Expert Routing in Tenders: Assigning Requirements to the Right People

Introduction

A specification for a rail vehicle contains hundreds or thousands of individual requirements. Traction, braking, fire protection, air conditioning, passenger information, accessibility, maintenance: each discipline has its own standards, evidence obligations, and technical vocabulary. No single engineer can assess all of them. The bid manager therefore distributes the requirements among subject-matter experts from different departments and disciplines. This process is called expert routing.

What sounds straightforward is one of the biggest bottlenecks in bid preparation. In 2014, a manager at a braking-system manufacturer told CONTACT Software that the number of requirements requiring comments had increased at least tenfold over the preceding decade. Over the same period, tender-document volumes had grown from a CD to a DVD. The volume has continued to rise since then, along with the number of participating subject-matter experts.

The figures are similar across industries: according to the Loopio 2025 RFP Response Trends Report (1,500+ companies across all industries), 48% of proposal teams name collaboration with subject-matter experts as their biggest challenge, for the fifth year in a row. In the rail industry, where 10 to 30 subsystem teams typically work in parallel, the real coordination burden is likely even higher.

Why expert routing matters

The quality of a bid depends directly on whether every requirement is assessed by the right subject-matter expert. A fire-protection engineer can competently classify a requirement on EN 45545 but not a requirement on ETCS compatibility. Conversely, a signalling engineer cannot make a statement on structural strength under EN 12663. The INCOSE Requirements Working Group defines requirements allocation, the assignment of requirements to responsible system elements and people, as one of the core activities of requirements management, on a par with traceability and verification.

Routing errors have direct consequences:

• Incorrect classification: A non-expert marks a requirement as "OK" even though it can only be met subject to conditions (OKB). That becomes a change request during the project.
• Delays: A requirement reaches the wrong department, is forwarded, and then sits in another queue. Days are lost. According to Loopio, 33% of teams say faster subject-matter expert responses would be their most important lever for winning more tenders.
• Duplicated work: Without clear ownership, two departments process the same requirement independently, potentially reaching conflicting conclusions.
• Gaps: A requirement is not assigned to any expert and remains unprocessed. In the worst case, the response is missing from the specification.

Source: CONTACT Software: From slow train to requirements management

The LH structure and EN 15380: how specifications are organised

In the rail industry, specifications are typically divided into thematic sections covering different subsystems and trades. A common structure uses the so-called LH categories (LH1 to LH8), which the VDB Requirements Management Guide describes as a structuring aid.

In parallel, the EN 15380 series of standards defines a standardised classification system for rail vehicles. Part 2 (Product Groups) divides a vehicle into named product groups, Part 4 (Function Groups) assigns functions such as traction, braking, or door control, and Part 5 (System Breakdown Structure) defines the hierarchical breakdown into main systems and subsystems. EN 15380 is relevant to expert routing because it supplies a stable, manufacturer-independent taxonomy: if every requirement is assigned to an EN 15380 subsystem, the routing decision can be automated or at least standardised.

Typical requirement volume by LH category (relative illustration; varies by project and customer).

The LH structure provides the natural basis for expert routing: LH3 requirements go to traction engineering, LH6 to interior fit-out, and LH7 to the IT department. In practice, though, assignments are rarely as clear-cut as the structure suggests.

The routing workflow in practice

Whether a company works with Excel, an ALM system, or a specialist platform, the routing process follows a similar pattern:

RACI and skills matrices: systematising routing

With large specifications, assigning requirements ad hoc is not enough. Companies that have systematised the process use two instruments:

RACI matrix for routing: For each LH category, or each EN 15380 subsystem, the company defines who is Responsible, Accountable, Consulted, and Informed. A traction engineer, for example, is Responsible for LH3 requirements and Consulted on LH4 requirements that have a traction interface. The RACI matrix prevents cross-cutting requirements from disappearing during assignment because the consulting role is explicitly documented.

Project-specific skills matrix: An assignment table that documents which standards, subsystems, and technologies each available subject-matter expert can cover. If a specification contains hydrogen-traction requirements for the first time, the skills matrix immediately shows whether the expertise exists in-house or must be procured externally. Most companies build their skills matrix from previous projects and update it for each project.

Where friction arises

On paper, this workflow sounds linear. In reality, it looks different.

1. The distribution problem

With a specification containing 1,500 requirements, the initial assignment alone can take several working days when it is done manually. The bid manager must read, understand, and assign each requirement to the right expert. The decision is not trivial for cross-cutting requirements, especially when the bid manager does not know every discipline equally well. In the technology sector, according to Loopio 57% of teams even report difficulties with subject-matter expert coordination, the highest figure across all industries.

2. Expert availability

Subject-matter experts in the rail industry are not dedicated bid employees. They work in parallel on ongoing projects, development tasks, and other tenders. If a fire-protection expert is tied up in an approval project for two weeks, their requirements stand still, regardless of which routing system the company uses. According to its own supplier page Alstom coordinates around 19,000 suppliers in 77 countries. Each of those suppliers has capacity bottlenecks of its own, which affect the OEM's response times.

3. Missing visibility into processing status

In document-based workflows, such as Excel or Word sent by email, the bid manager has no real-time view of which requirements have already been processed, which are still open, and where bottlenecks are forming. Status checks by email or in meetings take time and often produce inaccurate results. According to Loopio, 39% of teams struggle to find current, accurate answers to technical questions. In the rail industry, where requirements often reference normative sources that change between revisions, this problem is particularly acute.

4. Consistency problems

When 20 experts process requirements independently, contradictions can arise: Department A classifies an interface as "OK", while Department B classifies the corresponding counter-requirement as "OKB". Such inconsistencies often become apparent only during the late consolidation phase, when there is little time left for corrections. The EuroSpec standard for requirements management addresses this problem through its six core areas: requirement characteristics, syntax, attributes, traceability, validation/verification, and data exchange. However, meeting those standards requires all participating experts to work in the same system.

Organisational models for routing

How companies organise expert routing depends on their size, organisational structure, and level of tool support. Three models are common in practice:

Most companies use a hybrid model: the bid manager handles the high-level pre-distribution at subsystem level, based on LH categories or EN 15380 assignments, while subsystem owners distribute the requirements in detail within their area. The bid manager identifies cross-cutting requirements and assigns them multiple times. For large consortium tenders, such as the EUR 15 billion contract for Berlin's S-Bahn (DB, Siemens, Stadler), requirements are also divided among consortium partners, adding another routing layer.

Traceability: what EN 50126 and ISO 22163 require

Expert routing is also a regulatory requirement. The CENELEC EN 50126 defines the RAMS lifecycle (Reliability, Availability, Maintainability, Safety) and calls for complete traceability within the V-model. Every requirement must be assignable to a verification activity, an owner, and supporting evidence. The specification and allocation of system requirements is a distinct phase in the V-model lifecycle, with defined roles for each lifecycle phase. Who assessed which requirement is part of the safety case, not an optional extra.

ISO 22163:2023 (IRIS Rev. 04) requires traceability from customer requirements through to design evidence in its sections on design and development. IRIS goes beyond ISO 9001 and adds rail-specific requirements for project management, First Article Inspection (FAI), RAMS evidence, and lifecycle costs. The standard applies to the whole supply chain: from the development company through the manufacturer and supplier to the maintenance operator. In a bid context, this means that assigning requirements to subject-matter experts and recording their assessments must be documented in a traceable way, not only for the internal project team but also for external auditors and the IRIS certification body.

Tools for expert routing

The tools range from Excel spreadsheets to ALM systems. What a company uses determines how well routing, status tracking, and traceability work day to day.

Excel with assignment columns

The simplest approach gives every requirement row an "Owner" column containing the name of the subject-matter expert. Filter views for each expert allow rudimentary work distribution. The drawbacks are no notifications, no status tracking, and no parallel editing, which leads to the familiar question: "Who has the current version?" The CONTACT Software article reported in 2014 that even Deutsche Bahn had only "recently" introduced a database-backed requirements-management system.

IBM DOORS / DOORS Next

The market-leading requirements-management system in the rail and aerospace industries. DOORS offers rule-based assignments, traceability, and change-impact analysis. A concrete example: Rail Projects Victoria (Melbourne) selected DOORS Next as a SaaS solution for the AUD 11 billion Metro Tunnel Project. The system created a central, collaborative environment in which requirements could be managed in real time across multiple internal and external stakeholders. This was particularly relevant for routing: integrating hazard logs and interface registers ensured that controls and requirements were correctly assigned to the right parties. RPV now applies the knowledge gained to other projects, including Regional Rail Revival and Melbourne Airport Rail.

Siemens Polarion

Polarion is Siemens' alternative with the same functional scope. Hitachi Rail has integrated Polarion into its project workflows to replace Excel and Word as requirement carriers. Since 2025, Polarion has also offered AI-assisted features for automatic requirement extraction from tender documents. According to Siemens, the AI analyses requirement objects by domain and assigns the responsible departments based on the domains it identifies, reducing manual effort by up to 70%.

Specialised platforms

Alongside the large ALM systems, specialised platforms focused on the bid process are increasingly emerging. Tendric uses rule-based routing with a lower entry barrier than DOORS or Polarion, while providing targeted support for routing, classification, and specification export. Cross-industry data shows that teams with a structured content library achieve a 66% reuse rate and are almost twice as likely to achieve high win rates. Specialist platforms address the point at which Excel no longer scales but a full ALM system would be excessive.

Can AI automate routing?

Automatically assigning requirements to subject-matter experts is an obvious application for AI-based systems. The idea is that when a system learns from thousands of historically assigned requirements which formulations and subjects belong to which department, it can propose an initial assignment for new specifications.

This works well for requirements with clear assignments: a requirement containing "EN 45545" or "fire-protection class" can reliably be assigned to the fire-protection team. Modern systems such as Polarion with an AI layer go further: they understand the semantics of requirements, not only keyword matching, compare new requirements with existing ones, and automatically detect changes between revisions. Accuracy drops markedly for cross-cutting requirements or unfamiliar wording, for example hydrogen traction or FRMCS as the successor to GSM-R.

According to Loopio 68% of proposal teams across industries already use generative AI, double the 34% recorded in 2023. Seventy percent of those using AI do so at least weekly. In the rail industry, this change is still at an early stage, but the first integrated solutions, Polarion AI, DRIM, and Tendric, show the direction of travel.

Case study: Melbourne Metro Tunnel Project

The Melbourne Metro Tunnel Project is one of Australia's largest infrastructure projects: a new underground rail corridor for more than 500,000 additional passengers per week, with a volume of AUD 11 billion. Rail Projects Victoria (RPV) had to coordinate thousands of requirements across multiple work packages and hundreds of engineers from different organisations.

RPV chose IBM Engineering Requirements Management DOORS Next as a SaaS solution. The result was a central, security-backed collaboration environment, a "single source of truth", in which requirements could be managed in real time and selectively shared with every organisation, aligned with its functional areas in the project. Integrating hazard logs and interface registers ensured that requirements were correctly assigned to the right parties.

Based on this experience, RPV is now developing a standardised requirements-management framework intended to be transferred to other projects, including Regional Rail Revival and Melbourne Airport Rail.

Conclusion

Expert routing is the invisible bottleneck in rail-industry bid preparation. The challenge lies not in technology alone but in the combination of volume, hundreds to thousands of requirements; complexity, dozens of disciplines and standards from TSI through CENELEC to ISO 22163; and time pressure, with submission deadlines of only a few weeks.

A good routing system, whether spreadsheet-based or tool-supported, must do four things: distribute work quickly on the basis of a stable taxonomy (LH categories and EN 15380), provide transparent real-time status tracking, document assignments traceably for EN 50126, and balance workload across all participating experts.

IBM DOORS and Siemens Polarion cover the enterprise market, specialised platforms such as Tendric serve mid-sized businesses, and AI-assisted solutions support initial allocation. More important than the choice of tool, however, is the discipline behind it: RACI matrices, skills assignments, and consistent follow-up. The biggest risk is not the wrong tool. It is requirements assigned to no expert, or requirements whose processing status no one knows.