---
id: "evidence-productivity-benchmarks"
type: "evidence"
source_timestamps: ["¶2"]
tags: ["productivity", "benchmarks", "calibration"]
related: ["claim-productivity-boost", "concept-full-funnel-gen-ai", "concept-b2b-gen-ai", "contrarian-low-volume-ai"]
confidence: "medium"
sources: ["attention"]
sourceVaultSlug: "hbr-seg-attention"
originDay: 4
articleStem: "hbr-cl-90-genai-myths-sales-marketing"
sourceUrl: "https://hbr.org/2025/02/5-gen-ai-myths-holding-sales-and-marketing-teams-back"
sourceTitle: "5 Gen AI Myths Holding Sales and Marketing Teams Back"
---
# External benchmarks for the 15–20% productivity claim

## Evidence: External benchmarks for the 15–20% productivity claim

Calibrates [[claim-productivity-boost]] against outside research.

**Where the article aligns:**
- **Marketing 15%** aligns well with McKinsey's **5–15%** productivity estimate (share of total marketing spend).

**Where the article runs hot:**
- **Sales 20%** exceeds McKinsey's central **3–5% of sales spend** estimate — though other syntheses report up to **40%** in specific implementations, so 20% sits within the broader empirical range while being optimistic vs. the central case.

**Corroborating knowledge-work evidence:**
- **St. Louis Fed:** Gen AI users save ~**5.4% of work hours** (≈**33% higher productivity** in hours where Gen AI is used); ~1.1% aggregate productivity increase.
- **Wharton Budget Model:** labor-cost savings of ~**10–55%**, averaging ~**25%**.
- **Cross-border retail field experiment:** **0–16% sales uplift** from Gen AI workflows across millions of users, without changing prices — causal evidence complementing survey/case-study claims.

**Expert framing:** Distinguish **potential upper bounds** from **typical realized gains**. Treat **15–20% as a best-case/targeted ceiling**, contingent on data quality, process clarity, change management, and adoption. Full-funnel and B2B applicability themselves are strongly supported — see [[concept-full-funnel-gen-ai]], [[concept-b2b-gen-ai]], and [[contrarian-low-volume-ai]].