---
title: "Why the same AI adoption leads to different outcomes — the difference between redesign and adding tools"
url: https://blog.tyrano.dev/en/why-the-same-ai-adoption-leads-to-different-outcomes-the-difference-between-redesign-and-adding-tools
lang: en
site: Tech AI News - 테카이
category: "AI Case Studies"
tags: ["ai","artificial intelligence","ai adoption","workflow redesign","enterprise ai strategy","mckinsey","bcg"]
published: 2026-10-07T12:21:37.377Z
updated: 2026-10-07T12:21:37.412Z
sources:
  - https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-how-organizations-are-rewiring-to-capture-value
  - https://www.bcg.com/publications/2025/to-unlock-the-full-value-of-ai-invest-in-your-people
  - https://www.deloitte.com/kr/ko/issues/generative-ai/state-of-ai-in-enterprise.html
---

# Why the same AI adoption leads to different outcomes — the difference between redesign and adding tools

> Even after bringing in the same generative AI, some organizations create meaningful changes in profit and loss, while others end up only increasing costs. The fork in the road isn’t which model you used, but whether you rewired the work procedures themselves before layering on AI.

---

## At a glance

| 항목 | 내용 |
|------|------|
| 조사 범위 | [맥킨지](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-how-organizations-are-rewiring-to-capture-value) State of AI 2025, [BCG](https://www.bcg.com/publications/2025/to-unlock-the-full-value-of-ai-invest-in-your-people) AI investment research (November 2025), [딜로이트](https://www.deloitte.com/kr/ko/issues/generative-ai/state-of-ai-in-enterprise.html) global executive survey (surveyed Aug–Sep 2025, published 2026) cross-referenced |
| 사용률 vs 손익 임팩트 | Regular AI users 88%, genAI adoption 72% — yet only 39% report P&L (EBIT) impact; high performers about 6% (McKinsey) |
| 재설계 여부별 격차 | High performers that actually redesigned workflows 55% vs other companies 20% — 2.8× gap (McKinsey) |
| 프로세스 변화 수준 | Fundamental business model shifts 34% · Core process redesign 30% · Largely unchanged legacy procedures 37% (Deloitte) |

## Background — why do outcomes diverge with the same AI?

Generative AI adoption has already reached a high level. According to McKinsey’s State of AI 2025 (published March 2025), 88% of surveyed companies use AI regularly in at least one business area, and genAI adoption is 72%, more than doubling from the prior year (33%).

The issue comes next. In the same survey, only 39% said they saw a meaningful impact on profit and loss, and only about 6% were classified as “AI high performers.” Many studies highlight this gap between adoption and outcomes, but McKinsey zeroed in on “why only some see impact.” Among 25 organizational attributes, the single factor most strongly correlated with P&L impact was **workflow redesign**.

## The structural fork that separates outcomes

### 1) Whether workflows were redesigned — the strongest single variable

In the McKinsey survey, the share that actually rewired workflows was 55% among high performers versus 20% among the rest, a 2.8× difference. Here, “redesign” doesn’t mean adding an AI feature alongside existing work; it means re-architecting the decision structure itself—approval chains, sign-offs, role splits—on the premise of AI use.

> Analysis: These numbers don’t say “redesign causes performance,” but rather “among companies that achieved impact, the share that redesigned is much higher.” There’s no basis yet to flip causality, but the fact that redesign ranked first among 25 factors in the same study shouldn’t be taken lightly.

### 2) Executive involvement — setting goals beyond efficiency

In the McKinsey survey, 80% of respondents cited “efficiency” as the goal for AI adoption. High performers differed in that they also targeted revenue growth and innovation. If goals are narrowly defined as cost reduction, AI remains a tool that just speeds up existing tasks; organizations that include revenue and innovation have motivation to change the work procedures themselves. Such goal-setting and governance typically sit at the executive, not line, level.

### 3) Where budget and effort actually go

Deloitte surveyed 3,235 executives across 24 countries in Aug–Sep 2025 and published the results in 2026, grouping companies into three types: 34% fundamentally transformed their business models/processes; 30% are redesigning core processes around AI; 37% are using AI superficially while leaving existing processes largely intact. The third group commonly spends on tools—purchases and subscriptions—but allocates little to labor-intensive areas like process redesign, governance, and training.

BCG’s November 2025 analysis shows a similar distribution. Only about 5% created value at scale, and nearly 60% have seen little to no effect so far. In one example BCG cites, a European retail bank redesigned specific work around AI and lifted productivity by over 50%. The common thread was not “tool swapping,” but “changing the work design.”

## Add-on tools vs redesign: what is actually different?

| 항목 | 도구 추가형 | 재설계형 |
|------|------|------|
| AI의 위치 | Added as a separate tool alongside existing procedures | Decision points and approval flows are rearranged around AI |
| 목표 | Centered on efficiency and cost reduction | Efficiency plus revenue and innovation goals in parallel |
| 예산 방향 | Licenses and subscriptions | Process redesign + governance + training |
| 책임 소재 | Delegated to line functions | Executives directly own goals and governance |
| 전형적 결과 | Individual work speeds up but doesn’t show up in enterprise P&L | Significantly higher probability of translating into P&L impact |

## Limits and open questions

Most figures in this domain are based on self-reported enterprise surveys, and definitions of “redesign” vary by study. It’s unclear whether McKinsey’s “workflow redesign” matches Deloitte’s “core process redesign,” and there’s still insufficient evidence to assert causality from correlation. Few cases disclose concrete numbers like BCG’s European retail bank, so it’s hard to treat “redesign lifts productivity by 50%” as a general rule. Even so, multiple independent studies converging on the same conclusion—that whether you redesign separates outcomes—suggests structural consistency.

## Applying this to your organization

### Large enterprises
If you already have governance, first pivot it from “AI adoption approval” to “work procedure redesign approval.” Rather than department-by-department pilots, pick one critical decision point and redesign the procedure itself; this tends to offer better ROI.

### Small and midsize businesses
Your strength is scoping redesign narrowly. Instead of the whole business, pick one or two tasks with high repetition and simple approvals and rework those procedures; you’ll see results faster than by distributing tools across teams.

### Startups
Your small size makes procedures inherently flexible. But if you constrain goals to “efficiency,” you can fall into the same trap as large firms—tools get used, but impact is invisible in P&L. Linking AI use to revenue and product metrics from the outset is advantageous.