evidence model that preserves known differences alongside unresolved fields and separates content comparison from two-level packaging comparison.Twenty-four constructed pairs produce seven equivalent, five different and twelve unresolved content outcomes; a stricter packaging scope produces four, seven and thirteen respectively. Forty-eight outcome assertions, forty-eight symmetry assertions and seventeen missing-field checks pass. These are consistency checks, not marketplace accuracy measurements. The result is a transparent review method: state the comparison purpose, normalize only supported quantities, expose differences and unknowns, and keep buyer acceptance separate. No production capability, customer saving or real-world error rate is established.This is the unchanged working paper v1.0, originally published on Zenodo on 2 November 2026. It has not been peer reviewed. The canonical manuscript and reproducibility materials are available at https://doi.org/10.5281/zenodo.23897468.Mohammad Maani is Founder of QuickAIBuy and has a commercial interest. OpenAI ChatGPT/Codex assisted with source retrieval, synthesis and drafting. No independent human code replication is claimed. The manuscript contains the full contribution and rights statements.