How to build an AI business case that your CFO will approve

Chief financial officers approve AI investments when you present baseline costs, quantify value in currency terms, show a clear payback period, and prove you can govern the risk. More than 35 per cent of fashion executives already use generative AI in daily operations and rank AI as the single biggest opportunity facing the industry, according to the McKinsey and Business of Fashion State of Fashion 2026 report. Yet 92 per cent of fashion executives intend to increase their AI investment while only 1 per cent report having mature deployments. The gap between ambition and execution narrows when the business case addresses the four elements CFOs need: measurable baselines, documented value drivers, realistic cost and payback models, and credible risk mitigation.

What baseline metrics should you measure before deploying AI?

Measure the current cost of the process you intend to improve for at least four to eight weeks before switching on any AI tool. Every use case needs a financial view alongside its operating metrics: the pre-AI baseline cost, the total cost of the AI solution including integration and training, the payback period, and a defined threshold that triggers a scaling decision. If the pilot targets demand forecasting, document current forecast accuracy, markdown rates, stockout frequency, and the labour hours required to generate each forecast. If the use case is product imagery, track current cost per image, lead time from brief to final asset, and the number of variants produced per style.

One mid-market brand's baseline showed 120 SKUs per season requiring 2.8 sampling rounds on average, at 800 dollars for the initial sample plus 650 dollars per revision, totalling approximately 236,400 dollars in sampling costs per season. That figure becomes the reference point against which AI-generated design iterations must compete. Without a documented baseline, you cannot prove return on investment, and without proof, CFOs will not scale the pilot.

Which AI use cases deliver measurable value in fashion?

Focus on use cases where fashion businesses have published documented impact. McKinsey estimates generative AI could add 150 billion to 275 billion dollars to apparel, fashion and luxury operating profits within three to five years. That sector-wide figure translates into specific use cases with verified results. Zalando scaled its AI-generated product content from near zero to 90 per cent in one year, boosting content output by 70 per cent due to shorter campaign lead times. Zalando's shift to generative AI for campaign imagery cut production lead time from weeks to days and reduced related costs by around 90 per cent, Reuters reported.

H&M reported savings of 75 to 80 per cent in design and visual content time in February 2026, training over 100 designers on AI-design tools within a quarter. Returns mitigation also shows documented impact. The National Retail Federation's 2025 Retail Returns Landscape put total US retail returns at 849.9 billion dollars, or 15.8 per cent of everything sold, with online purchases returning at 19.3 per cent. AI size recommendation tools can reduce that rate, though results vary widely by approach.

Inventory optimisation delivers margin impact at scale. Fashion companies that have scaled AI across at least two functions report 8 to 12 per cent higher EBIT margins compared to peers that have not adopted AI, according to McKinsey's State of Fashion 2025 report. Demand forecasting improvements of 20 to 35 percentage points and markdown rate reductions of 5 to 15 percentage points are typical, industry analyses show. Brands using AI variant testing in 2026 report 15 to 30 per cent higher sell-through on extended ranges, which translates directly into lower carrying costs.

How should you model AI costs and payback periods?

Structure the investment in three tiers: pilot, multi-use case deployment, and enterprise scale. A focused pilot testing a single use case such as AI photography or design concepting typically costs 50,000 to 200,000 dollars and can be launched in one to three months. That pilot should run for a minimum of 90 days with measurable KPIs tied to the baseline you established. If the pilot proves a positive return, the next tier expands to two or three additional use cases at an investment of 500,000 to 3 million dollars, with expected ROI of 200 to 500 per cent across combined applications.

Fashion brands deploying AI-powered operations platforms report demand forecasting accuracy improvements of 20 to 35 percentage points, reduction in markdown rates of 5 to 15 percentage points, pattern development cycle time reductions of 40 to 60 per cent, and working capital improvements of 8 to 14 per cent, with enterprise deployments typically achieving full payback within 18 to 30 months of go-live. For smaller direct-to-consumer brands, the entry point sits lower. The total technology investment for a comprehensive AI transformation of a 20 million euro fashion business is typically in the range of 100,000 to 300,000 euros, with payback periods of 12 to 24 months based on documented results.

Include ongoing operational costs. Annual seat licences for fashion-specific AI platforms range from approximately 24,000 euros per seat for full deployment suites, with proof-of-concept fees adding another 12,000 to 15,000 euros upfront. Generic AI image generators cost 120 to 720 dollars annually, while enterprise platforms can run 6,000 to 60,000 dollars or more per year depending on scale and customisation.

What value drivers should the business case quantify?

Quantify value across three dimensions: time savings, cost reduction, and revenue or margin lift. Time savings appear first but matter least unless you can show how freed-up capacity converts into commercial output. A working pattern, bill of materials and measurement chart that used to take a freelance technical designer 6 to 10 hours now takes 20 to 45 minutes inside an AI-native tool, saving 300 to 700 dollars per style at freelance market rates of 55 to 90 dollars per hour. Multiply that by 30 styles per season and the saving reaches 19,500 to 48,000 dollars in cash that does not leave the business.

Cost reduction must account for both direct and indirect savings. Virtual fit and fabric simulation removes one to two physical sample rounds for the majority of garments that are not fit-critical, with each round saved worth 150 to 400 dollars in sample sewing, fabric and freight. Operators who switched to AI-native tech packs in 2025 reported 4 to 9 per cent lower per-unit production quotes on identical garments, with no change in factory or fabric. That is pure margin extracted from reduced friction and clearer specifications.

Revenue and margin lift come from higher sell-through, lower markdowns, and increased average order value. Personalisation engines deliver measurable uplift when properly implemented, though the range is wide. Cross-industry benchmarks suggest personalisation can lift revenue by 5 to 15 per cent and marketing ROI by 10 to 30 per cent. In fashion specifically, AI recommendations reportedly drive significant portions of online sales at retailers that have scaled the capability. Build the business case around the conservative end of verified ranges, not the headline figures vendors cite.

What governance and risk mitigation must you include?

Governance is no longer optional. Enforcement for high-risk AI systems under the EU AI Act begins 2 August 2026. High-risk systems require risk management frameworks, data governance, human oversight, transparency and conformity assessments. Most fashion AI applications do not qualify as high-risk under the Act, but dynamic pricing, certain recommendation engines and automated hiring tools may. AI governance in retail typically costs 10,000 to 15,000 euros for initial framework setup and 3,000 to 8,000 euros per month for ongoing monitoring and compliance.

The business case must include four governance components: an inventory of all AI systems in use or development, a risk classification against applicable regulations, defined roles and accountability for AI decisions, and monitoring procedures that track model performance and detect bias or drift. Data governance underpins everything. Without clean, representative training data and clear provenance, AI systems fail compliance audits and produce biased outputs. Budget for data quality work as part of the total cost.

Address intellectual property and model consent explicitly. Fashion brands using AI-generated imagery from real models must document consent, usage rights and compensation. Inditex stated it works collaboratively with models, agreeing any aspect on a mutual basis and compensating in line with industry best practice. That standard should appear in your governance framework and in the cost model, because proper consent and compensation carry a price.

How do you present the ROI calculation to the CFO?

Present ROI as a range with clear assumptions, not a single percentage. Structure the presentation around a three-year view: year one focuses on pilot and proof of value, year two on scaling to multiple use cases, year three on enterprise integration and sustained margin improvement. Show the baseline cost, the total investment including platform fees, integration, training and governance, the annual benefit broken into time savings, cost reduction and revenue lift, and the cumulative cash flow with payback period clearly marked.

Use a table to summarise the financial model. Include investment by phase, expected annual benefit, payback period, and three-year cumulative return. Add a sensitivity analysis showing how ROI changes if adoption is slower, if benefits realise at 75 per cent or 50 per cent of the base case, or if costs run over. CFOs trust models that acknowledge uncertainty more than models that promise certainty.

Define the decision gates. The business case should state explicitly what metric must be achieved by what date to justify moving from pilot to multi-use case, and from multi-use case to enterprise deployment. If the pilot does not achieve a minimum 150 per cent ROI within 12 months or does not demonstrate measurable improvement on the baseline KPIs, stop and reassess. That discipline protects capital and builds credibility.

Finally, assign ownership. The business case must name the executive sponsor, the project lead, the data owner, and the person accountable for governance and compliance. While immense amounts of money are being spent on AI pilots, very few of them make it to enterprise-grade adoption that delivers return on investment and value at scale, McKinsey's State of AI research shows. Accountability turns pilots into platforms.