From spreadsheet to platform: Why Excel is no longer enough for 1,000+ requirements
Everyone knows it: the Excel file with 47 tabs and 2,000 rows. An analysis of Excel’s seven limits in tendering and what specialised platforms do differently.
Introduction
Every bid manager in industry knows it: the Excel file with 47 tabs, 2,000 rows, and a filename like LH_v3_final_FINAL_revised_new(2).xlsx. It contains requirements, classifications, expert assignments, responses, and standards references. All in a tool built for spreadsheets.
In 2024, Poon et al. published a meta-study in Frontiers of Computer Science that evaluates 35 years of research on spreadsheet errors . The result: 94% of all business spreadsheets contain errors that can influence decision-making. In the rail industry, where a single overlooked standards reference can jeopardise vehicle approval, this error rate quickly becomes a tangible cost factor.
Database-based systems such as IBM DOORS or Siemens Polarion are widespread in the rail industry. But for operational tender processing, particularly classifying and responding to requirements, many teams still turn to Excel. The CONTACT Software Blog described the rail industry as a “slow train in requirements management” as early as 2014. Little has changed since then.
This article shows where Excel reaches its limits, what it really costs companies, what specialised platforms do differently, and when making the switch makes economic sense.
What is at stake: the market in numbers
To understand the significance of the tooling question, it is worth looking at the scale of the rolling stock market.
Alstom has an order backlog of EUR 100.3 billion. Siemens Mobility reached a backlog of EUR 52 billion at the end of fiscal year 2025. Stadler Rail reported CHF 29.2 billion at the end of 2024 (+20% year on year). In 2024, the EU invested EUR 7 billion in transport infrastructure through the Connecting Europe Facility , of which 80% is for rail.
Every one of these contracts began with a tender. Every tender consists of a specification with hundreds to thousands of requirements. And a considerable proportion of these requirements are processed in Excel even at companies with professional requirements engineering. The global rolling stock market reached a volume of EUR 65 billion in 2024 (+11% compared with 2022).
Why Excel is so popular
Excel dominates operational tender processing for good reasons. It is a sensible choice for small projects.
- It is installed on every computer, and everyone knows how to use it
- Columns, formulas, colours, filters: everything can be freely designed
- No rollout, training, or IT department required
- Already included in most Office licences
- Every new employee can get started immediately
Excel is entirely sufficient for a specification with 50 requirements and three people involved. Problems begin when projects grow. And in the rail industry, they almost always do. According to CONTACT Software, the specification documents for a single tender had already “grown from a CD to a DVD” by 2014. Their volume has continued to grow since then.
The 7 limitations of Excel for tenders
1. No real-time collaboration
As soon as more than one person works on a file, chaos begins. Who has the current version? What changes did the colleague make in their copy? The typical solution of distributing files by email or network drive creates version conflicts that must be resolved manually. For a specification being worked on simultaneously by traction technology, RAMS, braking systems, and ten other departments, this quickly becomes a full-time job.
Even SharePoint or OneDrive sharing only partially solves the problem: Excel files with thousands of rows and complex formulas regularly lead to save conflicts and data loss when edited simultaneously. An IDC study puts the time knowledge workers spend each week on content creation and management at 11.2 hours. Of that, 7 hours alone are spent on editing, review, and approval processes. That is $10,661 per employee per year.
2. No version control
When the client sends a new revision of the specification, changes must be identified manually. Row by row, across thousands of requirements. Classifications and responses already completed must be transferred without overwriting the new changes. The VDB requirements management guide therefore recommends using ReqIF, because automated change marking avoids time losses from extensive comparisons with previous working versions. In major regional passenger rail tenders, there are three to five revisions between the initial publication and contract award. Every one must be reconciled manually.
3. No permission management
In Excel, everyone sees everything. There is no way to make certain requirements visible only to particular teams. In practice, this means sensitive assessments, internal comments, and pricing information are unprotected in the same file. At the latest when external suppliers need to work on partial specifications, this becomes a security problem.
4. No traceability
Who changed which cell and when? In a regulated environment, complete traceability is not optional. The CENELEC standards EN 50126, EN 50128, and EN 50129 require complete traceability from requirements to evidence. ISO 22163:2023 (the rail industry quality standard, formerly IRIS) also requires documented processes. In 2023, the standard consolidated configuration management and change control in section 8.1.4. Excel offers no cell-level audit trail.
5. Error-prone in high-volume processing
Moving the wrong row, a copy error in a formula, a filter that was not reset. With thousands of rows, this happens regularly. Ray Panko, the founder of spreadsheet error research, documented cell error rates of 0.4% to 6.9% in field audits in his study “What We Know About Spreadsheet Errors” . It sounds small. But in an Excel file with 5,000 cells, a 2% error rate means 100 erroneous cells.
In regulated procurement procedures, discrepancies between compliance statements and the actual proposal can, according to ESA procurement guidelines , lead to downgrades or exclusion. The same principle applies in rail.
6. No standards comparison
When a requirement refers to TSI LOC&PAS, DIN EN 45545, and EN 50155, each of these standards references must be checked manually. Is the referenced version of the standard still current? Which sections are relevant? Excel has no way to automatically recognise, validate, or compare standards references against a standards database. The ERA alone defines 11 different TSIs for rolling stock. A change to one of them can affect dozens of requirements in the specification.
7. No knowledge transfer
When an experienced bid manager leaves the company, their Excel files, templates, and practical knowledge leave with them. There is no central knowledge base from which the next project can learn. According to Bidara , the cross-industry content reuse rate for proposal teams is 66%. But that only works when previous responses are searchable and structured. Not as xlsx files on personal network drives.
The McKinsey Global Institute puts the proportion of time knowledge workers spend searching for internal information at 19% of the working week. In bid departments, where this knowledge is scattered across dozens of Excel files, the proportion is likely to be even higher. The same study estimates that searchable knowledge systems can reduce search time by up to 35%.
Famous spreadsheet errors
The history of spreadsheet errors is long and expensive. Here are some of the best-known cases:
Documented losses caused by spreadsheet errors in USD millions (sources: Qashqade, Full Stack Modeller)
The JP Morgan case is the best known: a copy-and-paste error in a value-at-risk model caused the risk of a trading portfolio to be underestimated. Loss: USD 6 billion. At Fannie Mae , a logic error in a single Excel formula led to a USD 1.136 billion misstatement of equity. The share price fell by 6%.
And there is a case directly from the rail sector: in 2012, the UK Department for Transport awarded the West Coast Main Line franchise (worth GBP 5.5 billion) to FirstGroup, but had to cancel the entire procurement just weeks later because of spreadsheet errors in the financial model. Officials had miscalculated the required risk capital for each bidder. The cost to taxpayers: up to GBP 300 million for compensation and re-running the procedure.
In day-to-day tender processing, the errors are less spectacular: an incorrect standards reference in the compliance matrix, an overlooked change in binding force from “should” to “must,” a missing assignment in expert allocation. They do not make headlines, but such errors can cost a tender or delay approval.
Most business spreadsheets contain errors that can influence decision-making. Faulty spreadsheets lead to incorrect decisions, financial losses, and operational problems.
– Prof. Pak-Lok Poon, Central Queensland University (2024)Source: phys.org, Study finds 94% of business spreadsheets have critical errors (2024)
A concrete example from the rail sector: Network Rail replaced its paper- and spreadsheet-based safety processes with the digital RailHub platform. The result: 18% fewer safety-critical errors in Safe Work Packs and 43% fewer near misses. Engineers could process data 50% faster than before.
The hidden costs: what Excel really costs
Excel is free. The way it is used in bid processing is not.
According to McKinsey, knowledge workers spend 19% of the working week searching for information. IDC puts the time for editing and approval processes at 7 hours per week, or $10,661 per employee per year. For a 15-person bid team, that is more than $150,000 annually.
A ClusterSeven survey found that 58% of accountants rate the frequency of spreadsheet errors as ‘very high’ or ‘quite high.’ 72% rated ‘spreadsheet risk’ as a significant business risk. Siemens puts the savings potential of AI-supported bid processing at 21% lower error costs in the sales phase.
The replacement costs for a specialised knowledge worker are 150–200% of annual salary. It takes a successor 6–12 months to ramp up. When knowledge resides in personal Excel files rather than a searchable system, this onboarding time increases considerably.
Every hour a subject-matter expert spends searching for the current Excel version is missing from substantive work: requirements analysis, solution development, compliance review. McKinsey estimates that searchable knowledge systems can increase knowledge-worker productivity by 20–25%.
Sources: McKinsey Global Institute (2012), IDC (2012), ClusterSeven, Siemens/DRIMCO (2025)
What a specialised platform does differently
The difference between Excel and a specialised platform lies in the data model. Every requirement is a data record with relationships to classifications, sources, experts, and responses, rather than a row in a file.
Systems such as IBM DOORS and Siemens Polarion solve part of this problem for requirements engineering. What they often lack is the response process: classification, expert assignment, response generation, and specification export as a coherent workflow. Specialised solutions such as Tendric address precisely this gap.
The tooling ecosystem: what is available and what it can do
The range of tools for tender processing extends from general ALM systems to specialised solutions. The choice determines how much manual work is required at every step.
IBM DOORS: The de facto standard
IBM DOORS (Dynamic Object Oriented Requirements System) has been the standard in regulated industries since the 1990s. The system offers native support for baselines , meaning frozen snapshots of a requirements state. The baseline comparison function shows the exact difference between two states for every requirement. One concrete example: Rail Projects Victoria (Melbourne) selected DOORS Next as a SaaS solution for the Metro Tunnel Project.
Siemens Polarion: AI-supported bid processing
Siemens Polarion takes a document-centric approach with its LiveDoc functionality. Since 2025, Polarion has additionally supported AI-supported requirement extraction and bid processing. Integration with DRIMCO's RFQ/tender management enables faster decision cycles and is intended to reduce error costs in the sales phase by 21% while increasing EBITDA by 10%.
PTC Codebeamer: Product Line Engineering
PTC Codebeamer offers an approach tailored to Product Line Engineering with Streams, Baselines, and Delta Merge. This is particularly relevant for multi-product tenders, where the same specification must be answered for different vehicle variants. PTC announced new AI functionality for Codebeamer in 2026.
ReqIF: The exchange format
The ReqIF format (Requirements Interchange Format) enables the structured exchange of requirements between different tools. The ProSTEP iViP ReqIF Implementor Forum had conducted a total of six benchmarks by 2024; the most recent covered 56 system combinations and 2,800 evaluation criteria. Nevertheless, the benchmark found that a “lossless exchange of requirements is not always possible.” Vendor-specific extensions and version incompatibilities remain a problem.
Moving to a new system does not necessarily mean replacing existing systems such as DOORS or Polarion. Often the goal is to complement the requirements management system with the response and classification process. Tools such as Tendric are designed to work alongside existing systems. DOORS manages the requirements base well, but operational work (classification, expert assignment, specification export) still often takes place in Excel.
The economic impact
Bid processing for large rail projects ties up subject-matter experts from numerous departments, project managers, and bid managers for weeks. With win rates of 20–30% for complex tenders (CSK Management), two out of three bids do not win the contract. Every efficiency improvement lowers the cost per bid and can improve bid quality at the same time.
65% of proposal teams use specialised RFP software (up from 48% in 2024). 68% use generative AI, double the 34% in 2023.
– Loopio 2025 RFP Response Trends ReportSource: Loopio 2025 RFP Response Trends Report (cross-industry survey, 1,500+ teams)
The APMP survey among more than 1,750 members shows what manual processes mean for the people involved: 62% of proposal professionals work more than 40 hours per week on tenders, 88% report stress-related health problems. 77% say their current process is “not ideal.” Public tenders average 116 pages, and the resulting proposals average 144 pages.
Cross-industry tender processing statistics in percent. Sources: Loopio, Bidara, APMP
When making the switch makes sense
Not every company needs a platform immediately. But there are clear indicators that purely Excel-based work is reaching its limits. The Jama Software survey shows that almost one third of all development teams have no requirements management system and rely on email, documents, and shared spreadsheets.
- Your specifications regularly have 500+ requirements
- More than 10 people from different departments are involved in processing
- You process more than 5 tenders per year
- There are regular revision deliveries from the client
- You must document compliance with CENELEC standards or IRIS
- There have already been errors caused by version conflicts or incorrect assignments
- Subject-matter experts work on several tenders in parallel and lose track
- Experienced bid managers have left the company and taken their knowledge with them
The path to a platform: not all or nothing
The transition from Excel to a structured platform does not have to take place as a big-bang migration. In practice, a phased approach has proven effective:
Standardise the import of specifications, whether from ReqIF, Excel, or PDF. Each requirement receives a stable ID, an LH category, and a bindingness level. The existing DOORS or Polarion system provides the requirements base; the platform takes over the response process.
Move the core tender-processing workflow to the platform: Robel classification (OK/OKB/NOK/OKM/R), expert assignment, comments, and source references. This is where the greatest efficiency gain occurs, because coordination effort between departments falls dramatically.
Introduce automatic delta detection for revisions and a central knowledge base for previous responses. From this phase onward, every tender becomes faster than the previous one because the system learns from earlier projects.
Introduce automated export of specifications and partial specifications in the client format. Audit trails for EN 50126 and ISO 22163 are generated automatically. Compliance changes from a manual proof into a system capability.
According to industry analyses, more than 185 rail operators worldwide were engaged in digital modernisation programmes in 2024, an increase of 38% compared with 2020.
Conclusion
Excel is a good tool for spreadsheets. For processing complex tenders with thousands of requirements, it lacks versioning, permissions, traceability, and standards comparison. It was built for a different purpose.
94% of spreadsheets contain errors (Poon et al., 2024). Knowledge workers lose 19% of their working time searching for information (McKinsey). Replacement costs for lost knowledge are 150–200% of annual salary. With combined order backlogs of more than EUR 180 billion at Alstom, Siemens Mobility, and Stadler alone, these inefficiencies carry considerable weight.
Requirements volumes are rising, and regulatory requirements for traceability are becoming stricter. Companies that deliberately extend their existing systems with the response process, for example with platforms such as Tendric, save time and reduce errors.
Das tendric-Team entwickelt KI-gestützte Werkzeuge für die Ausschreibungsbearbeitung in der Industrie. Wir schreiben über Best Practices, Branchentrends und die Zukunft des Angebotsmanagements.