Imagine a small engineering firm preparing a bid for public work. On Monday, its team downloads the specification, prices the materials and books time with a specialist subcontractor. On Thursday, the procurement portal publishes a changed notice. The current version is available, but the firm’s practical question is harder: which requirements moved, and does the bid need to be rebuilt?
This is a hypothetical scene, but the design problem is real. A consolidated notice is good at answering what is valid now? It is much less efficient at answering what changed? When public systems publish a replacement without a clear field-level difference, they distribute the same comparison task to every potential supplier. The remedy is not another procurement portal. It is a small, verifiable layer: a change card that places the previous value, the new value and the effect on the deadline beside the authoritative notice.

The notice already has a history
The European Union’s eForms system provides a useful starting point because changes are not merely informal annotations. According to the official TED eForms documentation, a Change notice reproduces the parent notice in its consolidated state and adds a dedicated change section. It references the notice being changed through the Change Notice Version Identifier, requires a coded reason and may include explanatory text. Both the original and the Change notice remain published.
The schema goes further. Its Change Notice specification can identify changed sections, indicate whether procurement documents changed and record the date of that document change. This is more structure than a PDF posted under the label “updated documents”. It creates the ingredients for a machine to trace versions.
Yet a consolidated version and a difference view solve opposite reading tasks. Consolidation removes obsolete information so that a reader can act on the latest state. A difference view preserves the transition so that a reader can judge the work caused by it. Procurement needs both. A supplier must know the current deadline, but also whether a technical requirement, quantity, eligibility condition or evaluation rule changed after its estimate began.
Comparison is part of access
Procurement transparency is often treated as a publication question: was the notice made public? Access also depends on the cost of recognising what the publication means. A large bidder may have a dedicated monitoring team and software. A smaller firm may divide that work among people who are also estimating, designing and contacting suppliers. Publishing the same new document to everyone does not impose the same burden on everyone.
The OECD’s 2025 review of public-procurement practice connects transparency with fair access for suppliers, including smaller firms, while warning that published data must be meaningful for its users. A change log is one way to make that principle operational: it reduces duplicated detection work without changing the substantive rules of the competition.
Time matters as much as visibility. Article 47 of Directive 2014/24/EU requires contracting authorities to extend tender deadlines when significant changes are made to procurement documents, with the extension proportionate to the importance of the change. A readable difference does not decide whether a change is legally significant. It does, however, expose the facts that suppliers and authorities need in order to ask that question.
What a change card should contain
A useful change card would sit above the authoritative notice and answer five narrow questions:
- Which version changed? Link the new notice to its exact parent, not merely to a procurement title.
- When was the change published? Show a precise timestamp and the current submission deadline in the same view.
- Which fields or sections changed? Name them in the terminology used by the notice.
- What was the transition? Present old and new values side by side, while linking both source versions.
- Did supporting documents change? Flag their change date and link to the authoritative document set rather than pretending that structured fields capture every amendment.
The card should distinguish detected facts from interpretation. “Deadline changed from 20 to 27 November” is a field comparison. “This gives suppliers enough time” is a judgment. The first can be generated; the second remains contestable. That boundary matters because a helpful interface must not quietly become an unofficial legal notice.
This approach extends a principle already discussed in A Public Form Should Survive an Interruption: a public system should carry essential state instead of making each person reconstruct it. Here the state is not an unfinished application but the path from one procurement version to the next.
The strongest objection: a diff can lie
A redline looks objective, but document comparison can produce noise. Reordered sections may appear deleted and reinserted. A revised attachment can contain a consequential drawing change that no notice-field comparison explains. Translations can change wording without changing the underlying requirement. A neat visual could therefore create false confidence.
That is the strongest reason to keep the proposal modest. The change card should never replace the consolidated notice or procurement documents. It should identify its source versions, label machine-detected comparisons, expose unchanged context on demand and state when attachments fall outside the comparison. Public buyers should be able to add a human explanation, but not erase the underlying computed transition. If a parser cannot match two structures reliably, the card should say so rather than manufacture a clean answer.
A pilot using data that already exists
TED’s official Search API allows published notices to be searched and downloaded in XML for reuse without authentication. That makes a bounded civic experiment possible.
Choose one type of competition notice and collect a small set of procedures with published Change notices. For each pair, use the parent-version reference to retrieve both records, normalise the structured business terms and produce a draft change card. Then ask procurement practitioners and occasional suppliers to complete two tasks: identify the changed requirements and decide which parts of a sample bid would need review. Compare their completion time and missed changes using the normal consolidated view and the change card.
The pilot should record false differences, missed attachment changes and fields that cannot be compared safely. Success is not “users liked the interface”. It is a lower time-to-identification without an increase in missed material changes. The output could remain a browser-based demonstrator; no integration with a live submission system is needed to test the idea.
If the card helps, the next institutional question is ownership. A procurement portal could generate it at publication, a civil-society service could build it from open data, or suppliers’ tools could compute it locally. The important decision is simpler: should every bidder continue paying separately to discover the same public change, or should the public record make its own transition legible?
What concrete example would confirm or challenge this article?
Choose a reaction, write a comment, or ask your AI assistant to send a proposal for review. Comments are reviewed before publication.