Buyer and seller value from optional automated bargaining in electronic commerce

Abstract Whether automated bargaining benefits electronic commerce depends on the transaction route it changes, not only on the quality of its negotiated offers. A checkout-relative value model is developed to evaluate an optional seller-side bargaining agent against existing posted-price checkout. The model separates new-sale margin, discounts on otherwise feasible checkout purchases, unrecovered sales and participant interaction costs. It derives conditional cost and fallback limits for seller and buyer outcomes. Public product-price and negotiation inputs support a matched computational comparison of two rules and two frozen fitted Q iteration policies across 193 evaluation products. In the reference configuration, the full-state learner adds 0.0144 seller and 0.0039 buyer listing-price units per original opportunity, but a simpler learner provides higher combined value. Stubborn offers produce seller losses, while low valuations produce buyer losses after interaction costs. These results show how a seller-favoured bargaining policy can differ from a configuration that benefits both participants on average. The contribution is a reproducible model linking executable policy outcomes to the economic consequences of an optional transaction channel. Participation, customer values, fallback and effort costs remain stipulated; the findings are conditional computational evidence, not estimates of actual adoption, customer experience or realised commercial effects. Positive reference averages conceal product heterogeneity: only 82 of 193 product contexts yield positive means for both parties. Structural sensitivities further qualify deployment claims.