In short
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. Answers to its six compliance questions cited primary legal text, such as a directive on EUR-Lex, in 4% of cases (19 of 537) (Speya analysis of Peec AI data). Procurement suites and specialist tools were each named in about a third of the 3,047 answers. The engines differed most in coverage: Gemini named a tracked vendor in 81% of answers, Perplexity in 43%.
Vendor-owned websites supplied 45% to 68% of all citations in the index (Speya analysis of Peec AI data). The index is written for procurement and sourcing teams that use AI assistants for supplier work.
Speya built the index and chose its questions and its 42 tracked brands. Speya is one of those brands and publishes content aimed at these same questions. speya.ai is the most-cited domain in the data, cited in 26% of answers (Speya analysis of Peec AI data). A robustness check on the answers that cite no speya.ai page sits next to the main ranking, and it changes the ranking of some competitors.
Edition: November 2026. Data window: 16 October to 2 November 2026. This page is updated with each edition.
How we ran the index
The data comes from Peec AI, a service that sends the same questions to AI assistants every day and records the answers, the brands named and the sources cited. We read Speya's Peec project through its API, read-only, for the 18 days from 16 October to 2 November 2026.
Five engines answered: ChatGPT, Perplexity, Google AI Overviews, Gemini and Microsoft Copilot, each through its consumer interface. One answer is one question asked once on one engine on one day. That gives 3,047 answers: 612 per engine, and 599 for Google AI Overviews, which sometimes shows no overview (Speya analysis of Peec AI data).
Each of the 34 questions was asked about 90 times, once a day on each engine (Speya analysis of Peec AI data). The 3,047 answers are therefore not independent samples. For comparisons between question topics, the real sample is the number of questions, one to eight per topic. Small gaps between shares should not be read as differences.
Speya's tracking project holds 50 questions. We kept the 34 that a buyer could ask without naming any vendor, procurement suite, customer or Speya. All are in English. The full list is near the end of this page.
Speya chose the 42 tracked brands: Speya and 41 other vendors. We grouped the 41 into four types: 6 procurement suites, 20 specialist discovery, intelligence and sourcing tools, 10 directories and marketplaces, and 5 ESG and risk platforms. A type with more tracked brands has more chances to be named. Vendors outside the list are not detected. The full list is in the data appendix.
How the metrics are defined
Share of answers that name a vendor: answers mentioning the vendor at least once, divided by all answers.
Engine overlap: for each pair of engines, how many of their ten most-named vendors appear on both lists. Ties are broken alphabetically.
Source mix: every inline citation, grouped by Peec's domain classification. We checked the most-cited company sites by hand and moved two software vendors that Peec labels otherwise into the vendor group.
Primary legal text: an answer that cites an official law publication, such as a EUR-Lex law page or legislation.gov.uk. EUR-Lex summaries do not count.
What Speya discloses
Speya chose the questions and the tracked brands, from its own tracking project. Speya is one of the tracked brands. Speya publishes and updates pages aimed at these questions, including comparison and best-tool pages. Those pages raise its own figures. speya.ai is the most-cited domain. It was cited in 26% of answers and supplied 10% of all citations (Speya analysis of Peec AI data).
Those pages name other vendors, so they shape which competitors the engines name too. Speya is kept out of every vendor ranking and type count. The vendor ranking is also recomputed on the 2,253 answers that cite no speya.ai page (Speya analysis of Peec AI data).
AI answers to compliance questions rarely cite the law itself
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. Six of them deal with compliance: due diligence, risk screening, ESG checks and EU supply chain rules. Their 537 answers cited primary legal text in 4% of cases, 19 answers in all (Speya analysis of Peec AI data). Any institutional source, such as an EU, government or standards body site, was cited in 28%.
Compliance questions | Questions | Answers | Cite any institutional source | Cite primary legal text |
|---|---|---|---|---|
Ask how to comply (questions 15 and 24) | 2 | 178 | 61% | 9% |
Ask for tools or platforms (questions 8, 13, 19 and 32) | 4 | 359 | 13% | 1% |
All six | 6 | 537 | 28% | 4% |
Four of the six compliance questions ask for tools, and a tool shortlist rarely needs a legal source. The two questions that ask how to comply are the fairer test. Their 178 answers cited an institutional source in 61% of cases and primary legal text in 9% (Speya analysis of Peec AI data). Across all 3,047 answers in the index, 14% cited an institutional source.
Two limits apply. Only sites that Peec classes as institutional were scanned for legal text, so a law page hosted elsewhere would be missed. And two questions is a small base, so the 9% figure is indicative (Speya analysis of Peec AI data).
What it means for a procurement team: an AI answer about a due diligence duty rarely points to the legal text. For the duty itself, read the legal text, such as the CSDDD on EUR-Lex, and ask your legal team. Our CSDDD glossary entry and guide to the Omnibus I amendments cover the current state. This is general information, not legal advice.
Two procurement suites are the tracked vendors AI assistants name most often
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. Coupa was named in 27% of 3,047 answers and SAP Ariba in 23%, ahead of the other 39 tracked vendors (Speya analysis of Peec AI data). Scoutbee (17%), Thomasnet (15%) and Veridion (12%) complete the top five on all answers.
Speya's own pages shift which vendors are named, including competitors. On the 2,253 answers that cite no speya.ai page, Coupa and SAP Ariba stay first and second (Speya analysis of Peec AI data). Scoutbee falls from third to tenth (9%) and Veridion from fifth to twelfth (8%). Directories and ESG platforms move up. Much of the visibility of those two specialists comes from answers that also cite speya.ai.
Vendor | Type | Rank, all answers | Share, all 3,047 answers | Rank, no speya.ai citation | Share, 2,253 answers citing no speya.ai page |
|---|---|---|---|---|---|
Coupa | Procurement suite | 1 | 27% | 1 | 23% |
SAP Ariba | Procurement suite | 2 | 23% | 2 | 21% |
Scoutbee | Specialist tool | 3 | 17% | 10 | 9% |
Thomasnet | Directory or marketplace | 4 | 15% | 3 | 14% |
Veridion | Specialist tool | 5 | 12% | 12 | 8% |
Alibaba | Directory or marketplace | 6 | 12% | 7 | 11% |
Europages | Directory or marketplace | 7 | 12% | 4 | 13% |
EcoVadis | ESG and risk platform | 8 | 12% | 5 | 13% |
IntegrityNext | ESG and risk platform | 9 | 9% | 6 | 11% |
Kompass | Directory or marketplace | 10 | 9% | 8 | 11% |
Ivalua | Procurement suite | 11 | 9% | 11 | 8% |
Jaggaer | Procurement suite | 12 | 9% | 13 | 7% |
Prewave | ESG and risk platform | 13 | 8% | 9 | 10% |
Supplier.io | Specialist tool | 14 | 7% | 19 | 5% |
Keelvar | Specialist tool | 15 | 7% | 15 | 6% |
By vendor type, at least one procurement suite appeared in 33% of answers and at least one specialist tool in 34% (Speya analysis of Peec AI data). That gap is too small to read as a difference, and it comes from 6 tracked suites against 20 tracked specialists. A directory or marketplace appeared in 29% of answers and an ESG or risk platform in 18%.
Vendor type | Tracked brands | Share of answers naming at least one | Answers |
|---|---|---|---|
Procurement suite | 6 | 33% | 1,011 of 3,047 |
Specialist tool | 20 | 34% | 1,026 of 3,047 |
Directory or marketplace | 10 | 29% | 884 of 3,047 |
ESG and risk platform | 5 | 18% | 558 of 3,047 |
Answers that name vendors name several. Across the 3,047 answers, an answer named 2.4 tracked vendors on average. Among the 2,032 answers that named at least one, the average was 3.7 (Speya analysis of Peec AI data). Both are lower bounds, because vendors outside the tracked list are not counted.
What it means: AI answers to generic supplier questions tend to repeat the best-known names. A suite you already run will come up. A specialist that fits the category may appear on one engine and not on another, and its visibility can depend on which pages the engine happened to read. See supplier intelligence and supplier discovery for how the specialist categories differ.
The type of vendor named differs sharply by question topic
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. The six compliance questions named an ESG or risk platform in 77% of 537 answers, and EcoVadis in 51% (Speya analysis of Peec AI data). The five European sourcing questions named a directory in 57% of answers, led by Europages at 52%.
The six best-tool questions named a suite in 78% of answers and a specialist in 88%, and Coupa in 72% (Speya analysis of Peec AI data). The eight problem questions named directories more often than suites or specialists, in 34% of answers.
Question topic | Questions | Answers | Suite | Specialist | Directory | ESG and risk | Most-named vendor |
|---|---|---|---|---|---|---|---|
Problem questions | 8 | 718 | 14% | 17% | 34% | 3% | Thomasnet (25%) |
Best-tool questions | 6 | 537 | 78% | 88% | 40% | 13% | Coupa (72%) |
Use-case questions | 6 | 537 | 32% | 27% | 18% | 9% | SAP Ariba (24%) |
Compliance questions | 6 | 537 | 28% | 7% | 2% | 77% | EcoVadis (51%) |
European sourcing questions | 5 | 450 | 10% | 34% | 57% | 1% | Europages (52%) |
Platform comparison (indicative) | 1 | 90 | 77% | 97% | 56% | 0% | Scoutbee (79%) |
RFx question (indicative) | 1 | 89 | 52% | 3% | 0% | 1% | Coupa (43%) |
Definition question (indicative) | 1 | 89 | 2% | 11% | 6% | 0% | None at 10% or more |
Three topics hold one question each, about 90 answers per topic. Their figures are indicative. Topics also differ in wording and country, so the table shows association between topic and vendor type, with no claim about cause.
What it means: the way you phrase the question shapes the longlist. Ask about a problem and directories come up more often. Ask for the best platform and suites and specialists appear side by side. Ask about compliance and ESG ratings providers dominate.
The engines agree on the leading vendors and differ in how often they name any
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. Any two engines had 6 to 8 of their ten most-named vendors in common, 6.8 on average (Speya analysis of Peec AI data). Coupa and SAP Ariba sit in every engine's top five.
Coverage differed more. Gemini named at least one tracked vendor in 81% of its 612 answers and Perplexity in 43% of 612 (Speya analysis of Peec AI data). Thomasnet was named in 27% of Gemini answers and in 0 of 612 Perplexity answers. We read ten of those Perplexity answers in full: they named no brand at all.
Engine | Answers | Share naming at least one tracked vendor | Five most-named vendors |
|---|---|---|---|
ChatGPT | 612 | 68% | SAP Ariba, Coupa, Thomasnet, Scoutbee, Europages |
Perplexity | 612 | 43% | Coupa, SAP Ariba, Scoutbee, Veridion, Europages |
Google AI Overviews | 599 | 72% | Coupa, SAP Ariba, Veridion, Scoutbee, Ivalua |
Gemini | 612 | 81% | Coupa, Thomasnet, SAP Ariba, Alibaba, Kompass |
Microsoft Copilot | 612 | 69% | Coupa, SAP Ariba, Scoutbee, Thomasnet, Alibaba |
Engine pair | Vendors shared in top ten |
|---|---|
ChatGPT and Perplexity | 8 of 10 |
ChatGPT and Google AI Overviews | 7 of 10 |
ChatGPT and Gemini | 7 of 10 |
ChatGPT and Microsoft Copilot | 7 of 10 |
Perplexity and Google AI Overviews | 7 of 10 |
Perplexity and Gemini | 6 of 10 |
Perplexity and Microsoft Copilot | 6 of 10 |
Google AI Overviews and Gemini | 6 of 10 |
Google AI Overviews and Microsoft Copilot | 8 of 10 |
Gemini and Microsoft Copilot | 6 of 10 |
Ties in the top-ten lists are broken alphabetically, so a vendor at the edge of a list can move with one more answer.
What it means: the leading names are stable across engines. The length of the longlist is not. A buyer who asks only Perplexity will often get an answer that names no vendor, while Gemini usually names several. Our comparison of Speya and ChatGPT for supplier sourcing covers the general-purpose assistant case.
A large share of what AI answers cite comes from vendors' own websites
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. The 3,047 answers carried 21,180 inline citations. Between 45% and 68% of them pointed to sites owned by software, data or platform vendors (Speya analysis of Peec AI data). Independent sources, meaning trade press, user-generated content, institutions and reference sites, made up 25%.
The vendor-owned share is a range because Peec groups vendors with consultancies and analysts as company sites. We checked the most-cited company domains by hand. The low end counts checked vendor sites only, the high end adds every unchecked company site (Speya analysis of Peec AI data). Of the citations that remain once speya.ai is removed, 39% to 65% are vendor-owned.
Source group | Citations | Share of all citations |
|---|---|---|
Tracked vendors' sites | 3,182 | 15% |
Speya's own site (speya.ai, findmyfactory.eu) | 2,035 | 10% |
Other vendor sites, checked by hand | 4,331 | part of company sites |
Company sites checked by hand: consultancies, analysts and others | 838 | part of company sites |
Company sites not checked | 4,938 | part of company sites |
All company sites (the three rows above) | 10,107 | 48% |
Trade press, news and blogs | 1,821 | 9% |
User-generated (YouTube, LinkedIn, forums) | 1,422 | 7% |
Institutional (government, EU, standards bodies, associations) | 1,167 | 6% |
Reference | 826 | 4% |
Other | 620 | 3% |
Domain | Source group | Citations | Share of answers citing it |
|---|---|---|---|
speya.ai | Speya's own site | 2,031 | 26% |
youtube.com | User-generated | 633 | 9% |
suplari.com | Tracked vendor | 478 | 8% |
procurementmag.com | Trade press or blog | 473 | 7% |
veridion.com | Tracked vendor | 331 | 6% |
procurementaiagents.com | Trade press or blog | 329 | 6% |
europa.eu | Institutional | 324 | 3% |
sourceready.com | Tracked vendor | 311 | 5% |
linkedin.com | User-generated | 310 | 4% |
mercanis.com | Other vendor (checked by hand) | 276 | 4% |
supplier.io | Tracked vendor | 265 | 4% |
ivalua.com | Tracked vendor | 244 | 5% |
apexanalytix.com | Other vendor (checked by hand) | 243 | 3% |
guideflow.com | Other vendor (checked by hand) | 227 | 4% |
integritynext.com | Tracked vendor | 227 | 3% |
Peec's labels for trade press and user-generated sites were used as given, except for two software vendors listed in the data appendix. Some sites labelled as trade press publish product rankings, so the independent share is an upper bound.
What it means: much of what AI answers cite on supplier tools is written by vendors of those tools. Vendor comparison pages and best-tool lists are marketing, and the answer inherits their framing. That includes Speya's own pages.
What this means if you use AI assistants for supplier work
AI assistants are a reasonable place to start a longlist. Six habits make their answers safer to use.
Open the source behind each claim. Click through to the cited page and check who wrote it. A vendor's page about its own category is a sales document.
Ask more than one engine. The engines agree on the leading names and differ in how often they name any vendor. A second engine widens the longlist.
Phrase the question several ways. Problem questions, best-tool questions and compliance questions surfaced different types of vendor.
Use primary sources for compliance. Read the regulation, the regulator's guidance or the certification body's register. Treat an AI summary of a legal duty as a pointer.
Treat vendor-written listicles as marketing. That includes Speya's own comparison and best-tool pages. They are useful for names and a poor basis for ranking.
Keep the paper trail. Record the question, the engine, the date and the sources for any supplier that reaches your shortlist. Answers change from day to day.
For the next step, our checklist for evaluating supplier discovery tools and supplier vetting checklist list the checks to run. How procurement teams use AI agents covers the wider picture.
What we could not verify
Speya chose the questions and the tracked brands. They reflect the buying situations Speya cares about: supplier discovery, Europe and compliance. A different set would give different numbers.
Untracked vendors are invisible. Mercanis, for example, is among the most-cited vendor domains and is not a tracked brand. The vendor ranking holds among tracked vendors only.
English only. Buyers asking in German, Swedish or other languages may get different answers.
Answers vary by day, version and location. The figures describe this 18-day window. Each question ran from the country set on it, mostly in Europe and the US.
Classifications are approximate. Source groups are Peec's labels plus our hand check of the most-cited company domains. Vendor types are our own grouping.
Legal text scan is partial. Only institutional-class sites were scanned for primary legal text.
AI answers are not buyer behaviour. The index records what the engines said. It says nothing about which vendors buyers chose, or about the quality of any vendor.
Where Speya appears in this data
Read this section as Speya's own position in an index Speya built, on questions Speya chose and writes content for.
The AI Procurement Answers Index asked 34 generic supplier questions daily on five AI assistants, from 16 October to 2 November 2026. Speya was named in 21% of 3,047 answers and named before every other brand in 14% (Speya analysis of Peec AI data). In the 2,253 answers that cite no speya.ai page, Speya was named in 5%. Most of Speya's visibility comes from answers that read its own pages.
The 34 questions in the index
Each question ran once a day on each of the five engines. Questions 15 and 24 are the two compliance questions that ask how to comply.
# | Question | Topic |
|---|---|---|
1 | How do procurement teams find suppliers for a category they have never sourced before? | Use-case questions |
2 | How to find qualified manufacturers in Germany for industrial components | European sourcing questions |
3 | Best AI agents for procurement and sourcing teams in 2026 | Best-tool questions |
4 | How to find a second-source supplier to reduce single-source risk | Problem questions |
5 | How do large companies find and qualify new suppliers before onboarding? | Problem questions |
6 | How to build a defensible supplier shortlist that holds up in an audit | Problem questions |
7 | Supplier discovery platform for nearshoring production to Eastern Europe | European sourcing questions |
8 | Best supplier due diligence tools for enterprise procurement teams in Europe | Compliance questions |
9 | What is supplier intelligence and how is it different from supplier discovery? | Definition question |
10 | How to send an RFI, RFP, or RFQ to multiple suppliers at once | RFx question |
11 | Best supplier discovery platform for Scandinavian and Nordic procurement teams | European sourcing questions |
12 | Best alternatives to manual supplier research for procurement teams | Best-tool questions |
13 | Supplier risk monitoring platform for European procurement teams | Compliance questions |
14 | AI procurement tools for manufacturing companies sourcing components | Use-case questions |
15 | How to check supplier ESG scores and financial stability before onboarding | Compliance questions |
16 | How do I build a qualified supplier shortlist for an RFQ quickly? | Problem questions |
17 | Platform for compliance teams to screen and monitor supplier risks | Use-case questions |
18 | AI tools for procurement teams to find suppliers faster | Best-tool questions |
19 | AI platform for supplier compliance screening - ESG and financial health and certifications | Compliance questions |
20 | How do sourcing teams use AI to find new suppliers in 2026? | Use-case questions |
21 | How can R&D teams find specialist component suppliers quickly? | Use-case questions |
22 | AI-powered sourcing platform based in Europe with GDPR compliance | European sourcing questions |
23 | How to find reliable manufacturers in Europe for nearshoring | Problem questions |
24 | How do procurement teams comply with EU supply chain due diligence requirements? | Compliance questions |
25 | What is agentic supplier discovery and how does it work? | Use-case questions |
26 | How to reduce supplier discovery time in procurement | Problem questions |
27 | How to automate supplier research for procurement teams | Problem questions |
28 | Top supplier intelligence platforms with AI-powered search | Best-tool questions |
29 | How to find European manufacturers as alternatives to Chinese suppliers | European sourcing questions |
30 | What is the fastest way to find and vet new suppliers? | Problem questions |
31 | Best tool to find and vet new suppliers quickly | Best-tool questions |
32 | Best tools for CSDDD supply chain due diligence compliance | Compliance questions |
33 | What is the best AI-powered supplier discovery platform for enterprise procurement teams? | Best-tool questions |
34 | Supplier discovery platform comparison 2026 - which has the best AI? | Platform comparison |
Data appendix: tracked brands, groups and reclassified domains
Every tracked brand, its type and its share of answers, on all 3,047 answers and on the 2,253 answers that cite no speya.ai page (Speya analysis of Peec AI data). Speya is listed for completeness and is excluded from every ranking above.
Brand | Type | Share, all answers | Share, answers citing no speya.ai page |
|---|---|---|---|
Coupa | Procurement suite | 27% | 23% |
SAP Ariba | Procurement suite | 23% | 21% |
Speya | Speya (publisher of this index) | 21% | 5% |
Scoutbee | Specialist tool | 17% | 9% |
Thomasnet | Directory or marketplace | 15% | 14% |
Veridion | Specialist tool | 12% | 8% |
Alibaba | Directory or marketplace | 12% | 11% |
Europages | Directory or marketplace | 12% | 13% |
EcoVadis | ESG and risk platform | 12% | 13% |
IntegrityNext | ESG and risk platform | 9% | 11% |
Kompass | Directory or marketplace | 9% | 11% |
Ivalua | Procurement suite | 9% | 8% |
Jaggaer | Procurement suite | 9% | 7% |
Prewave | ESG and risk platform | 8% | 10% |
Supplier.io | Specialist tool | 7% | 5% |
Keelvar | Specialist tool | 7% | 6% |
GEP | Procurement suite | 6% | 6% |
Fairmarkit | Specialist tool | 6% | 6% |
GlobalSources | Directory or marketplace | 5% | 7% |
Matchory | Specialist tool | 5% | 2% |
Tealbook | Specialist tool | 5% | 4% |
osapiens | ESG and risk platform | 4% | 6% |
Suplari | Specialist tool | 4% | 3% |
Wonnda | Specialist tool | 3% | 3% |
Arkestro | Specialist tool | 3% | 2% |
Zycus | Procurement suite | 2% | 2% |
Supply EU | Directory or marketplace | 2% | 2% |
SourceReady | Specialist tool | 2% | 2% |
Globality | Specialist tool | 2% | 2% |
Craft | ESG and risk platform | 2% | 1% |
Kodiak Hub | Specialist tool | 2% | 2% |
Made-in-China | Directory or marketplace | 1% | 1% |
DeepStream | Specialist tool | 1% | 1% |
MFG | Directory or marketplace | 1% | 1% |
Alpas | Specialist tool | 0% | 0% |
Xometry | Directory or marketplace | 0% | 0% |
Lio | Specialist tool | 0% | 0% |
Beroe Inc | Specialist tool | 0% | 0% |
Maker’s Row | Directory or marketplace | 0% | 0% |
Altana | Specialist tool | 0% | 0% |
Tocco | Specialist tool | 0% | 0% |
ensun.io | Specialist tool | 0% | 0% |
Group | Tracked brands |
|---|---|
Procurement suite | 6 |
Specialist tool | 20 |
Directory or marketplace | 10 |
ESG and risk platform | 5 |
Speya | 1 |
Total | 42 |
Domain | Peec label | Counted as | Reason |
|---|---|---|---|
apexanalytix.com | UGC | Vendor-owned | Supplier-management software vendor (apexanalytix); Peec labels it UGC. |
mercanis.com | Corporate | Vendor-owned | Procurement software vendor (Mercanis); Peec labels it Corporate, now counted as a checked vendor site. |
Method and data: this page. Shares are whole percentages computed once from raw counts. The question set, brand list and source groups will be reviewed before each edition.
Frequently asked questions
Which AI assistant is best for supplier research?
The engines answer differently. Gemini named a tracked vendor in 81% of answers and Perplexity in 43%, while any two engines shared six to eight of their top ten vendors (Speya analysis of Peec AI data). Ask at least two engines and open the sources they cite. Our Speya and ChatGPT comparison covers general assistants.
Do AI assistants favour big procurement suites?
Coupa and SAP Ariba were the two most-named tracked vendors, in 27% and 23% of answers (Speya analysis of Peec AI data). Which type appeared depended on the topic: best-tool questions named a suite in 78% of answers, problem questions in 14%.
Can I trust AI answers on supply chain due diligence rules?
Use them as a pointer. Answers to the six compliance questions cited primary legal text in 4% of cases, and answers to the two how-to-comply questions in 9% (Speya analysis of Peec AI data). Check the obligation in the legal text and with your legal team. This is general information, not legal advice.
Where do AI assistants get their information about supplier tools?
Mostly from company websites. Vendor-owned pages supplied 45% to 68% of citations, and independent sources 25% (Speya analysis of Peec AI data). The range reflects company sites we could not classify by hand.
Why is Speya in its own index?
Speya runs the tracking project, chose the questions and brands, and writes content for these questions. Its pages are among the cited sources. Speya is kept out of the rankings and reported in a separate, caveated section.
Next edition: February 2027, on this page. Method and data: this page.