Why counting TED notices overstates the number of procurement opportunities

A notice is a published record, not necessarily a new procurement opportunity. One procedure can generate planning information, a competition notice, amendments, an award result and later changes. Counting every publication as a separate opportunity can overstate both market activity and the amount of work available to suppliers.

TED’s search guidance explains that its latest-version option filters out earlier versions while leaving those older publications available in the system. A search result count therefore depends partly on the chosen scope. Two analysts can retrieve different totals without either encountering a different underlying market. The TED defence service search guide identifies relevant service buying through the requirement and buyer, not one broad keyword.

Decide what the analysis is counting before collecting data. Is the objective to measure notices, procurement procedures, lots, contracts or selected suppliers? Each can be a legitimate unit, but they answer different questions. A report should name the unit rather than use the ambiguous word “opportunities” for all of them.

Consider a hypothetical procurement with three lots. It is published, amended twice and later produces awards for two lots while the third is discontinued. That sequence can contain several notices, one procedure, three proposed lots and a different number of actual contracts. A raw count cannot communicate the commercial outcome.

Maintain links between those objects. Keep the notice publication identifier, buyer’s reference, procedure identifier where available, lot and contract references, date and record type. A later notice should update the interpretation of the earlier record without erasing the source history.

Value fields need similar treatment. A planned overall amount, a lot estimate and an awarded contract value may overlap. Adding them together does not create a larger market. If a total cannot be reconciled because public information is incomplete, report the limitation rather than choose the largest number.

Company research also benefits from this structure. A supplier may appear in several versions of the same award notice. That is repeated evidence about one event, not proof of repeated wins. Conversely, several distinct contracts can arise from one procedure. Deduplication must preserve real commercial distinctions rather than collapse everything with a similar title.

TED’s API documentation supports systematic retrieval of published notices. The data model and editorial rules still belong to the analyst. Downloading records more efficiently does not resolve identity, missing fields or the meaning of a maximum contract value.

When presenting a chart, state the population and counting rule. For example, “unique published procedures identified in the reviewed sample” is more informative than “defence deals” if that is what the data actually contains. Separate exact matches from records linked through analyst judgment.

Follow the relationship between records

The eForms documentation distinguishes a procurement procedure from the notices describing it. Most competition, result, change and modification notices can share a procedure identifier, even when different publishing systems submit them. Planning notices are an important exception: they lack a procedure identifier because one plan may lead to several procedures. Later records can refer back to the planning notice through dedicated references. The Publications Office's explanation of notice relationships

That exception has an immediate analytical consequence. Grouping every row with an empty procedure identifier into one record would collapse unrelated plans together. Dropping those rows would erase evidence about future purchasing intentions. A research dataset should instead retain planning records as planning records and use explicit later references when establishing a relationship. The missing identifier has a procedural meaning; it is not simply a blank cell to fill with a guessed buyer-and-title match.

Legacy TED XML notices introduce a second distinction. The official guidance says they do not contain the eForms procedure identifier. Links from later eForms records must use the appropriate reference to the earlier publication. A market study covering both formats therefore needs to preserve those references instead of assuming a modern identifier exists throughout the historical series. Guidance on procedures spanning legacy and eForms notices

For a commercial analyst, this is a reason to separate exact relationships from inferred ones. An explicit notice reference can support a reproducible link. A similar title from the same buyer may warrant investigation, but it could also describe a recurring purchase under a new procedure. Combining those cases without a distinction would make the resulting market count look more certain than the underlying evidence permits.

A change notice carries a consolidated account

The dedicated change-notice specification explains that a change notice has its own notice identifier and refers to the notice it changes. It contains the original information with the changes applied. A later change consolidates earlier changes too, while its change section describes the latest amendments. Changes can also concern procurement documents outside the notice itself. The eForms change-notice specification

This structure explains why comparing titles or publication dates alone is inadequate. A new notice identifier can represent an updated account of an existing purchase. Conversely, two purchases can have nearly identical titles while remaining separate procedures. The analyst needs the relationship fields and the substantive scope, rather than a rule that simply removes every repeated title or keeps every new identifier.

A useful dataset can retain both a current view and a historical sequence. The current view supports the question of what the buyer presently says about a procedure. The sequence supports questions about how its timetable or description changed. Removing superseded records from the analytical total need not mean discarding the evidence that explains the current state. Keeping that history also makes later corrections to the research easier to audit.

Allocate amounts to the thing they describe

Consider a hypothetical procurement for a business application with separate implementation and support lots. A planning estimate describes the intended purchase before bids. Later results may identify contracts for the two lots. A further modification may update one contract. Adding every amount encountered along that history would mix expectations, awards and later changes, even after the notice records had been grouped correctly.

The market model should instead select the economic measure it wants to report. A study of announced awards uses the relevant award evidence and its date. A study of planning activity retains estimated values under that label. A study of contract changes investigates what each modification says about the existing agreement. These measures can sit beside one another in a report, but they should not silently feed the same revenue total.

The same discipline applies when identifying suppliers. A company name repeated in several notices about one contract does not establish several wins. Two distinct contracts awarded to the same company are different observations. If an award involves a group of companies, the published total should not be assigned in full to each member as though the notice disclosed their individual revenue shares.

The resulting report should make its counting rule readable: which purchasing stage it covers, what constitutes one observation and how linked records were handled. A compact explanation of unresolved relationships is more useful than a falsely precise headline total. This approach allows procurement evidence to support commercial decisions while preserving the distinction between what the public record identifies and what an analyst has inferred from it.

The practical result is less dramatic but more useful intelligence. A company can see whether a buyer has launched new work, revised existing requirements or completed an award. That makes it easier to decide where to spend proposal effort. Accurate counts depend on preserving procurement history, not simply collecting more notices. The SIPRI military spending analysis explains why a national spending total cannot substitute for a supplier's addressable demand.