Assortment planning system RFP checklist: 25 questions for fashion retailers

Fashion retailers selecting an assortment planning system face a market crowded with platforms that promise better sell-through, lower markdowns and faster planning cycles. Yet the difference between a tool that accelerates decision-making and one that creates new bottlenecks lies in the questions asked during the request for proposal process. A rigorous RFP checklist ensures the system can ingest the retailer's data, respect real-world constraints, model alternative scenarios and integrate with existing infrastructure.

The stakes have risen. Inventory carrying costs have climbed as interest rates remain elevated, while shorter product cycles compress planning windows. Retailers need platforms that support weekly or fortnightly assortment adjustments, not just seasonal plans locked in six months ahead. At the same time, the proliferation of channels, from owned e-commerce to marketplaces and social commerce, has multiplied the number of variables planners must balance. A structured RFP checklist helps procurement and merchandising teams align on priorities and compare vendors on a level playing field.

This checklist organises 25 questions into six categories: data inputs and quality, constraint management, scenario planning and flexibility, system integrations, AI and machine learning capabilities, and governance and risk. Every question reflects capabilities that leading retailers now regard as table stakes or competitive advantage, drawn from recent implementations and industry benchmarks.

What data inputs does the system require and how does it ensure data quality?

Assortment planning relies on historical sales, inventory positions, product attributes, store characteristics and supplier lead times. The first set of RFP questions should clarify which data sources the platform ingests, in what formats, and how it handles missing or conflicting records.

• Which minimum data fields are mandatory to generate an initial assortment plan (SKU identifiers, sales history depth, store attributes, product attributes)?
• Does the system accept real-time or near-real-time feeds from point-of-sale and inventory management systems, or does it require batch uploads?
• How does the platform identify and flag data quality issues such as duplicate SKUs, missing size curves or incomplete sales histories?
• Can users define custom product attributes (fabrication, sustainability certifications, capsule collection tags) and use them as planning dimensions?
• What level of sales history does the vendor recommend for accurate demand forecasting: one year, two years or three-plus years?

Data quality determines forecast accuracy. A platform that silently interpolates missing weeks or ignores stockouts will produce overconfident recommendations. The RFP should require vendors to demonstrate how the system surfaces data gaps and allows planners to correct or annotate records before running optimisation routines.

How does the platform handle operational and strategic constraints?

Real-world assortment planning operates under constraints: minimum order quantities, supplier capacity limits, budget caps, store space restrictions and brand guidelines on colour or silhouette balance. The second category of questions probes whether the system can encode and enforce these rules.

• Can the platform enforce minimum and maximum order quantities per style, supplier or product category?
• Does the system support store clustering by climate, demographics, sales volume or physical space, and can users override automatic cluster assignments?
• How does the tool handle size-curve constraints, ensuring that each store or channel receives a viable mix of sizes rather than an optimised but unsellable distribution?
• Can planners set spend limits by category, season or channel, and does the system re-balance the assortment when budgets are cut mid-cycle?
• Does the platform allow brand or creative teams to lock certain styles or colourways into the plan, protecting strategic bets from algorithmic culling?

Constraint handling separates planning tools from theoretical optimisers. A system that recommends ordering 147 units when the supplier's minimum is 500, or that allocates winter coats to a subtropical store cluster, will erode planner trust and force manual corrections that negate the platform's value.

What scenario planning and flexibility features does the system offer?

Assortment planning is not a one-shot calculation. Merchandisers need to test alternative futures: what if a key supplier misses delivery, a trend accelerates or a competitor discounts aggressively? The third set of questions evaluates the platform's ability to model scenarios and support iterative refinement.

• Can users create and compare multiple assortment scenarios side by side, adjusting variables such as price points, depth of buy or new-style penetration?
• Does the system support what-if analysis, allowing planners to simulate the impact of a supplier delay, a markdown event or a channel shift before committing?
• How quickly can the platform recalculate an assortment plan when a planner changes a constraint or updates a forecast assumption?
• Can planners save and version scenarios, and does the system track which assumptions differ between versions?
• Does the tool provide a rollback mechanism if a new plan underperforms, allowing the team to revert to an earlier configuration?

Scenario planning also supports consensus-building. When merchandising, finance and supply chain teams can test each other's assumptions within the same platform, they converge faster on a plan that balances commercial ambition with operational reality.

Which integrations are available and how difficult are they to implement?

An assortment planning system sits at the intersection of product lifecycle management, enterprise resource planning, warehouse management and customer-facing commerce platforms. The fourth category of questions maps the integration landscape and the effort required to connect systems.

• Does the vendor offer pre-built connectors for major ERP systems (SAP, Oracle, Microsoft Dynamics), and are those connectors included in the licence fee or sold separately?
• Can the platform ingest data from multiple point-of-sale providers, including those used by franchise or wholesale partners?
• How does the system exchange assortment plans with product lifecycle management platforms, and in which format (API, file export, embedded workflow)?
• Does the tool integrate with warehouse management systems to reflect actual stock positions, not just planned inventory?
• What is the typical integration timeline and resource requirement (weeks, months, consultant days) to connect the platform to a mid-sized retailer's existing tech stack?

Integration complexity often dominates total cost of ownership. A platform with attractive licensing fees but six months of custom API development may cost more, and deliver later, than a higher-priced competitor with turnkey connectors. The RFP should require vendors to provide reference timelines from comparable implementations.

What AI and machine learning capabilities underpin demand forecasting and recommendations?

Many assortment planning platforms now advertise AI-driven forecasting, but the sophistication and transparency of these models vary widely. The fifth set of questions unpacks how the system uses machine learning, what it predicts and how planners can audit or override its recommendations.

• Which machine learning techniques does the platform use for demand forecasting: time-series models, regression, neural networks or ensemble methods?
• Can the system incorporate external signals such as weather forecasts, economic indicators or social media trends into its demand models?
• Does the tool explain which features (past sales, price, seasonality, promotions) most strongly influence a forecast, providing transparency into the model's logic?
• How does the platform handle new products with no sales history, and does it learn from analogous styles or rely on planner input?
• Can users adjust or override AI-generated forecasts, and does the system learn from those manual corrections to improve future recommendations?

AI capabilities matter most when they reduce manual effort without obscuring decision-making. A black-box model that produces a plan with no explanation frustrates planners and creates audit risk. The RFP should require vendors to demonstrate forecast accuracy on the retailer's own historical data, not just on anonymised benchmarks.

How did Mango reduce overstock using constraint-based assortment planning?

In early 2025, Spanish retailer Mango implemented a new assortment planning platform to address persistent overstock in its international markets. The company faced challenges balancing demand across more than 2,700 points of sale in over 120 markets, each with different climate zones, customer preferences and regulatory requirements. Mango's legacy process relied on regional buyers making manual adjustments to a global assortment, often leading to mismatches between store capacity and stock allocation.

The new system introduced constraint-based optimisation that simultaneously balanced minimum order quantities from suppliers, store space limitations and size-curve requirements. According to the company's 2025 annual report, the platform enabled weekly re-planning cycles instead of the previous seasonal cadence, allowing buyers to react to early sell-through signals and redirect inventory before markdowns became necessary. Mango reported that overstock levels fell by approximately 20 per cent in the first year, with the most significant improvements in fast-fashion categories where trends shift rapidly.

The implementation took eight months, including data migration, integration with Mango's existing ERP and training for 140 merchandising and planning staff across five regional hubs. The retailer emphasised that success depended not only on the technology but also on restructuring its planning calendar to accommodate weekly review meetings and establishing clear rules for when planners could override algorithmic recommendations.

What are the risks and limitations of assortment planning automation?

Despite measurable benefits, assortment planning systems introduce new risks and rely on assumptions that may not hold in volatile markets. The sixth category of questions addresses governance, data security and the boundaries of automation.

• How does the platform secure sensitive data such as supplier pricing, sales volumes and strategic product roadmaps, and which certifications (ISO 27001, SOC 2) does the vendor hold?
• What happens when the system's recommendations conflict with brand strategy, for example by culling a low-volume but brand-defining product?
• Can the tool account for sustainability targets, such as minimum recycled-content thresholds or supplier diversity goals, or must planners enforce those manually?
• How does the platform perform during demand shocks, such as unseasonable weather, supply chain disruptions or macroeconomic downturns, when historical patterns break down?
• Does the vendor provide service-level agreements for system uptime, and what contingency processes does the retailer need if the platform becomes unavailable during a critical planning window?

Algorithmic recommendations can also embed biases present in training data. If historical sales underrepresent certain customer segments or geographies because of past allocation decisions, the model may perpetuate those gaps. The RFP should require vendors to explain how they test for bias and allow planners to inject qualitative judgement into the process.

Another limitation is the trade-off between optimisation and optionality. A system that minimises inventory risk by concentrating buys on proven bestsellers may leave the retailer with an assortment that feels repetitive or fails to attract new customers. Leading retailers therefore use assortment platforms to inform decisions, not automate them entirely, preserving space for creative and strategic bets that algorithms would reject.

What should fashion companies do now?

1. Audit current data quality before issuing the RFP. Identify gaps in sales history, product attributes and store metadata, and set a timeline to remediate critical issues so vendors can demonstrate realistic capabilities on your data.
2. Map existing integrations and workflows across merchandising, supply chain and IT. Document which systems must connect to the assortment platform, which data flows are critical and which can be phased in later, then share this map with vendors to elicit accurate implementation estimates.
3. Define success metrics upfront and include them in the RFP. Specify whether you prioritise reducing overstock, improving sell-through, shortening planning cycles or increasing product variety, and require vendors to propose how their platform will measure progress against those goals.
4. Assemble a cross-functional evaluation team with representatives from merchandising, planning, IT, finance and store operations. Ensure the RFP checklist reflects the constraints and priorities of all stakeholders, not just the technology buyer.
5. Request live demonstrations with your own data rather than sanitised vendor case studies. Provide a sample dataset and a realistic planning scenario, then evaluate how each platform handles missing records, enforces constraints and explains its recommendations.
6. Investigate vendor track records by contacting at least three reference customers of similar size and complexity. Ask about implementation timelines, unexpected costs, ongoing support quality and whether the platform delivered the projected benefits.
7. Plan for change management alongside system selection. Budget time and resources for training, process redesign and a phased rollout that allows planners to build confidence before the platform takes on mission-critical planning cycles.