Organizations: University of Science and Technology of China, China, Hefei · Hohai University, China, Nanjing · National University of Singapore, Singapore, Singapore
With the rapid growth of online video consumption, video advertising has become increasingly dominant in the digital advertising landscape. Yet diverse users and viewing contexts make one-size-fits-all ad creatives insufficient for consistent effectiveness, underlining the importance of personalization. In practice, most personalized video advertising systems follow a retrieval-based paradigm, selecting the optimal one from a small set of professionally pre-produced creatives for each user. Such static and finite inventories limit both the granularity and the timeliness of personalization, and prevent the creatives from being continuously refined based on online user feedback. Recent advances in generative AI make it possible to move beyond retrieval toward optimizing video creatives in a continuous space at serving time. In this light, we propose NextAds, a generation-based paradigm for next-generation personalized video advertising. We articulate its key principles and conceptualize an agentic workflow for closed-loop creative optimization, spanning planning, generation, verification, and reflection. We further explore how personalized creatives can evolve across impressions through feedback-driven refinement and progressive storytelling. To further ground this vision, we formalize two representative tasks and conduct exploratory experiments on corresponding lightweight benchmarks, revealing both the promise and limitations of existing GenAI. Moreover, we discuss potential research opportunities and the broader implications of increasingly personalized and native advertising, highlighting issues that warrant attention from the community and aiming to provide insights for future research and responsible deployment. Our code, benchmarks, and demonstrations are available at https://anonymous.4open.science/w/NextAdsProjectPage/.
Figures & tables
Figure 1. The evolution of video advertising has largely followed two dimensions: personalization and nativeness, yet high production costs and manual creative bottlenecks prevent the industry from reaching the upper-right corner.
Figure 2. Paradigm shift: from retrieval to generation. NextAds transforms personalized video advertising from retrieval over a discrete creative inventory to generative optimization in a continuous space. A comparison between retrieval-based video advertising and generation-based NextAds optimization.
Figure 3. Agentic workflow of NextAds , where four core agents, Director , Producer , Verifier , and Reflector , collaborate with Hierarchical Memory to enable closed-loop refinement through pre-deployment verification and post-deployment user feedback.
Figure 4. NextAds in action: (a) For standalone creative generation, NextAds transforms user history and product information into a personalized creative plan and realizes it as a video ad. (b) For context-aware creative integration, NextAds additionally conditions on the host video to identify a natural insertion point and generate an ad segment that blends into the viewing context. Beyond one-shot generation, Verifier and Reflector close the loop by (c) revising problematic creatives before deployment and (d) evolving creatives across impressions through feedback-driven refinement and progressive storytelling.
#PCG-Bench
Personalization
Product Alignment
Integration Quality
Diversity
Video Quality
Presentation
Content
MS
DD
AQ
IQ
GenericAds (Closed)
1.77
1.52
8.48
-
7.10
0.9756
0.9742
0.5494
0.6524
NextAds (Closed)
7.12
5.09
8.58
-
6.57
0.9406
0.9797
0.5259
0.6339
NextAds (Open)
6.01
3.35
3.44
-
5.40
0.9693
0.9800
0.4064
0.6199
Table 1. Overall performance comparison of existing GenAI in both PCG and PCI tasks. The best results are highlighted in bold.
Figure 5. Overview of the PCG pipeline.
Figure 6. Examples generated by NextAds on PCG-Bench.
Appendix figures & tables2 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 7. Overview of the PCI pipeline.
Figure 8. Examples generated by NextAds on PCI-Bench.
With the rapid growth of online video consumption, video advertising has become a dominant component of digital advertising. Most personalized video advertising systems follow a retrieval-based paradigm, which matches ad content with user preferences in a feature space and recommends the most similar candidates. However, since the candidate creatives are pre-produced, this paradigm limits the granularity of personalization and leads to performance disparities across diverse user groups. To address this limitation, we propose a multi-agent personalized script-driven video advertising system (MAPS). We design four collaborative agents that form a closed-loop pipeline of user profile modeling, script generation, video synthesis, and feedback-driven iterative refinement. To train the Scripter Agent, we develop a reinforcement learning framework equipped with a multi-dimensional reward function, which steers the agent toward producing scripts that simultaneously align with user content preferences and target product information. Extensive offline experiments on representative datasets demonstrate that our approach achieves encouraging performance in generating and integrating personalized creatives. Online A/B experiments in a specific scenario in the Kuaishou advertising system show that MAPS improves CTR by 2.87%, CVR by 2.74%, and ARPU by 9.85% over the generative baseline.
Advertising video generation is not only a video synthesis task, but also a product-conditioned reasoning problem whose success is measured by online business metrics. Recent video foundation models can generate realistic clips from multimodal conditions, yet they do not optimize how a product should be transformed into an effective advertisement or how future generation should be improved from online business feedback. To close this loop, we propose AgenticGen, a reward-guided agentic framework that decomposes advertising video generation into two trainable reasoning stages, strategy selection and draft generation, thereby exposing optimization targets that online business feedback can supervise. AgenticGen learns a performance-based reward from accumulated online feedback and a complementary rubric-based reward aligned with human quality standards, then uses them to supervise policy optimization. DPO first moves the agentic policies toward online preferences, and GRPO further refines both stages with process and outcome rewards. Offline experiments validate the reward models and successive policy optimization. Online A/B experiments in the TikTok advertising system show that AgenticGen after DPO and GRPO improves CTR by 2.72%, CVR by 2.63%, and Advv by 9.61% over the SFT baseline.
Xingyuan Bu, Chengru Song, Hao Zhou +9
ByteDance · Tsinghua University · The Chinese University of Hong Kong
Generating realistic and user-preferred advertisements is a key challenge in e-commerce. Existing approaches utilize multiple independent models driven by click-through-rate (CTR) to controllably create attractive image or text advertisements. However, their pipelines lack cross-modal perception and rely on CTR that only reflects average preferences. Therefore, we explore jointly generating personalized image-text advertisements from historical click behaviors. We first design a Unified Advertisement Generative model (Uni-AdGen) that employs a single autoregressive framework to produce both advertising images and texts. By incorporating a foreground perception module and instruction tuning, Uni-AdGen enhances the realism of the generated content. To further personalize advertisements, we equip Uni-AdGen with a coarse-to-fine preference understanding module that effectively captures user interests from noisy multimodal historical behaviors to drive personalized generation. Additionally, we construct the first large-scale Personalized Advertising image-text dataset (PAd1M) and introduce a Product Background Similarity (PBS) metric to facilitate training and evaluation. Extensive experiments show that our method outperforms baselines in general and personalized advertisement generation. Our project is available at https://github.com/JD-GenX/Uni-AdGen.