advertising platform engineers

Abstract: Click-through rate (CTR) prediction relies on historical behaviors and contextual features to estimate clicks on a target item. Existing Transformer-based CTR models often adopt index-based or temporal positional encodings and token-level causal interaction. However, we observe that most existing positional encodings do not outperform NoPE, while relaxing causal constraints within target-relative temporal groups also yields gains. These observations motivate a fundamental question: what positional encoding and interaction granularity does CTR prediction actually require? To address this question, we propose \textbf{TRIP}, which combines target-aware rank--time positional encoding with temporal-group interaction modeling. Specifically, Factorized Rank Encoding represents each behavior's ordinal position within the historical sequence through a low-rank factorization, while Target-Relative Time Encoding characterizes its temporal distance to the target on a coarse-to-fine scale. This joint design grounds each behavior's positional representation in its sequence context and temporal relation to the current target. Furthermore, Temporal Group-Causal Interaction enriches intra-group representations through bidirectional interaction and masked reconstruction while preserving inter-group causal order. Finally, behavior representations are condensed into compact group summaries for efficient group-level modeling in subsequent layers. Experiments on four public benchmarks show that TRIP improves AUC by up to 4.45\% relative to the strongest baseline.

Abstract: Large language models (LLMs) have demonstrated strong reasoning capabilities, motivating their use as the backbone of foundation recommendation models (FRMs). Existing methods enhance recommendations through explicit Chain-of-Thought (CoT) reasoning under a Think-then-Answer paradigm. However, explicit CoT incurs substantial inference overhead by generating lengthy reasoning traces and relies on manually designed templates that struggle to capture diverse, dynamic user interests. We propose OneLatent, an efficient latent reasoning framework that compresses explicit reasoning traces into several learnable latent tokens, enabling Latent-Reason-then-Answer inference without generating verbose traces. OneLatent first introduces Multi-View Adaptive CoT (MV-ACoT), which creates diverse, high-quality teacher-generated supervision by exploring user interests from multiple perspectives and automatically adapting reasoning complexity to each instance. Building on pretrained FRMs, it then uses a three-stage latent-token alignment paradigm to progressively internalize CoT traces into learnable latent tokens. Finally, a multistage curriculum-based post-training strategy activates latent-token reasoning for downstream recommendation tasks. Experiments on an industrial-scale Kuaishou dataset and the public Kuaishou LLM-Rec benchmark show that OneLatent consistently outperforms explicit CoT-based methods and traditional baselines. Compared with the Think and No-Think variants of FRMs, OneLatent improves SID@64 by 17.44% and 9.33%, respectively, while achieving over 17x higher online inference throughput. We further develop a production serving system for scalable, real-time FRM inference. An online A/B test in Kuaishou's local-services advertising scenario shows that deploying OneLatent with this system yields an estimated 9.6% revenue lift over strong online baselines, including OneRec and OneReason.

Abstract: In sponsored-shopping auctions, one advertiser may win several positions, and a product that is never displayed may determine the prices of several winners. Under generalized second-price (GSP) pricing without self-pricing, these interactions couple allocation and payment through a single global ranking. We prove that exact autobidding equilibria under budget and return-on-investment (ROI) constraints can be computed in polynomial time whenever either the number $n$ of advertisers or the number $K$ of display slots is fixed. Our algorithms rely on two structural reductions. With few advertisers, the payment polytopes of tied bid layers combine by a Minkowski sum, giving a fixed-dimensional representation of exponentially many rankings. With few slots, budget equilibria admit threshold recovery and an exact submodular description of attainable payments at the display boundary. For ROI, we prove that every group-consecutive equilibrium has a representative with at most $K+2$ nonunit multipliers, preserving the item and price in every displayed slot. In each case, the reduction leads to polynomially many systems of polynomial equations and inequalities in bounded dimension. The algorithms compute exact multipliers and explicit distributions over valid group-consecutive rankings, with support at most $n+1$; for ROI with few slots, this improves to $\min\{n+1,K+3\}$.

Abstract: A feed ranker scores candidates with a value model, a weighted sum of engagement predictions. Each prediction conflates a baseline, what this viewer, item or request would produce anyway, with the incremental amount by which this candidate exceeds it. The literature mostly removes the baseline. We argue that a multi-signal value model should keep the raw prediction and add the incremental part as a separately weighted term. On Instagram Feed, three debiasing families deployed independently, for viewer habit, item popularity and request context, are one parameterised operator, and a design procedure picks a point in it. Media-type (format) bias goes instead to a debiased label chosen by the same procedure. In a controlled four-arm ablation of deployed Feed terms (neither term, each alone, both), the two-term arm beats both single-term arms at production weights on point estimates in 11 of 13 judged (term, metric) pairs. Of the 26 direct contrasts, 17 are significantly for it and only 1 against it, on impressions. The ablation also suggests that the offline diagnostic should screen signals, not rank them. Such terms have served Instagram Feed production for most of a year. Beyond Feed, Meta, whose apps serve billions of people, has adopted the operator across surfaces, apps and both ranking stages. Its launches there drive sessions and daily active users, the core visitation and retention metrics.

Abstract: As AI-generated content becomes commoditized, economic value increasingly shifts from final artifacts to the upstream expertise used to specify inputs that steer LLM generation. Yet organizations lack methods for valuing this expertise and identifying which individual skills or configurations causally improve performance. This study examines generative marketing, where structured expert labels serve as skill primitives conditioning LLM-generated advertisements. Estimating their causal effects is difficult because outcomes reflect stochastic LLM mediation, high-dimensional label interactions, and multimodal confounding. This study proposes D-VAE (Deconfounder Variational Autoencoder), a counterfactual-inference framework integrating a multi-treatment deconfounder with causal representation learning. D-VAE recovers causally disentangled expert-labeling logic and derives PAC-based bounds on omitted-variable bias from single-treatment confounders. Applied to 29,177 foundation-makeup ad creatives, D-VAE outperforms state-of-the-art benchmarks. A field experiment shows that recommended configurations increase engagement by 32.7% and perform statistically indistinguishably from human experts, while a simulation study confirms causal attribution under complex confounding and treatment interactions. Overall, the study shifts value assessment from AI-generated products to skills, providing a foundation for pricing and optimizing AI-augmented skill portfolios.

Abstract: Large language models (LLMs) are now a standard source of query expansion terms but they hallucinate. They inject non existent entities and off-topic terms that add ranking noise and can hurt strong retrievers. Existing remedies such as corpus grounding, mutual verification and preference alignment treat hallucinated terms as an undifferentiated nuisance to be filtered by agreement heuristics, without a mechanistic account of why a given term appears or a principled criterion for how many terms to keep. We propose Subsequence-Association Filtered, breadth-aware query expansion (SAFE-QE), a training-free, retriever agnostic method that draws on the subsequence-association account of hallucination. We score each candidate expansion term by its corpus conditioned cross-context reproducibility, terms that reappear under randomised query contexts yet lack support in the pseudo-relevant set are flagged as products of dominant hallucinatory associations and down-weighted. Drawing on the characterisation of language generation as a trade-off between consistency (no hallucination) and breadth (coverage), we cast term selection as constrained coverage maximisation over the pseudo-relevant set, keeping a set that is approximately broad but consistent and stable across resampling. We also introduce the Hallucinated Expansion Term Rate, a query side faithfulness metric that counts injected expansion terms no relevant document supports. We evaluate our approach on benchmarked TREC and BEIR collections. Our SAFE-QE is statistically indistinguishable from generative and corpus-steered expansion on nDCG@10 while injecting fewer unsupported terms 4.3 per query against 27.8 for Query2doc and 57.5 for HyDE and on a strong learned sparse retriever, where every expansion method regresses. Faithfulness is therefore available at no measurable cost in ranking quality.

Abstract: Search advertising platforms routinely spend beyond an advertiser's average daily budget on high-traffic days, so long as total spending over the month stays within the monthly budget. Motivated by this practice, we study a $D$-day generalization of the Adwords problem (Mehta et al. 2007), where each advertiser $i$ has a nominal (average) daily budget $B_i$ and a total horizon (monthly) budget $DB_i$. Given a flexibility parameter $δ$, the platform may spend at most $δB_i$ on advertiser $i$ on any single day, subject to the horizon spending limit of $DB_i$. We quantify the power of $δ$-flexible budgets by benchmarking against the inflexible offline optimum, which may spend at most $B_i$ on advertiser $i$ on each day. We show that no amount of flexibility helps direct generalizations of the classical algorithm of Mehta et al. (2007). By contrast, for every fixed $δ$, we design an algorithm whose competitive ratio converges to $1-e^{-δ}$ as $D\to\infty$, and we show that this is asymptotically optimal. Perhaps surprisingly, this matches the optimal competitive ratio in a more permissive setting where the algorithm receives a fresh spending limit of $δB_i$ each day and may spend up to $δD B_i$ over the horizon. Along the way, we characterize the exact optimal competitive ratio for every pair $(D,δ)$ on high-traffic instances, where the offline benchmark exhausts every advertiser's budget on every day.

Abstract: We study the efficiency of auctions with budget-constrained bidders and derive tight price of anarchy (PoA) bounds for coarse correlated equilibria (CCE). To this end, we develop a general framework that reduces proving PoA guarantees to finding feasible solutions to the dual of a linear program. The LP is formulated over a small number of per-bidder equilibrium statistics, incorporates constraints implied by the auction setting and equilibrium behavior, and captures a novel smoothness notion tailored to budget-constrained environments; we provide a reduction from classical smoothness to this new notion. Our approach enables simple and tight PoA analyses across a broad range of auction formats. Using our framework, we obtain tight liquid welfare guarantees for simultaneous first-price, second-price, and all-pay auctions, simultaneous auctions with restricted uniform bidding interfaces, discriminatory and uniform price multi-unit auctions, and generalized first-price position auctions. Beyond budget-constrained settings, we derive new lower bounds for the budget-free uniform price auction and generalized second-price auction; in particular, our lower bound for the uniform price auction matches the upper bound of $3.146$ by de Keijzer et al. (2013).

Abstract: Click-Through Rate prediction, a core task in recommendation and advertising systems, relies on modeling interactions among sparse categorical features. Explicit cross networks are a central paradigm for CTR prediction, and recent progress has largely come from increasing the interaction capacity of a single predictor through deeper cross networks and more expressive cross operators. We revisit whether continually increasing interaction capacity remains the most effective way to improve predictive performance, and find that its benefits quickly exhibit diminishing returns even as capacity continues to grow. This motivates a complementary scaling direction that we call estimator scaling, where additional resources are used to incorporate multiple related estimators rather than only enlarging a single predictor. Through theoretical analysis, we show that the gains from estimator scaling are governed by the amount of non-shared predictive variation available across estimators. However, exploiting this variation naively can be expensive: independently trained models provide substantial estimator diversity but require deployment cost to grow with ensemble size. This motivates a parameter-efficient realization of estimator scaling that can incorporate diversity from multiple estimator sources without maintaining multiple full models. Building on this view, we introduce RECursive Averaged Predictor (RECAP), a parameter-efficient recursive CTR model that operationalizes estimator scaling at three levels: distillation across independently trained models, exponential moving averaging over training trajectories, and aggregation over inference-time routes within a weight-shared recursive backbone. Experiments across multiple benchmarks establish new state-of-the-art predictive performance on standard benchmarks, while placing the RECAP on a favorable performance-parameter Pareto frontier.

Abstract: Conversational agents, generative recommenders, and personalized advertising all rest on one capability: understanding each user from raw behavior. Prevailing industrial practice is task-specific: for each task, a relevant subsequence is extracted from the full history and a dedicated model trained on it. In production it hits two bottlenecks. First, even after filtering, a single-task sequence stays extremely long: content-interest summarization reads several hundred items per user, tens of thousands of tokens once serialized as prompt text. Second, profiles are refreshed routinely: a billion users weekly, roughly 100K QPM in aggregate, which under a fixed GPU budget sets a hard throughput floor. Compression is therefore mandatory, yet truncation or coarse compression can silently distort the profile, introducing four hallucination types (fabrication, omission, date misattribution, broken logic) that, with no way to evaluate the compressed representation itself, surface only as diffuse degradation in downstream metrics. We present KuaFu, a unified behavior-compression layer whose minimal unit is one behavior item. A two-axis projector compresses each item into 2-4 tokens of width 128-256 (about 10x along the token axis, 20x along width; per-item cache 10 KB to 0.5 KB), with fidelity-oriented four-stage training and layered intermediate evaluation. Across four production profiling tasks it matches or exceeds uncompressed single-task production models on all five headline metrics, raises per-GPU throughput by 37%-350%, and saves 190 GPUs. On public benchmarks it nearly always beats prior compressors at the same compression ratio (up to +17.7 EM on out-of-domain MRQA); on RecBench, a 4B model surpasses its 8B counterpart by 1.90 points. KuaFu has run on the Tencent advertising and recommendation platform for ten months, lifting overall GMV by 1.37%.

Abstract: Numerical embeddings for click-through rate (CTR) prediction are built on a convenient but restrictive premise: a scalar has one representation. This premise conflates where a value lies with what it means for the current sample. On the Criteo validation split, the same numerical interval carries residual click evidence with opposite signs across categorical and numerical contexts, even after additive main effects are removed. Production pipelines compound this mismatch because externally normalized features require transformations and statistics to remain synchronized between training and serving. We introduce ScalarLens, a numerical embedding that preserves what a value is while adapting how it should be interpreted. A monotone local mesh constructs a stable coordinate from the focal scalar alone; bounded low-rank dynamics then produce a contextual response without moving that coordinate or replacing categorical tokens and the CTR backbone. In a 1,539-run primary evaluation covering 19 representations, three datasets, nine backbones, and three seeds, ScalarLens ranks first in 25 of 27 settings on original numerical scales and second in the remaining two. Matched ablations show that scale correction, additional local capacity, and generic conditioning do not reproduce the gain. A controlled study further recovers categorical, numerical, and mixed response mechanisms under context shift while the focal coordinate remains exactly invariant. A complete rerun under shared standardization retains significant advantages over DEER, DAES, and NaryDis, showing that the result is not explained by tolerance to raw scales alone. ScalarLens therefore recasts numerical embedding as a measurement problem: coordinates belong to values, while predictive responses belong to values in context.

Abstract: Search advertising connects user intent with commercial content and plays a critical role in platform monetization. Recent systems typically align pretrained generative models with a single business reward, such as eCPM, or use naive reward fusion for preliminary multi-objective alignment. However, an ideal search advertising system must jointly account for heterogeneous objectives, including relevance, click propensity, and commercial value, to balance user experience and business value while mitigating globally suboptimal performance caused by gradient competition. We propose UniPolicy, an objective-aware multi-policy alignment framework. UniPolicy combines objective-specific prefix tokens, sparse MoE-LoRA routing, and objective-specific residual FFNs to hierarchically decouple parameters within a shared backbone, providing differentiated parameter and policy-expression spaces for different business objectives. It further constructs pairwise preferences from multi-stage behavioral feedback, supplementing the relative preference information in exposed-but-unclicked samples and strengthening the relative advantage of clicked candidates in the generation distribution. At inference, UniPolicy supports parallel, business-customizable multi-policy beam search, flexibly allocating candidate quotas across objectives under a fixed retrieval budget. Large-scale offline experiments show that UniPolicy delivers balanced improvements across multiple metrics while preserving retrieval quality, outperforming single-objective reinforcement learning and naive reward-fusion baselines. In a 7-day online A/B test on a real search advertising system, UniPolicy improves CTR by 0.71%, RPS by 1.58%, and advertising revenue by 1.32%, while maintaining stable serving latency.

Abstract: We describe our 10th-place solution to the KDD Cup 2026 Tencent UniRec Challenge, industrial click-to-conversion (CVR) prediction over 34.82M records, and we ask which mechanisms actually move held-out AUC. Starting from the official PCVRHyFormer baseline, a 15-step single-variable chain raises test AUC from 0.813237 to 0.827816, and our final submission reaches 0.828535. A leave-one-out ablation from the full model attributes the gain: removing the dense-feature representation stack costs 0.0095 AUC and removing the orthogonalized optimizer costs 0.0028, while no sequence-modeling component (merged single-stream backbone, polarity channel, auxiliary head, per-token FFN) costs more than 0.0005, within or adjacent to a $\pm$0.0004 seed band. We also report a generalization hazard: the row-group train/validation split shares one time window, so validation AUC overstates the leaderboard by about 0.014; anti-memorization and high-cardinality-ID changes even invert sign against it, a divergence that traces to dump-to-dump distribution shift and survives a time-ordered re-split. Dense representation and optimization, not finer sequence modeling, drive CVR AUC at this scale, and verdicts must come from the held-out leaderboard.

Abstract: In online A/B tests for real-time bidding (RTB), control and treatment models are typically trained on a shared serving log that includes data generated by the counterpart model. This shared-log training biases each model's training data through two channels: the counterpart model may have selected a different ad from the ad-candidate pool (ad-ranking disagreement) and may have bid a different price (bid-pricing disagreement), potentially distorting the A/B test outcome. Log-splitting eliminates the bias but sacrifices training data; log-sharing retains all data but leaves the bias unaddressed. We formalize the Bid-Aware Filter Family (BAFF), a class of (k,l)-parameterized hard filters that controls tolerance to each channel independently, providing a structured search space between these two extremes. We further propose a three-stage online measurement protocol that enables evaluating data-sharing strategies by their deviation from an interference-free reference model in production. In offline simulation, a (k,l) sweep surfaces operating points with smaller deviation from the interference-free reference model than both log-sharing and log-splitting. In a live RTB deployment on a demand-side platform (DSP), filter-based variants preserve the reference model's business metrics (e.g., CPC, CTR) more closely than both baselines. The best operating point is setting-dependent, underscoring the practical value of the search space itself.