Large language model (LLM)-native advertising has emerged as a viable monetization channel as LLM service providers seek sustainable revenue streams. However, integrating native ads into LLM responses introduces two largely understudied challenges: (1) controlling ad intensity -- how prominently advertising content appears -- and (2) ad labeling -- identifying which segments of the generated text constitute advertising. We propose AdKnob, a framework jointly addressing both challenges. For ad intensity control, we introduce new ad control tokens without pretrained semantic associations and align the model via direct preference optimization to generate ads at each desired level. For ad labeling, we propose an attention rollout-based attribution mechanism that identifies ad segments by tracing their attribution to input ad-related tokens, requiring no additional inference cost. We further construct MI-KnobSet, the first multi-level ad intensity preference dataset for LLM-native ads. Experiments across six LLMs and human evaluations show that AdKnob achieves uniformly spaced ad intensities where baselines collapse to a narrow range, achieves the highest ad labeling accuracy while being over 500x faster than the strongest baseline, and is preferred over baselines by human evaluators up to 89.6% of the time.
Figures & tables
Figure 1: AdKnob addresses two challenges in LLM-native advertising: (a) ad intensity control conditioning the model to generate responses at a desired advertising intensity and (b) ad labeling identifying ad segments within the response without additional inference costs.
Figure 2: Overview of AdKnob. The LLM is aligned via DPO with level-dependent preference pairs conditioned on ad control tokens, enabling the model to generate responses across different ad intensities. We also strengthen attribution from input brand/ad copy tokens to identify ad segments by supervising attention rollout scores, enabling ad labeling without additional inference costs.
Figure 3: Candidate responses show heavily overlapping advertisement scores, indicating that natural language prompts are insufficient for ad intensity control.
Figure 4: Ad intensity results across four base models. Our AdKnob (L1–L4) achieves a wide and uniform range of ad intensities, while baselines are constrained to a single or a narrow intensity range.
Qwen3-4B
Phi-4-Mini (3.8B)
Method
AUC
Prec.
Rec.
F1
Time (ms)
AUC
Prec.
Rec.
F1
Time (ms)
3–4B
Marker
0.58
0.62
0.37
0.47
–
0.53
0.47
0.14
0.22
–
Models
BERTScore
0.68
0.64
0.73
0.68
1,785
0.70
0.56
0.78
0.65
1,769
AttnLRP
0.48
0.57
0.81
0.67
179
0.53
0.30
0.83
0.44
151
Ours w/o Lattr
0.69
0.64
0.61
0.62
3.3
0.75
0.46
0.71
0.56
2.5
Ours
0.77
0.67
0.89
0.76
3.3
0.76
0.62
0.93
0.75
2.5
Table 1: Ad labeling performance at thresholds δ∗=1.0 , τ=0.005 . Time denotes average labeling time per response (– for Marker , as it generates tags alongside the response). Best results in bold .
Figure 5: Qualitative example of ad intensity control and ad labeling. At Level 1, the advertisement is subtly integrated alongside other options, while at Level 4, the response is heavily focused on advertising. Regions highlighted with [AD] indicate that AdKnob correctly labels ad segments.
Table 2: User study results. (a) Human-rated ad intensity and response quality (1–4 Likert scale). (b) A/B preference test win rates (%) of AdKnob (Level 3) vs. baselines. Green background indicates favorable comparison ( > 50%). (c) Ad labeling evaluated on human-annotated labels.
Figure 6: Ablation studies. (a) Ad control tokens vs. natural language descriptors. (b) Effects of the rejected samples in preference pair construction. (c) Sensitivity to γ in preference pair construction.
Appendix figures & tables29 assets
Supplementary material from the paper’s appendix.
Appendix
Figure A1 : Annotation interface for metric validation and ad labeling
Figure A2 : Annotation interface for the preference test
User Preference
Click Preference
Baseline
Helpful
Discloses Ad
Natural
Relevant
Not Pushy
Stands Out
Two-step
88.4
74.4
55.8
73.7
89.5
7.9
Prompt-based
84.6
59.0
38.5
78.4
64.9
5.4
JU-NLP
86.8
55.3
28.9
63.2
71.1
13.2
NL-prompt
58.3
80.6
50.0
52.8
77.8
11.1
Average
79.5
67.3
43.3
67.0
75.8
9.4
Appendix
Table A1 : Justification reasons (%) for preferring AdKnob over each baseline, reported separately for user preference and click preference axes. Values indicate the fraction of annotators who selected each reason among those who preferred AdKnob.
Metric
Pearson r
Spearman ρ
shit
0.898
0.906
ssent
0.854
0.781
shit⋅(ssent+1)/2
0.875
0.896
srel
0.846
0.789
sad
0.927
0.909
Appendix
Table A2 : Instance-level correlation between advertisement score components and human-perceived ad intensity. All correlations are statistically significant ( p<0.001 ).
Combination form
Pearson r
Spearman ρ
Equal
0.905
0.874
Product
0.882
0.906
Learned
0.934
0.911
sad (Ours)
0.927
0.909
Appendix
Table A3 : Correlation with human-perceived ad intensity across combination forms of the advertisement score. Learned fits linear regression weights on human-perceived intensity (5-fold cross-validated).
Task
# HITs
Payment / HIT
Est. time / HIT
Hourly rate
Metric validation & ad labeling
913
$2.00
8–10 min
∼ $12
Preference test
1,220
$0.50
2–3 min
∼ $12
Appendix
Table A4 : Summary of annotation statistics for each evaluation task.
Figure A3 : Ad intensity results on LLaMA-3.2-3B and LLaMA-3.1-8B. Our AdKnob (L1–L4) achieves a wide and uniform range of ad intensities, while baselines are constrained to a single or a narrow intensity range.
Figure A4 : Comparison with closed-source LLMs. NL-prompt is applied to GPT-4o and Claude Opus 4.8, and compared with AdKnob on Qwen3-4B.
Figure A5 : Zero-shot ad intensity results of AdKnob (Qwen3-4B) on real, non-synthetic inputs, with user requests from WildChat-1M [ 46 ] and ad copies from the Programmatic Ad Creatives dataset [ 26 ] .
Method
AUC
Prec.
Rec.
F1
Marker
0.67
0.57
0.47
0.52
BERTScore
0.63
0.49
0.69
0.57
AttnLRP
0.48
0.46
0.89
0.62
Ours w/o Lattr
0.81
0.50
0.79
0.61
Ours
0.83
0.61
0.90
0.73
Appendix
Table A5 : Zero-shot ad labeling performance of AdKnob (Qwen3-4B) on real, non-synthetic inputs.
Figure A6 : Generalization to unseen domains on Qwen3-4B trained on 8 of the 10 meta-topics (Healthcare and Gaming held out). Seen and Unseen denote test instances from the training and held-out meta-topics, respectively.
Figure A7 : Zero-shot multilingual results of AdKnob (Qwen3-4B) trained on English MI-KnobSet. User requests and ad copies are translated into Spanish, Chinese, and Hindi. Scores are measured with multilingual evaluators.
Language
AUC
Prec.
Rec.
F1
English
0.77
0.67
0.89
0.76
Spanish
0.73
0.65
0.90
0.75
Chinese
0.73
0.61
0.84
0.71
Hindi
0.81
0.64
0.87
0.74
Appendix
Table A6 : Zero-shot multilingual ad labeling performance of AdKnob (Qwen3-4B) trained on English MI-KnobSet.
Ad level for each turn
Turn 1
Turn 2
Turn 3
L1-L1-L1
0.247
0.267
0.260
L4-L4-L4
0.543
0.490
0.491
L1-L1-L4
0.247
0.267
0.498
L4-L1-L1
0.543
0.254
0.233
Appendix
Table A7 : Advertisement scores of AdKnob (Qwen3-4B) in 3-turn conversations under four per-turn ad level setups.
Figure A8 : Generalization across different numbers of intensity levels. AdKnob maintains uniform intensity control and response quality across different numbers of levels.
Figure A9 : Ad control token embedding initialization. Our random initialization achieves the best balance of intensity control and response quality.
Figure A10 : Continuous ad intensity control on Qwen3-4B trained on 4 discrete levels, obtained by linearly interpolating between the embeddings of adjacent ad control tokens in intervals of 0.25. Larger markers denote the discrete levels (L1–L4).
Method
w/o masking
w/ masking
Δsad
Ldpo only
0.399
0.318
−0.081
Ldpo+Lattr (Ours)
0.388
0.213
−0.175
Appendix
Table A8 : Faithfulness of ad attribution on Qwen3-4B. We report sad before and after masking the response tokens with the highest attribution scores. A larger drop indicates that the attribution better identifies ad-related tokens.
Model
Method
δ∗=0.5
δ∗=1.0
δ∗=1.5
δ∗=2.0
AUC
Prec.
Rec.
F1
AUC
Prec.
Rec.
F1
AUC
Prec.
Rec.
F1
AUC
Prec.
Rec.
F1
Qwen3-4B
Marker
0.59
0.73
0.36
0.48
0.58
0.62
0.37
0.47
0.57
0.54
0.37
0.44
0.56
0.47
0.36
0.41
BERTScore
0.66
0.72
0.71
0.71
0.68
0.64
0.73
0.68
0.69
0.59
0.74
0.65
0.70
0.55
0.75
0.63
AttnLRP
0.50
0.67
0.82
0.74
0.48
0.57
0.81
0.67
0.49
0.52
0.81
0.64
0.48
0.48
0.81
0.60
Ldpo only
0.69
0.70
0.58
0.64
0.69
0.64
0.61
0.62
0.69
0.58
0.61
0.60
0.69
0.54
0.62
0.58
Ours
0.75
0.76
0.87
0.81
0.77
0.67
0.89
0.76
0.77
0.60
0.90
0.72
0.77
0.56
0.91
0.69
Appendix
Table A9 : Ad labeling performance across pseudo-label thresholds δ∗ at inference threshold τ=0.005 . Best results in bold .
Decoding strategy
AUC
Prec.
Rec.
F1
Temp 0.7
0.76
0.66
0.89
0.76
Temp 1.0
0.75
0.67
0.90
0.77
Greedy (Ours)
0.77
0.67
0.89
0.76
Appendix
Table A11 : Ad labeling performance across decoding settings on Qwen3-4B at δ∗=1.0 , τ=0.005 .
Figure A11 : DPO vs. SFT. Our DPO achieves a wider intensity range and higher response quality than SFT.
λ
squal ↑
AUC ↑
F1 ↑
0 (= Ldpo only)
0.656
0.69
0.62
0.01
0.651
0.74
0.67
0.1
0.650
0.75
0.69
1.0 (Ours)
0.647
0.77
0.76
10.0
0.636
0.84
0.79
Appendix
Table A12 : Analysis on λ . We report squal averaged across all levels and labeling performance.
# Instances (train / test)
3,946 / 913
# Candidate responses per instance
25
# Total candidate responses
98,650
# Preference pairs per instance
12
# Total preference pairs
47,352
Avg. user request length (tokens)
14 ± 4
Avg. brand length (tokens)
3 ± 1
Appendix
Table A13 : Summary statistics of MI-KnobSet.
Banking
Car
Gaming
Healthcare
Real Estate
Restaurant
Shopping
Streaming
Vacation
Workout
Total
Train
390
455
522
323
352
392
409
485
355
263
3,946
Test
72
105
105
63
58
86
36
100
232
56
913
Appendix
Table A14 : Distribution of instances across meta topics.
Figure A12 : Distribution of candidate responses in the advertisement and response quality score space, colored by system prompt type. The candidate responses span a wide range of both scores, enabling us to generate preference pairs with clear distinctions. Despite using prompts with different levels of ad integration, responses from different prompt categories are heavily intermixed, demonstrating the need for explicit ad intensity control via ad control tokens and DPO alignment.
Figure A13: Qualitative comparison between Prompt-based and AdKnob (Level 4) on Qwen3-4B. Green highlights indicate mentions of the user-requested topic ( Modern Warfare 2 ), red highlights indicate mentions of the advertised brand ( Call of Duty: Warzone ), and blue highlights indicate mentions of the ad copy content. Prompt-based fails to address the user request while inserting ad content with weak relevance to the ad copy, whereas AdKnob naturally integrates all three elements in a balanced response.
Figure A14: Qualitative example of ad intensity control on Qwen3-4B across four intensity levels. Red highlights indicate mentions of the advertised brand ( Bank of America Advantage Plus ), and blue highlights indicate mentions of the ad copy content. As the level increases, AdKnob progressively incorporates more brand mentions and ad copy content into the response.
Figure A15: A failure case of qualitative language descriptor on Qwen3-4B. Orange highlights indicate descriptor words ( e.g. , “medium-lo”, “low”) that appear contextually irrelevant in the generated response, confirming that pre-trained semantics of natural language descriptors degrade response quality.
Figure A16: A failure case of quantitative language descriptor on Qwen3-4B. Orange highlights indicate descriptor words ( e.g. , “Level 4”) that appear contextually irrelevant in the generated response, confirming that pre-trained semantics of natural language descriptors degrade response quality.
Online advertising studies how digital ads are targeted, ranked, priced, measured, and delivered while balancing budgets, business goals, and user interests.
Large language model (LLM) services face rapidly rising serving costs. To sustain free or affordable access for a broad user base, alternative monetization mechanisms are needed. Native advertising offers a promising approach, as it creates economic value by aligning user intent with merchant offerings. Yet how to integrate sponsored content naturally into LLM responses remains largely underexplored. Existing approaches primarily rely on prompt-level injection, multi-pass rewriting, or fine-tuning, which can compromise response quality or introduce additional computational costs. We propose AdSteer, a lightweight decoding-time advertising framework that adaptively adjusts token probabilities to integrate contextually appropriate advertisement into responses. AdSteer calibrates the strength and duration of advertising influence, making sponsored content a natural part of the response. Experiments on GEM-Bench across three application scenarios show that AdSteer achieves an average improvement of 14.1% in overall score over the strongest baselines and 22.5% fewer total tokens on average than rewrite-based methods. These results demonstrate that AdSteer provides an effective and practical solution for native advertising in LLMs.
Shuhang Lin, Chuhao Zhou, Zhenyu Wang +4
Rutgers University · Nanyang Technological University · UIUC +2
Reconciling platform revenue with user experience in LLM advertising motivates a data-centric foundation. We introduce NaiAD, the first comprehensive dataset for LLM-native advertising comprising 58,999 carefully constructed ad-embedded responses paired with user queries. NaiAD is organized around theoretically grounded evaluation metrics that separately and comprehensively capture user and commercial utility. To mitigate the dimensional collinearity of aligned LLMs, we propose a decoupled generation pipeline that produces structurally diverse samples, ranging from responses that explicitly disentangle stakeholder utilities to responses that are uniformly strong or weak across dimensions. We further provide score labels calibrated by a Variance-Calibrated Prediction-Powered Inference (VC-PPI) framework, aligning automated scoring with human annotations. Mechanistic analyses reveal that successful ad integration relies on reasoning paths that cluster into four distinct semantic strategies. Models leveraging NaiAD internalize these strategies to simultaneously improve user and commercial utility, while enabling independent control over these distinct objectives via in-context learning. Together, these results position NaiAD as a foundational infrastructure for developing future LLM-native ad systems.
Yihang Zhang, Zimeng Huang, Ren Zhai +2
Tsinghua University Beijing, China · College of AI Tsinghua University Beijing, China · Department of Literature, Arts and Communication Anhui International Studies University Anhui, China +1
How to monetize large language models (LLMs) by naturally integrating sponsored content into their responses, known as LLM-native advertising, has recently emerged as a critical problem. However, existing solutions entangle advertising with content generation inside a single model, which is incompatible with modern API-only or workflow-based LLM applications and inevitably compromises the original response quality. To address this, we propose PILA, which reformulates ad insertion as a conditional response rewriting problem and decouples it from the upstream service as a lightweight sidecar module. PILA is model-agnostic and can be seamlessly integrated with existing LLM services without modifying the base model or its workflow. It also exposes a controllable trade-off between user-side naturalness and ad-side exposure, offering a practical interface for downstream pricing and deployment. Experiments across diverse upstream models show that \pila consistently improves ad effectiveness while preserving response quality, highlighting its promise as a practical solution for LLM-native advertising.
Zhaowei Zhang, Yuhan Fu, Yihang Zhang +6
Institute for AI, Peking University · College of AI, Tsinghua University · University of Michigan +2