The 2027 Public Tender Report: what 304,000 tender notices and 670,000 contract awards tell bidders

Key findings

• One in five tender notices gave suppliers under 10 days. 56,654 of 281,669 notices (20.1%), published November 2025 to October 2026.
• A US federal bidder gets 12 days. An EU bidder gets 31. Medians on SAM.gov (115,686 notices) and TED (118,972), published February to October 2026.
• Nine in ten US federal notices give less than 30 days. 88.8% on SAM.gov.
• Four in ten EU notices give less than 30 days from publication. 42.7% on TED, which is not a count of breaches.
• September is the crunch month in US federal buying. In September 2026, 68.6% of SAM.gov notices closed within 14 days and 46.3% gave under 10 days.
• One in three EU open-procedure contract lots drew a single tender. 32.9% of lots whose TED result notice states the tender count; 35.3% without framework agreements.
• Three in four notices do not state a value. Only 24.2% of 304,463 notices give a contract value.
• Construction is the largest category in Europe. 17.4% of notices with a CPV code (25,578 of 146,764), median stated value EUR 492,000.

1. What we counted

Two sources carry most of the data: TED, with 132,917 notices, and SAM.gov, with 118,755.

On 7 November 2026 (06:12 UTC) our index held 343,814 public notice records from 21 sources. This report uses the 304,463 tender notices published from 1 November 2025 to 31 October 2026, after removing duplicates and Australian contract-award records, from buyers in 193 countries.

The United States leads the buyer countries with 121,178 notices, then Germany (26,343), Poland (22,207), France (17,330) and Spain (16,587). Monthly counts measure our coverage, not the market.

Full source table and top 20 countries in the appendix

2. How long suppliers get

A US federal bidder gets 12 days to respond. An EU bidder gets 31.

The response window is the deadline stated in the notice minus its publication date, in days, for notices published 1 November 2025 to 31 October 2026 (full definition in the Methodology).

Full table in the appendix

Across 18 sources (grants and EU funding calls left out), 281,669 notices, the median was 20.7 days (middle half 11.0 to 31.9). 8.6% gave 7 days or less, 20.1% under 10, 34.1% 14 or less and 67.3% 30 or less. SAM.gov and TED make up 83% of this pool, so read the per-source figures first.

Against the legal floors

SAM.gov: 88.8% of notices give less than 30 days, and three in four give 16 days or less. US federal rules (FAR 5.203) require 30 days only for non-commercial purchases above the simplified acquisition threshold, so most federal notices fall outside the rule.

TED: 42.7% of notices give less than 30 days from publication, 30.9% give 30 to 35 days, and 26.4% give more than 35. 14.6% give less than 15 days. The EU directive's usual minimum for an open procedure is 35 days from sending the notice, 30 with electronic submission. These are not breaches: TED carries procedures other than open ones, and the legal clock starts when the notice is sent, a few days before it is published.

The legal texts in full are in the appendix

By country

Two European markets stand out for short windows: Poland (17.5% under 10 days) and Lithuania (15.9%), against 7.0% in Germany and 3.0% in France.

Full table in the appendix

3. Where the work is

Construction is the largest category in Europe: 17.4% of notices with a CPV code.

The Common Procurement Vocabulary (CPV) is the EU's classification for public contracts. 146,764 notices from European and UK sources published November 2025 to October 2026 carry a CPV code. We count each notice once, by the first code it lists.

Full table with median values in the appendix

Three in four notices do not state a value.

73,641 of 304,463 notices (24.2%) state a contract value, and only 28,756 (9.4%) state it with an explicit currency of euros. Sums mislead. In IT services, the 10 largest notices hold 81.9% of the total stated value; the largest single notice states EUR 10 billion. These are framework ceilings, not spend. We rank by count and median, not by sum.

Across all sources, Lucius sector labels put Defence first with 48,916 notices, then Construction 45,904, Engineering 39,692, Manufacturing 35,876 and Facilities Management 31,982; 196,798 of 301,620 notices (65.2%) have a label. Defence ranks first because US federal notices are 39% of this sample.

4. Pass/fail gates: what this data cannot show

Our notice data cannot show which pass/fail gates suppliers miss.

A bid can fail before anyone scores it: a missing certificate, an unsigned declaration, a turnover threshold. Public notices do not report why a tender was rejected, and the notice text rarely lists the gates. In October we measured UK IT notices (309 notices): Cyber Essentials was named in 1.0% and ISO 27001 in 0.6%. The gates sit in the tender documents, not the notice.

Our June 2026 study read 136 full tender document sets and found a median of 45 mandatory requirements per tender. That remains our only pack-level figure.

How we plan to measure the gates, in the appendix

5. The finding: one in three EU open-procedure contract lots drew a single tender

One in three EU open-procedure contract lots drew a single tender; without framework agreements, 35.3%, above the European Commission's 28% for a typical EU country.

We used TED result notices, one row per awarded lot: 670,661 lots awarded under an open procedure from 1 November 2025 to 31 October 2026, from 225,065 notices. A lot counts as single tender when its notice states that one tender was received.

Where a notice leaves the count out, our table holds the number of winners instead, so the raw 44.0% mixes real single tenders with missing counts. A check of 400 lots against the notices separated them: an estimated 81.7% of lots state their count.

• 32.9% of open-procedure lots that state their tender count received a single tender (95% range 30.6% to 34.9%).
• 35.3% when framework agreements are also left out, as the European Commission counts it (95% range 29.9% to 41.5%). Framework agreements made up an estimated 41.8% of open-procedure lots.

The EU Single Market Scoreboard puts a typical, mid-ranking EU country at 28% for 2024; the two measures differ in period, unit, procedures, buyers and sample. There is no country ranking this time, because the missing counts sit in a few countries. The chart is at the top of this report.

The check, the five differences and the country gaps, in the appendix

What to expect in 2027

US federal windows did not move, September was the exception, and EU windows did not shorten.

We only list a trend where our 2026 data shows one.

• US federal windows did not move. The SAM.gov median was between 11.0 and 13.0 days in every month from March to October 2026. Nothing in 2026 points to a change in 2027.
• September was the exception. 68.6% of SAM.gov notices closed within 14 days in September 2026, the last month of the US federal fiscal year, against 56.0% to 62.7% from March to August. We have one fiscal year of data. If the pattern repeats, September 2027 will be the tightest month again.

• EU windows did not shorten. The TED median was 31.2 to 31.3 days in each month from July to October 2026 (29,000 to 37,000 notices a month).
• No single-tender trend yet. On the raw count, the share of single-tender lots fell from 47.4% (results published July to September 2025) to 39.8% (July to September 2026). The raw count mixes stated counts with missing ones, so this fall may reflect how notices are filled in rather than more competition. We will measure the trend on stated counts only in the next edition.
• No volume forecast. Our 2026 monthly counts track the growth of our own coverage, not the market.

Rules for bidders in 2027

1. 1

   On SAM.gov, have your standard documents ready before the notice appears.

   The median SAM.gov window is 12.0 days, and 34.9% of notices give under 10.

2. 2

   Keep September clear for US federal bids.

   In September 2026, 46.3% of SAM.gov notices gave under 10 days.

3. 3

   Treat an EU prior information notice as your early warning.

   After one, the EU minimum for an open procedure can drop to 15 days (Article 27); 14.6% of TED notices gave less than 15.

4. 4

   In the EU, make the go/no-go call in week one.

   The TED median is 31.3 days, and 42.7% of notices give less than 30.

5. 5

   Do not filter opportunities by contract value.

   Only 24.2% of notices state one; on SAM.gov, 3 in 118,755 do.

6. 6

   Read the award statistics before you pick a field.

   About four in five lots in TED result notices state how many tenders were received (81.7%). Use them to find the buyers and categories where few suppliers bid.

7. 7

   Check whether a lot is a framework agreement.

   An estimated 41.8% of open-procedure lots on TED were; the orders come later, through call-offs or mini-competitions.

8. 8

   Search EU notices by CPV division, and US notices by keyword and US codes.

   CPV codes cover 146,764 European notices; US federal notices use NAICS and PSC codes instead.

9. 9

   Read the whole pack, not a one-pass summary.

   A general AI model asked once found 66.2% of the requirements on long packs (47 to 84 pages); see the next section.

Where general AI tools go wrong on tender packs

A general AI model asked once for every requirement in a tender pack left requirements out on every pack.

We asked Google's Gemini 4 model, on 1 November 2026, in one request per pack, to list every requirement in 18 packs of 2 to 84 pages. We compared its list with a reference from a slower reading that works through each pack in parts. The reference is machine-made, so these figures show what a one-pass answer leaves out, not absolute accuracy.

• With the pack attached and default settings, it listed 70.7% and 70.3% of the reference requirements in two runs, below the slower reading on all 18 packs. With the full text and its deepest reasoning setting, it reached 80.3% (16 packs).
• On the five longest packs (47 to 84 pages), the best setting found 66.2% of the requirements; on the other 11, 86.7%.
• Where a file passed the input limit of the tool used in the test, the rest never reached the model. That happened on 6 of 18 packs, and the answers gave no sign.
• What it listed was mostly right: 92.8% and 91.8% of its lines matched a reference requirement, and its page references were accurate 96.2% and 96.3% of the time.

In these tests the problem was not wrong answers. It was what the model left out, without saying so.

The full test, in the appendix

Methodology

• Data source. The Lucius AI index of public procurement notices, production database, queried read-only on 7 November 2026, 06:12 to 06:22 UTC (single-tender check 06:31 to 06:45 UTC). Public notices only. No customer data.
• Base. Notices published 1 November 2025 to 31 October 2026 (304,463 tender notices after removing records marked as duplicates and 18,793 AusTender contract-award records). Award results: TED result notices for lots awarded in the same period (670,661 open-procedure lots). The "670,000 contract awards" in the title are these 670,661 lots awarded under an open procedure, as stated in TED result notices.
• Sources reported. Some national and international portals are included in totals but not reported separately, under their terms of use.
• Response window. Deadline in the notice minus publication date, in days. Deadline after publication, window of 365 days or less. Sources with 200 or more notices. Grants (Grants.gov) and EU funding calls (Funding & Tenders Portal) are shown but left out of pooled figures.
• Legal floors. Directive 2014/24/EU, Article 27 (legislation.gov.uk/eudr/2014/24/article/27); FAR 5.203, FAC 2026-01 (acquisition.gov/far/5.203), both read 7 November 2026.
• Coverage. Each source joined the index on its own date (see the source table in the appendix). Figures describe the notices we hold, not every notice published.
• Deadlines. 94.2% of notices state a deadline. Notices without one are left out of window figures. Some tender packs set a different submission date, and clarification questions often close earlier. TED publication can follow the sending of a notice by a few days, so windows counted from publication are slightly shorter than legal ones counted from sending.
• Duplicates. Records marked as duplicates are removed. A notice published on both TED and a national portal can appear in both; per-source figures are unaffected, country totals may count it twice.
• Values. Only explicit euro values are used for value figures. Totals are dominated by framework ceilings.
• Categories. CPV division = first two digits of the first CPV code a notice lists. Sector labels are assigned by Lucius and can be several per notice.
• Single tender. The number of tenders received as stated in the TED result notice (eForms statistics code "tenders"). Our award table stores that count, or the number of winners where a notice leaves it out. We separated the two with a fixed random sample of 300 single-tender lots and 100 multi-tender lots, read from each notice's XML on TED on 7 November 2026 (394 notices), weighted by the size of each group. Framework agreements were identified from each lot's framework-agreement field in the same XML. Ranges are 95% bootstrap intervals. Counts are per lot.
• Changes from earlier figures. Since 25 October 2026: a fixed 12-month window, duplicates removed, the AusTender figures withdrawn, some portals reported in totals only. Since June 2026: a different source mix and a 365-day cap instead of two years.
• AI tests. Run by us on 1 November 2026, as described in the section above. Lucius AI makes software that reads tender packs; the tests above were run by us on public documents.
• Queries. Every figure has a recorded query, run time and row count. Available on request.

Download the data

The key tables of this report in one CSV file, one figure per row, under CC BY 4.0. The full report is also available as a PDF.

Columns

table

Which table the row belongs to (listed under Tables).

label

The row: a source, country, month, CPV division or measure.

measure

What the value measures, for example median_days or share_under_10_days.

value

The figure, as published (percentages to one decimal).

unit

notices, lots, days, percent or EUR.

period

The period the figure covers, as an ISO 8601 interval.

definition

How the figure was counted.

Tables

notices_by_source

Tender notices by source, November 2025 to October 2026.

top_countries

Top 20 buyer countries by notices.

window_by_source

Response window by source: notices, median, p25, p75, share under 10 days, share 14 days or less.

window_pooled

Response window across 18 tender sources together.

window_by_country

Response window by country, 1,000 or more notices.

sam_gov_monthly

SAM.gov by month, March to October 2026.

cpv_divisions

Top 10 CPV divisions, European and UK sources.

single_tender

Single-tender share of EU open-procedure lots, with 95% ranges.

Sources and licences

This report uses aggregate figures computed by Lucius AI from public notices. The underlying notices remain under their publishers' terms.

• TED (ted.europa.eu): © European Union, CC BY 4.0.
• SAM.gov: public domain (US Government work).
• Find a Tender, Contracts Finder, Public Contracts Scotland: Contains public sector information licensed under the Open Government Licence v3.0.
• Plataforma de Contratación del Sector Público (contrataciondelestado.es): Origen de los datos: Ministerio de Hacienda.
• eTenders Ireland: Contains Irish Public Sector Data licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) licence.
• CanadaBuys: Contains information licensed under the Open Government Licence, Canada.
• AusTender: AusTender materials are released by the Department of Finance. AusTender materials are subject to change and should be verified on the AusTender website to ensure the information is up to date and correct: www.tenders.gov.au. CC BY 3.0 AU.
• World Bank: The World Bank: Procurement Notice: World Bank Group. CC BY 4.0.
• Inter-American Development Bank: Inter-American Development Bank: IDB Project procurement bidding notices and notification of contract awards dataset. CC BY 4.0.
• eTenders South Africa: National Treasury, Republic of South Africa: eTender Publication Portal. PDDL 1.0.
• Grants.gov: This product uses the Grants.gov API but is not endorsed or certified by the U.S. Department of Health and Human Services.
• European Commission, Funding & Tenders Portal: CC BY 4.0.
• Uganda eGP, Zambia e-GP: used under each portal's terms and conditions.
• European Commission, Single Market Scoreboard (single-market-scoreboard.ec.europa.eu): cited for comparison.

How to cite