With the help of co-branding to launch new products has become an important way for enterprises to break through their own development shackles, but how to reverse the negative impact of co-branding? Based on the adaptive learning theory, this paper adopts three experiments to empirically examine the impact of co-branding on the evaluation of new products by weak brands and its underlying mechanisms. The results show that: (1) Co-branding between strong and weak brands indeed exerts a negative impact on the evaluation of new products by weak brands. (2) Co-branding tightness plays a significant moderating role. High co-branding tightness can reverse this negative impact, and when co-branding tightness is low, theme-similar (vs. category-similar) co-branding mitigates this negative impact by enhancing consumers’ brand perception capability. (3) Brand width also serves as a significant moderator. Great brand width can reverse the aforementioned negative impact, and when brand width is narrow, category-similar (vs. theme-similar) co-branding alleviates this negative impact by boosting consumers’ brand perception capability. This paper expands the explanatory boundary of the co-branding theory and provides a unique theoretical perspective for reversing the negative impact of co-branding, offering significant practical implications for co-branding management.