How PVH uses AI and data platforms across Calvin Klein and Tommy Hilfiger

PVH Corp, the New York listed owner of Calvin Klein and Tommy Hilfiger, has spent the past two years building an enterprise data architecture that connects product development, merchandising, wholesale and retail channels on a single cloud platform. The initiative, disclosed in regulatory filings and investor presentations through 2025 and 2026, is designed to speed up decision making, improve forecast accuracy and reduce excess inventory across two global brands that together generated more than 8 billion US dollars in revenue in 2023.

The stakes are high. PVH has faced margin pressure from promotional activity and elevated inventory levels, prompting a multi year restructuring dubbed PVH Plus that targets 200 million US dollars in annualised cost savings by the end of 2025. Technology and data analytics are central to that plan, with the company committing 50 million US dollars to capability building in 2024 alone. Early results show improved sell through rates and fewer markdowns in categories where AI driven demand forecasting has been deployed, according to management commentary in earnings calls.

The PVH case offers a detailed look at how a legacy apparel group is attempting to modernise its technology stack under competitive pressure, and what obstacles remain. For fashion executives weighing similar investments, the experience highlights both the operational gains available and the organisational friction that can slow returns.

What is PVH's enterprise data strategy?

PVH began consolidating disparate data sources onto a cloud based platform in 2023, a move described in the company's 2024 annual report as part of its digital transformation roadmap. The platform ingests product master data, point of sale feeds, e commerce transactions, wholesale order books and supply chain events, creating what the company calls a single source of truth for inventory, demand and product performance.

In its fourth quarter 2024 earnings call in March 2025, chief executive Stefan Larsson said the data foundation was enabling faster insights across geographies and channels. The company had previously relied on regional systems and manual reporting, which meant decisions on replenishment, markdowns and allocation often lagged by weeks. The new architecture aims to surface anomalies and opportunities in near real time, allowing regional teams to adjust assortments and pricing before inventory problems compound.

PVH has not disclosed the specific cloud vendor or integration partners involved, but regulatory filings note investments in software as a service subscriptions and external consulting for data engineering and analytics. The platform was rolled out first in North America and Europe for Tommy Hilfiger, then extended to Calvin Klein in select markets during 2025, with further international expansion planned through 2026.

How is PVH applying AI to demand forecasting?

Demand forecasting is one of the earliest use cases for machine learning at PVH. The company has deployed models that analyse historical sales, promotional activity, weather patterns, macroeconomic indicators and social media sentiment to predict demand at the style and colour level. These forecasts feed into assortment planning and help buyers decide how much inventory to commit to wholesale partners and owned retail.

In the second quarter 2025 earnings call in September, management reported that AI enhanced forecasting had contributed to a reduction in markdown rates for Tommy Hilfiger North America compared with the prior year period. The company attributed the improvement to better initial buys and faster in season adjustments, though it did not quantify the markdown reduction in basis points.

PVH's approach combines demand signals from owned channels, where it has granular transaction data, with aggregated sell through reports from wholesale partners. The models are retrained quarterly to incorporate fresh data and are tuned separately for each brand and region, reflecting differences in seasonality, promotional calendars and customer behaviour. The company has said that forecast accuracy, measured as the percentage of styles within 10 per cent of actual sell through, has improved, but has not published specific accuracy metrics.

Which other functions are using AI and analytics?

Beyond forecasting, PVH has piloted AI in product design, pricing optimisation and supplier collaboration. In design, the company has experimented with generative tools that suggest colour palettes, print variations and silhouette adjustments based on trending keywords and past bestsellers. These tools are used by design teams as inspiration rather than autonomous creation, and the company has emphasised that human designers remain central to the creative process.

Pricing and promotion are another focus. PVH has tested dynamic pricing algorithms for e commerce that adjust list prices and discount levels in response to inventory position, competitor pricing and conversion rates. The algorithms recommend price changes to merchandisers, who retain final approval. In wholesale, the company is using analytics to identify which retail partners have the strongest sell through for specific categories, informing allocation decisions and co marketing investments.

Supply chain visibility has also improved through data integration. PVH now tracks shipments, production milestones and quality inspections on a unified dashboard shared with strategic suppliers. This visibility allows the company to spot delays earlier and reroute inventory to higher demand markets. In a May 2025 investor presentation, PVH said lead times for certain product categories had shortened by an average of two weeks, though it did not specify which categories or geographies.

What results has PVH reported so far?

PVH has linked its technology investments to tangible operational improvements, though it has been cautious about attributing margin gains solely to AI. In the full year 2024 results, announced in March 2025, the company reported that operating margin expanded by 80 basis points to 8.4 per cent, helped by lower markdowns and tighter inventory discipline. Management cited the data platform and enhanced analytics as contributing factors, alongside broader restructuring actions such as store closures and organisational simplification.

Inventory turns, a key efficiency metric, have also improved. At the end of the second quarter 2025, PVH reported inventory of 1.65 billion US dollars, down 7 per cent year on year, while revenue was roughly flat. The company said better demand forecasting and faster markdown decisions had reduced aged stock, freeing up working capital. The PVH Plus programme, which includes the technology investments, was on track to deliver its 200 million US dollar savings target by the end of 2025, according to the September 2025 earnings call.

E commerce performance has benefited from personalisation engines that recommend products based on browsing and purchase history. PVH said in its 2024 annual report that digital channels accounted for approximately 20 per cent of total revenue, and that average order values and conversion rates had both increased following the deployment of recommendation algorithms, though it did not provide specific percentage uplifts.

How does PVH's approach compare with other apparel groups?

PVH is not alone in pursuing enterprise AI, but the scale and pace of adoption vary widely across the apparel sector. Inditex, the Spanish owner of Zara, has long invested in data analytics and automated replenishment, using real time sales data to guide production decisions and in season reorders. Nike has built a direct to consumer technology stack that includes personalisation, demand sensing and supply chain optimisation. Fast fashion players such as Shein have used algorithmic design and on demand manufacturing to shrink lead times to weeks.

What distinguishes PVH's case is the legacy context. Unlike digitally native brands, PVH operates a hybrid model with substantial wholesale revenue, multiple brands, and entrenched regional systems. The company has had to retrofit data integration and AI onto an existing technology landscape, a process that is slower and more complex than building greenfield. Industry observers note that many traditional apparel groups face similar challenges, making PVH's progress a useful benchmark for peers.

A 2025 report by McKinsey on AI in fashion found that leading apparel companies are achieving 10 to 20 per cent improvements in forecast accuracy and 5 to 10 per cent reductions in excess inventory through machine learning, but that adoption remains uneven. The report emphasised that technology alone is insufficient and that success depends on data quality, cross functional collaboration and willingness to act on algorithmic recommendations even when they conflict with intuition.

What are the risks and limitations?

Despite the progress, PVH's AI journey faces several headwinds. Data quality remains a persistent challenge. The company sources information from dozens of wholesale partners, each with different reporting standards and latencies. Incomplete or inaccurate data can degrade model performance, and PVH has acknowledged in filings that data governance and master data management are ongoing priorities. Building trust in AI recommendations is another hurdle. Merchandisers and regional leaders are accustomed to relying on experience and intuition, and overriding an algorithm that contradicts their judgement can be difficult, even when the model has a better track record.

There are also questions about return on investment. PVH invested 50 million US dollars in technology in 2024, and while the company attributes part of its margin improvement to these capabilities, isolating the impact of AI from other restructuring actions is difficult. Some industry analysts have cautioned that apparel companies can overestimate the near term benefits of AI, particularly if they do not account for the cost of change management, training and system maintenance.

Regulatory and ethical considerations are emerging as well. AI driven pricing and personalisation raise questions about fairness and transparency, especially if algorithms inadvertently discriminate or create perceptions of price manipulation. PVH has not disclosed specific policies on algorithmic accountability, but the broader industry is beginning to grapple with these issues. In the European Union, the AI Act, which came into force in stages from 2024, imposes transparency and risk management requirements on high risk AI systems, including some used in consumer facing applications.

Finally, there is the risk of commoditisation. As AI tools become more accessible, the competitive advantage they confer may erode. If every apparel company achieves similar improvements in forecast accuracy and inventory efficiency, the gains will be competed away through lower prices or higher promotional spending. PVH's long term differentiation will depend not just on adopting AI, but on embedding it into distinctive capabilities such as brand building, product innovation and customer experience.

How are wholesale partners responding?

PVH's wholesale business, which accounted for roughly half of total revenue in 2024, depends on collaboration with department stores, specialty retailers and e commerce platforms. The company has sought to bring key partners into its data ecosystem, sharing sell through data and jointly planning assortments. Some partners have welcomed the transparency, seeing it as a way to reduce markdowns and improve full price sell through. Others have been more cautious, concerned about sharing proprietary sales data or ceding merchandising authority to a supplier.

In practice, the level of integration varies. Large strategic accounts such as Macy's and Nordstrom in North America have established data sharing agreements and use PVH's analytics to guide their buys. Smaller independents and international distributors often lack the systems or bandwidth to participate fully. PVH has responded by offering tiered levels of collaboration, from basic sales reports to advanced joint business planning supported by AI forecasts.

The shift towards data driven wholesale relationships mirrors broader trends in the industry. Suppliers and retailers are moving away from transactional interactions and towards partnerships built on shared visibility and mutual performance metrics. For PVH, this means not only providing better products and service, but also becoming a source of market intelligence and planning support for its wholesale customers.

What should fashion companies do now?

1. Audit data quality first Before deploying AI, ensure product, sales and inventory data are accurate, consistent and accessible across systems. Poor data quality will undermine even the best algorithms.
2. Start with high impact use cases Focus early AI efforts on areas with clear business value and measurable outcomes, such as demand forecasting, markdown optimisation or supplier lead time reduction, rather than trying to transform everything at once.
3. Build cross functional teams AI initiatives succeed when merchants, planners, supply chain leaders and IT work together. Create small pilot teams with joint accountability for both technical deployment and business results.
4. Secure executive sponsorship AI adoption requires sustained investment and organisational change. Ensure senior leadership understands the timeline, costs and trade offs, and is prepared to champion the programme through setbacks.
5. Partner with wholesale accounts For brands with significant wholesale revenue, collaborate with key retail partners on data sharing and joint planning. Shared visibility can reduce markdowns and strengthen commercial relationships.
6. Plan for change management Train merchandisers, planners and regional teams to interpret AI recommendations and integrate them into their workflows. Expect resistance and allow time for trust to build.
7. Monitor algorithmic accountability Establish policies for transparency, fairness and oversight of AI systems, particularly those affecting pricing, personalisation or hiring. Regulatory scrutiny of AI in consumer industries is increasing.