Loading
Life sciences · Preprint
arXiv · August 12, 2026
Early or partial results. Treat as a signal, not a conclusion.
FunnelCausalNet is a novel uplift estimator designed to jointly model conversion and conditional order value through a multiplicative funnel structure, validated on industrial coupon allocation logs and semi-synthetic benchmarks. The method shows modest improvements over baselines in controlled ablation (18–48% GMV error reduction) and descriptive frontier consistency across ROI anchors, but results are not presented as independently significant and the work lacks peer review.
Methodological study with semi-synthetic benchmark validation and observational industrial RCT log analysis. Semi-synthetic: Criteo-MT7 coupon dataset. Industrial: de-identified Hotel-Coupon RCT logs. Public benchmarks: sparse binary-spend datasets.. Intervention: FunnelCausalNet uplift estimator with funnel-aware joint conversion-revenue modeling and Lagrangian budgeted allocation. Compared with: Eleven baselines (named leading feature-interaction baseline highlighted); direct GMV regression in ablation; revenue-focused rankers and uplift-curve proxies on public benchmarks.
Controlled ablation reduces GMV effect error versus direct GMV regression by 18–48% across tested zero-inflation regimes On industrial Hotel-Coupon RCT logs with ~4.9 million hold-out exposure records per seed, FunnelCausalNet has best seed-averaged mean DeltaROI at all seven correlated anchors from 10% to 60% On semi-synthetic Criteo-MT7, mean AUUC_GMV within one seed standard deviation of leading feature-interaction baseline among eleven baselines
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
A novel methodological proposal with algorithmic validation on semi-synthetic and industrial observational data, but without peer review, no confirmatory RCT of the full method, and results presented as descriptive frontier consistency rather than independent statistical significance.
As stated by the source record.
Quoted from the source exactly as published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Coupon campaigns seek to lift both conversion and revenue, but gross merchandise value (GMV) follows a deterministic funnel from conversion to conditional order value and is zero-inflated and heavy-tailed. We propose FunnelCausalNet, an uplift estimator coupling a binary conversion head with a nonnegative conditional-value head through $μ_{\mathrm{gmv}}=μ_{\mathrm{conv}}μ_{\mathrm{val}}$. Under explicit RCT, support, rate-gap, and cross-head covariance-control assumptions, an idealized leading-order MSE comparison identifies a regime in which funnel composition can reduce pointwise variance; this is a heuristic, not a guarantee for the shared-representation neural model. The estimator is paired with marginal split-conformal CATE summaries, combined through a Bonferroni union as audit bands, and a Lagrangian budgeted allocator using RCT-anchored estimates for subsidy-aware ROI accounting. On semi-synthetic multi-tier Criteo-MT7, FunnelCausalNet's mean AUUC_GMV is within one seed standard deviation of the leading feature-interaction baseline among eleven baselines, while a controlled ablation reduces GMV effect error versus direct GMV regression by 18--48% across tested zero-inflation regimes. On de-identified industrial Hotel-Coupon RCT logs with about 4.9 million hold-out exposure records per seed, expected-outcome evaluation sweeps full LP frontiers; FunnelCausalNet has the best seed-averaged mean DeltaROI at all seven correlated anchors from 10% to 60%, which we treat as descriptive frontier consistency rather than independent significance. On sparse binary-spend public benchmarks, revenue-focused rankers can dominate uplift-curve proxies, defining an explicit regime boundary.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.