Quick Facts
- Only 25% of AI initiatives deliver expected ROI, according to IBM, while MIT found 95% of AI pilots produce zero measurable P&L impact.
- 88% of employees use AI daily, but only 5% use it in advanced ways, creating a hidden proficiency gap that undermines returns.
- Just 4% of retailers have embedded AI as a core business strategy, even as 70% say they are exploring it.
Brands are spending on artificial intelligence. They are not getting their money back. The gap between AI investment and AI returns has become one of the most expensive problems in retail and fashion, and it is not closing.
IBM’s CEO study found only 25% of AI initiatives deliver expected ROI. Separately, Business of Fashion reports that 56% of CEOs see zero significant financial benefit from AI spending. S&P Global found that 42% of companies abandoned most of their AI projects in 2025, more than double the prior year.
The numbers inside fashion are no better. Only 4% of retailers have made AI a core business strategy. Meanwhile, 35% of fashion executives report using generative AI for functions like customer service, image creation, or copywriting. Most remain stuck in pilot mode.
The Proficiency Problem
The core issue is not bad software. It is that most employees do not know how to use AI effectively.
EY’s 2025 Work Reimagined Survey of 15,000 employees across 29 countries found 88% use AI daily. Only 5% use it in advanced ways. OpenAI’s 2025 State of Enterprise AI report identified a fourfold productivity gap between power users and typical employees working with the same tools inside the same organization.
Meta quantified this gap directly. AI tools produced a 30% output increase for the average engineer. Power users saw an 80% improvement year over year. The difference is proficiency, not access.
High adoption rates mask low proficiency. A company where 90% of staff logs into an AI tool at Level 1 skill is not the same as one where 30% operates at Level 2 and 10% at Level 3, even if both report similar adoption figures. ROI depends on where people fall on the proficiency curve, not whether they have logged in.
Data and Measurement Gaps
Weak data is the second structural failure. Fashion supply chains run across multiple suppliers and borders. Data sources are fragmented and inconsistent. AI needs clean, connected data to produce useful outputs.
Between 2023 and 2025, many brands invested in AI design tools. Design teams generated thousands of concept images. When those concepts reached product development, the process stalled because the underlying data could not support execution.
How companies measure AI success compounds the problem. Half of companies track data quality improvements. Forty-eight percent measure employee productivity. Far fewer tie AI spending directly to margin impact or revenue. Operational efficiency is not a business outcome. A company can improve its processes while losing market share.
The Scale Barrier
McKinsey’s 2025 State of AI survey found 88% of organizations regularly use AI, but only 6% achieve enterprise-wide impact defined as a 5% or greater EBIT contribution. Up to 90% of AI initiatives never scale past the pilot stage.
The reason is not tool performance. Most pilots fail because the business case was never tied to specific, trackable metrics. Companies launch with enthusiasm. Reviews reveal ambiguity. Projects are canceled before scaling.
Ninety-four percent of CEOs and CHROs identify AI proficiency as their top in-demand skill for 2025. Yet only 35% of leaders say they have prepared employees effectively for AI roles. The top barriers to AI readiness are lack of talent at 46%, data privacy concerns at 43%, poor data quality at 40%, and high implementation costs at 40%.
What This Means for Operators
For DTC and retail operators, the data points to a consistent pattern. Buying AI tools is the easy part. Building the internal capability to use them at a high level is where most companies fall short.
Brands that want returns need to measure proficiency, not just adoption. They need clean data pipelines before they run AI on top of them. And they need to connect AI spending to specific P&L outcomes, not proxy metrics like efficiency scores.
The companies reporting real returns are not the ones with the most AI tools. They are the ones that treated workforce skill-building and data infrastructure as prerequisites, not afterthoughts.
