Original Research · Arc Shift Ventures

Delegations don’t fail.
They land — and then fly home.

Innovation corridors run the same play on every continent: Nordic→Singapore, Germany→APAC, France→North America, Gulf sovereign programs→everywhere, Australia’s landing pads, Korea and Japan’s inbound centres, India’s bilateral bridges, Chinese municipalities’ launch-pads into ASEAN. We pulled every published corridor and trade-mission outcome we could find, across every region we could reach, back toward its source. The pattern does not change by geography, budget, or income level: independent evaluation is nearly absent, self-reported “success” hides an undefined denominator, and the week works — the conversion happens before the flight out and after the flight home, in the stretch nobody owns. This is not a Nordic-Singapore quirk. It is structural to how governments run these programs, everywhere.

Talk about a corridor you’re running See the numbers

Four numbers that frame the problem

9.4pp
vs 6.7pp

the two-year gain in a small firm’s odds of becoming an exporter from direct partner-search and matchmaking support, against market-intelligence-only support — the cleanest causal comparison we found anywhere between “make an introduction” and “send information.” Introductions win.

Munch & Schaur, American Economic Journal: Economic Policy, 2018 (Danish Trade Council data) · Peer-reviewed, firm/year fixed effects · Grade A

1 of ~10

internationalisation-assistance programmes — across a 2026 review spanning roughly ten countries, Korea to Canada — has ever had a formal evaluation. That one evaluation measured whether the programme felt well-run, not whether it moved a number. India and China are not in the review at all.

OECD, Incubation in Entrepreneurial Ecosystems, 2026 (our count of the chapter’s contents) · Grade A

≈50%
& ≈50%

the share of supported teams that reach a registered local entity, in two unrelated national inbound programmes on opposite sides of the East China Sea, working from different definitions in different years. Both stop counting at incorporation. Neither publishes what happens after.

A Korean and a Japanese government inbound programme, self-reported figures — our division and synthesis · Grade EST

+5.3%
+4.4%

revenue and export growth, respectively, from a Singapore market-entry grant — measured against similar non-recipients with firm and year fixed effects. The counter-evidence: the money and the door-opening are measurable, and here they show an effect. It is the ceremony that does not.

MTI Singapore economics team, Economic Survey of Singapore, 3Q2021 · Disclosed-method, government-authored · Grade A

The core finding

The week is measured. The corridor is not — anywhere we looked.

Every corridor program we found — landing pad, innovation house, inbound accelerator centre, soft-landing bridge, acceleration hub, whatever the local brand — publishes numbers from the delegation week or the cohort’s demo day: meetings brokered, consultation value logged at the booth, MOUs signed at the closing ceremony, prize money disbursed. Almost none publishes what those became twelve months later. Where a same-cohort trail does exist, the in-week metric runs roughly an order of magnitude ahead of the paid outcome — a Nordic matchmaking program’s 60% “further discussions” against ~9% paid collaborations is the cleanest published pair we found, in any region. That multiple is a synthesis across sources, not a single survey line — Grade EST.

This is not a quirk of one region’s diplomatic style. A 2026 OECD review of internationalisation-assistance programmes across roughly ten countries — Korea, Japan, Germany, Canada, the UK, Denmark, the Netherlands, Austria, France among them — found a formal evaluation of exactly one, and that evaluation measured process quality and self-reported perceptions, not impact. India and China are not in the review at all — not audited and found wanting, simply absent from the exercise. Two national inbound programs on opposite sides of the East China Sea, working from different definitions in different years, converge on the same number for the one thing both can count administratively: a legal entity formed, in roughly half of all supported teams. Neither publishes what happens after incorporation.

The studies with disclosed samples agree on where success is actually decided, and they did not coordinate with each other. Spence (n=190, UK trade missions): prior knowledge of the market, contact established with the counterpart before the mission, and regular follow-up visits after it. Denmark’s Trade Council data, tested with firm and year fixed effects: direct partner-search and matchmaking support raised small firms’ odds of becoming exporters by nearly a third more than general market-intelligence support did, two years out. Singapore’s own market-readiness grant, tested the same way against non-recipients: a statistically significant lift in revenue and exports — for firms smaller and less export-ready than average to start with. Every one of these studies finds an effect concentrated on the door-opening and the follow-through. None finds an effect concentrated on the week.

The falsifiable claim at the centre of this page: no disclosed-sample study we could find, across corridors researched in Northern Europe, Western Europe, the Middle East, South Asia, East Asia, Southeast Asia and Oceania, attributes corridor conversion to anything that happens during the delegation week itself. Where rigorous evaluation does exist — and it is rare everywhere — it measures the money and the introduction, not the ceremony. The week is the most visible, most funded, most photographed part of every corridor we looked at, on every continent — and the least decisive. If a program can show a cohort whose conversions trace to the week itself, this finding falls. We’d genuinely like to see it.

Where it actually breaks

Not at the meeting. At the flight home — and increasingly, programs know it.

The metric ends at the airport. The agency’s KPI is delivered the moment the cohort disbands: meetings held, consultation value, attendance. Everything after wheels-up belongs to nobody’s scorecard. What gets measured ends exactly where conversion begins.

The MOU is the deliverable — and one government has now said so in writing. Corridor weeks run on a diplomatic clock: something must be signed by Friday. This is no longer inference: one government’s own published definition of programme success lists a signed MOU or commercial agreement as an example of “substantial business and market traction” — the ceremonial artefact, scored as the commercial outcome, in a footnote on an official factsheet.

Nobody local has a mandate. Call it a landing pad, an innovation house, an inbound accelerator centre, a soft-landing bridge — the service bundle is nearly identical everywhere we looked: a desk, a visa pathway, a mentor list, an events calendar. Almost none of it includes a person whose job — with a number attached — is converting this startup’s warm meetings into a paid pilot. The startup flies home; the local corporate’s “let’s continue the conversation” enters a procurement and budget cycle with no one physically present to push it.

The budget-cycle mismatch. A Singapore corporate’s operating budget was locked months before the delegation landed. A first meeting in late October competes for money allocated in Q2. The follow-through stretch isn’t just unowned — it’s mistimed, unless someone plans backwards from the buyer’s fiscal year rather than forwards from the event calendar.

The industry’s own tacit admission. One Asian government’s flagship inbound programme has redesigned itself three times in five years — most recently splitting a single cohort into a three-phase, ten-month pathway, explicitly because, in the programme’s own words, “conventional programs…focus on a single moment of market entry” rather than carrying a startup through one. A neighbouring country’s flagship outbound acceleration programme has made its in-person week optional: eight weeks of virtual work now precede a single, optional week abroad. Neither programme frames this as a concession. Both are behaving as if the delegation week was never where the value sat.

What actually works

The saves sit outside the week — before selection, or in the months after

Same rule as the pilot research: every intervention that shows up in the evidence is a structure set up outside the delegation week — before selection, or in the months after. None is in-week heroics. Two new threads sharpen this without changing its shape. First, the clearest causal comparison available says direct introductions beat information sessions — funding a workshop is not the same intervention as funding a warm intro, and the evidence can now tell them apart. Second, the corridors with the most legible handover all report the same structural choices: a maintained local partner rather than a one-off MOU, selection matched to the firm’s actual stage, and mentors who hold standing on both sides of the corridor.

If you run a corridor program

Select for pre-existing conversations, not applications. The strongest disclosed-sample predictor of mission success is contact established with the counterpart before the trip. Make “name your three target accounts; show the first call happened” a selection criterion, not a nice-to-have.
Fund direct introductions over information sessions. The best available causal comparison found matchmaking support outperforming market-intelligence support by roughly a third, two years out. If your programme’s budget still favours seminars and market reports over paid time for someone to make the actual introduction, the evidence says you have the ratio backwards.
Fund the six months after, not just the flights. Travel grants fund the week. The evidence says conversion needs return visits. Reserve a tranche of every grant for trip two and trip three.
Give someone on the ground a mandate and a number. A named local owner per startup, carrying “paid pilots signed” — not “meetings held” — as their metric. This is the single cheapest structural fix, and almost no corridor has it.
Find your local partner before you find your cohort. The most-cited “what converts” finding across the programmes we reviewed is a maintained partnership in the target market — a local accelerator or corporate, competitively selected and replaced on performance, not a one-time signing. One government housed its outbound centre inside a home-country conglomerate’s overseas subsidiary specifically to inherit an existing sales channel and physical presence on day one.
Publish the follow-up data you already collect. At least one national programme makes a post-programme questionnaire a condition of eligibility — “follow-up questionnaires and business status surveys will be conducted at any time during or after the program” — and has never published a cohort result from it. If you already gather this, publishing it costs nothing but a bad quarter.
Retire consultation value, MOU counts and funds-raised as headline metrics — or publish your own definition, the way one government now has, and let the reader see that a signed MOU is already counted as the outcome.

If you’re the startup landing

Book meetings before you’re selected. Treat the delegation as your second meeting with each target, not your first. If you can’t get a first call from home, the trip won’t fix that.
Fly with a priced pilot, not a deck. The week’s job is finding the person who can sign a purchase order — awareness is a by-product, not the goal.
Don’t sign MOUs you can’t convert. Ask what the programme’s own paperwork counts as success. If a signed MOU already qualifies as “market traction” in someone else’s reporting, it will look identical to a real deal on their dashboard — which tells you nothing about whether you’ve actually landed.
Plan the return visit before the first one. In the disclosed-sample studies, relationships that converted were cultivated through regular visits, not follow-up emails. Budget trip two before you take trip one.
Name who owns this market for the next six months. If the answer is “the CEO, when there’s time,” you are not landing — you are visiting.
Ask what stage the programme actually selects for. Programmes that report the clearest handover tend to select firms already at an advanced stage of internationalisation, and lean on relational support, not training. If you’re still building the product-market fit the programme assumes you already have, the mismatch is structural, not a reflection on you.
Use the landing pad for what it is. Entity setup, visa, desk, mentor introductions: yes. Sales channel: no. The pipeline is owned by a person, not a program.

The citation audit

Six corridor numbers that don’t survive a source check

Same discipline as the pilot research: we tried to trace each widely-circulated number to a primary source and a definition. These six are quoted as outcomes, across four regions. None survives as one.

North America

“The trade missions generated tens of billions of dollars in new business deals”

Claimed by: a national government’s flagship trade-mission programme

Debunked by econometrics

These are announced-deal values — self-reported by participants at the mission, never audited against realized contracts. When the same missions were tested against actual bilateral trade with country-pair fixed effects, the effect was “small, negative, and mainly insignificant.” The tens-of-billions figure is the archetype of every corridor’s press release: it counts the press conference, not the purchase order.

Oceania

“92% of graduated startups continued doing business across Southeast Asia”

Claimed for: one national landing-pad programme’s Singapore pad

Definitional & survivorship games, stacked

“Continued doing business across” a region is compatible with sending emails from the home office — it is not revenue, entity, or contract. The denominator is graduated startups (drop-outs excluded), the figure appears in a promotional “record success” article, and no methodology, sample size or time window is disclosed. A second, independent search pass confirms no national audit office has ever reviewed this programme — one of the most-cited landing-pad brands in the world, never once the subject of a formal performance audit.

Southeast Asia

“More than a quarter of projects achieved substantial business and market traction”

Claimed by: a Southeast Asian government’s national innovation-alliance programme

The definition is published — and that makes it worse

This is not an undefined term; it is a defined one, in a footnote on the government’s own factsheet: “signed MOUs or commercial agreements, embarked on joint ventures, established in-market entity, or received investment from market partners.” A signed MOU counts as market traction. Funding raised also counts. And the denominator quietly switches mid-sentence — 600+ startups supported, more than a quarter of projects achieved traction — with the number of projects never disclosed. The 25% figure that circulates is a projects-rate read as a startups-rate, and it cannot be recomputed from the published text.

East Asia

“226 local corporations established, 425 startup visas issued” — and, elsewhere, “195 corporations, 364 visas”

Claimed by: an East Asian government’s flagship inbound accelerator, in two undated releases

Two incompatible outcome series, live at once

Both figures are attributed to the same decade-long programme; neither release states its vintage, so both circulate as current. The gap (+31 corporations, +61 visas) is plausible as a single year’s growth — but with no dated primary source for either number, a reader cannot tell whether they are looking at last year’s total or this year’s. And “corporation established” is an administratively verifiable event, not a claim about revenue, a customer, or survival — the closest thing to an honest funnel in this batch of research still stops at incorporation.

Western Europe

“1,800+ companies. $17B+ raised by participants” — also cited as “500+/$12B,” “1,200+/$16.7B,” and, elsewhere, “290 completers, $4.9B raised”

Claimed by: a European national accelerator, across its own site and ecosystem press, over time

The wrong variable, four ways

Funding raised by alumni measures the startups’ fundraising, not whether the programme’s market entries converted — a startup can raise $50M at home having sold nothing in its target market. A fourth, independent figure surfaced in a later research pass (290 completers, $4.9bn raised) neither matches nor reconciles any of the programme’s own three published counters. The number that would actually answer the corridor question — what share of participants generate revenue in the target market within 24 months — is published by no major national accelerator we could find, anywhere. The silence is the finding.

South Asia

“Targets 500 corporate–startup partnerships and over ₹250 crore of corporate investment over three years”

Claimed by: a South Asian state government’s new corporate-convergence programme

A target, already being relayed as a result

This is a three-year target, announced in 2025, already appearing in headline form across multiple outlets without the word “target” attached. It is recorded here, before the fact, specifically so that when it resurfaces in 2027 phrased as an achievement, the original framing is on record. This is the predictable next entry on this list — and a template for what to watch for in every corridor that publishes a target before it publishes a result.

A note on one figure we chose not to feature, and why. A separate research pass found that the original registration domain named in one flagship inbound programme’s own 2024 launch announcement — live in that release at the time — now resolves to an unrelated commercial site with no connection to the programme; we independently re-verified this in August 2026 and it still resolves that way. We are not treating this as a statistic and it does not appear as an audit card above, because a lapsed domain is not itself an outcome claim. But per our own standard of not silently dropping a caveat, it is recorded here, in category terms, as exactly the kind of second-hand, unverifiable detail that should never be cited as evidence of anything except how much institutional attention a corridor receives once a cohort ends.

Running a corridor into Singapore? I’m on the receiving end.

I sit on the Singapore side of several of these corridors — German, Swedish, Korean, Australian and Indian startups land in my inbox the week after their delegations end, usually with a folder of warm contacts and no owner for what happens next. I mentor at BLOCK71, NUS GRIP and ATUM Ventures, so I also see which landings convert and which quietly stop replying to their corporate contacts by month three.

If you run a delegation, a landing pad or a corridor program — anywhere in the world — or you’re a startup about to land — I’d like to compare notes. Especially if your data contradicts this page; that’s the most useful conversation of all. Nothing to sell in that call.

Book a 30-min conversation Connect on LinkedIn

Method & honesty note

How this was put together

Corridors researched: Northern Europe (Sweden, Finland, Denmark, Norway), Western Europe (Germany, France), the Middle East (UAE, Saudi Arabia, Qatar), South Asia (India, national and state level), East Asia (Korea, Japan), Southeast Asia (Singapore as receiving hub, intra-ASEAN structures) and Oceania (Australia). Every statistic on this page was traced back toward its primary source and graded: A = peer-reviewed study or disclosed-sample primary research · B = industry-published or self-reported data, sample not independently verified · EST = reasoned estimate, labelled as such. Where a number could not be traced, it is labelled as such or was cut. One source (a CEPR/VoxEU survey column summarising several of the peer-reviewed studies below) is a researcher-authored policy commentary, not itself a peer-reviewed publication — it is graded B and used only as a pointer; the underlying journal studies are cited and graded independently on their own terms. Body text describes source categories rather than naming organisations; the full named citations sit in the appendix below.

A correction, stated plainly. An earlier version of this page claimed Singapore’s “substantial business and market traction” language was never defined. That was wrong. It is defined, in a footnote on the government’s own factsheet, and the definition includes signed MOUs and investment received. We have corrected this throughout the page. The correction makes the underlying point about ceremonial metrics stronger, not weaker, and we would rather print the correction than quietly fix it.

Four limits worth stating plainly. First, most corridor outcome numbers are self-reported by the body that runs the corridor — and this holds across every region researched, not just the original Nordic/German/Korean set. The independent, disclosed-method measurements on this page are not uniformly unflattering, though: alongside the two peer-reviewed academic studies and the Swedish state’s counterfactual evaluations, Singapore’s own market-readiness grant evaluation is a disclosed-method government study with a genuinely favourable result. What is consistent is not the verdict — it is the rarity. Second, the order-of-magnitude gap and the ~50% cross-corridor convergence are syntheses, not measurements (Grade EST); other sources corroborate direction, not magnitude. Third, where an independent evaluation of a named programme genuinely does not exist — which is most of them, in every region — that absence is treated on this page as a finding, not a gap to apologise for. A 2026 OECD chapter written specifically to survey this field found a formal evaluation of one programme in roughly ten; that is the state of the field, not a limitation of this research. Fourth, the author sells into this problem: Arjun operates Singapore-side landing and conversion work for exactly these corridors. The incentive cuts the obvious way; the mitigations are the ones on this page — graded sources, published definitions, a falsifiable core claim, a standing invitation to programs whose data contradicts it, and printed corrections when we get something wrong.

The full limits file — eight notes, unabridged

Most corridor outcome numbers are self-reported by the body that runs the corridor. This held for the original Nordic/German/Korean research and holds without exception across the expanded set: matchmaking programmes count their own collaborations, national agencies grade their own alliances, ministries report their own trade-show consultations, landing-pad agencies narrate their own pads. Across the full set of programmes researched, independent, disclosed-method evaluation was found for: the peer-reviewed academic literature (below), the Swedish state agency’s evaluations, and Singapore’s own MTI-authored grant study. That is roughly three sources of independent verification against dozens of self-reported programmes across nine-plus governments. The rarity is the finding, not any single result.
Survivorship bias is structural, not incidental, and it recurs on every continent. “92% of graduated startups,” alumni funding totals, and success-story pages all condition on survival. Startups that landed, stalled, and went home quietly appear in no denominator anywhere — from Sydney to Seoul to Abu Dhabi. Our own inbound-corridor anecdotes share this bias: we disproportionately meet the startups still trying.
Definitional games do the heavy lifting, and one government has now put the game in writing. “Market entry” (a legal entity? a first invoice?), “continued doing business” (an email thread?), “substantial business and market traction” (defined — and a signed MOU qualifies), “contract” (including MOUs?), “consultation value” (declared intent, priced at peak optimism), “beneficiary” (never defined on the page that publishes it). Across every programme audited, the definition softens exactly where the conversion evidence would be, or in one case, is published loosely enough to include the ceremonial artefact as the outcome. When a term is undefined, we read it at its weakest plausible meaning and say so; when it is defined, as with Singapore’s footnote, we quote the definition rather than assume one.
The academic anchors don’t map perfectly onto startup corridors, and we now have more of them, not fewer. Head & Ries measure country-level goods trade around prime-ministerial missions; Spence studies 1990s UK SME missions; the Swedish and Singaporean state evaluations assess grant and export-promotion schemes generally; the Danish, Belgian and Egyptian firm-level studies measure export-promotion and market-access effects on goods-exporting SMEs, not venture-backed software startups moving into a delegation-style programme specifically. None is a study of venture-backed startup delegations in the exact form this page describes. We use them as the best available disciplined evidence on neighbouring mechanisms — brokered introductions, market access, grant support — and flag the extrapolation rather than hide it.
The order-of-magnitude gap and the ~50% cross-corridor convergence are syntheses, not measurements (Grade EST). The cleanest same-source pair for the first is the Nordic programme’s 60% further-discussions vs ~9% paid; the second combines two unrelated national inbound programmes’ incorporation counts under different definitions and years. Both are labelled EST wherever they appear. If a corridor publishes a full funnel that narrows either multiple, this page should be updated — and we’d count that publication as a win for the field.
Genuine absence of evaluation is reported here as a finding, not a gap. No formal national-audit-office, ministry-commissioned, or academic evaluation was found for the large majority of programmes researched across every region — confirmed, in several cases, on a second independent search pass specifically looking for one. Where we searched and found nothing, we say so and treat the silence as evidentiary, consistent with the core claim of this page.
Second-hand links in the chain are flagged at the point of use, not silently dropped. Some figures on this page (an Asian ministry’s trade-show consultation data; the Swedish evaluation summaries) reach us via press relay or agency summary rather than a full primary read, and are confined to the citation audit or flagged accordingly. One additional, non-statistical detail — a lapsed programme domain now resolving to an unrelated commercial site — was independently re-verified before publication and is recorded in the citation audit section as an explicit example of an unverifiable, second-hand chain, not used as evidence of anything beyond itself.
The author sells into this problem. Arjun operates Singapore-side landing and conversion work for exactly these corridors. The incentive cuts the obvious way. The mitigations are the ones on this page: graded sources, published definitions, a falsifiable core claim, a printed correction when the page itself got something wrong, and a standing invitation to programs whose data contradicts it.

Appendix

Every number on this page, with source, grade and provenance

Sources appendix — 33 entries, plus the searches that found nothing

Named sources appear here so every number can be checked; the body text deliberately describes categories rather than singling out organisations.

1 · 8,000+ curated matchmaking meetings; 715+ (“more than 700”) paid collaborations (~9%); “six out of ten leading to further discussions”; follow-up rate 45–50%; 3,000+ startups since 2017 Grade B
Ignite Sweden · Ignite Nordic

Program self-published running totals, 2017–2025. Disclosed counts but no external audit; “paid collaboration” includes paid pilots, not only production contracts. The 9% is our division of their two published numbers.

2 · Canadian trade missions: effects “small, negative, and mainly insignificant” with country-pair fixed effects; government claimed tens of billions in new deals from same missions Grade A
Head, K. & Ries, J., “Do trade missions increase trade?”, Canadian Journal of Economics 43(3), 2010 (PDF) · SSRN

Peer-reviewed econometric study, bilateral trade data, Canada 1994+ Team Canada missions. Measures country-level goods trade, not startup deals.

3 · Singapore GIA: 600+ tech startups supported since 2019; “more than a quarter of projects achieved substantial business and market traction,” defined in footnote 2 as “signed MOUs or commercial agreements, embarked on joint ventures, established in-market entity or received investment from market partners”; 24 innovation hubs; GIA Discovery (2–5 weeks) launched 2025 for Tokyo and Mumbai Grade B
MTI Singapore, “Factsheet on Enhancements to Global Innovation Alliance” (COS 2025) · EnterpriseSG media release MR00625, March 2025

Government self-report. Definition is published (footnote 2 of the source PDF), correcting an earlier version of this page. Denominator switches from “startups” (600+) to “projects” (count never given) mid-statement — the circulating 25% figure cannot be recomputed from the published text. GIA is outbound (SG startups going abroad).

4 · Trade-mission success determinants: market diversification strategy, prior market knowledge, communication with partners established before the mission, post-mission cultivation incl. regular visits; n=190, logit model Grade A
Spence, M., “Evaluating Export Promotion Programmes: U.K. Overseas Trade Missions and Export Performance”, Small Business Economics 20, 2003 · open PDF at repository.mdx.ac.uk

Peer-reviewed, disclosed sample (190 UK mission participants). UK SMEs, 1990s-era missions — extrapolation to startup delegations is an inference, flagged in the limits file.

5 · Swedish export promotion: firms without exports more likely to start exporting after contact; no or marginal effects for experienced exporters; counterfactual method Grade A (caveat)
Tillväxtanalys (Swedish Agency for Growth Policy Analysis), PM 2020:15 “Effektutvärdering av Sveriges exportfrämjande”

State evaluation agency, counterfactual design. Report in Swedish; findings taken from the agency’s own published summary.

6 · NSID 2026 delegation: Singapore, 26–30 Oct 2026; a Nordic agency travel grant of SEK 25,000 (outside-Europe rate); multi-agency organisers; program runs during a major Singapore innovation trade event Grade A (event facts)
vinnova.se event page · sisp.se project page · ignitenordic.org event page

Official agency event pages, cross-confirmed on three organiser sites.

7 · Korea inbound accelerator, decade totals: 449 international teams supported → 226 local corporations (~50.3%, our division) → 425 startup visas; earlier-vintage series for the same programme: 195 corporations / 364 visas, undated; 21,537 applicants since 2016; KSC Singapore 2025: 29 startups, KRW 6.6bn overseas revenue, KRW 69.5bn investment, 106 jobs Grade B
KoreaTechDesk programme coverage · Businesswire, 2026 programme release

Ministry/agency self-report relayed through trade press and official releases; two non-reconciling outcome series in simultaneous circulation, undated. “Corporation established” and “visa issued” are administratively verifiable but not revenue, customer, or survival measures.

8 · Austrade Landing Pads: five hubs, 90-day residency model, no travel/living-cost funding; 340+ businesses since 2016; Singapore pad: 92% of graduated startups “continued doing business across Southeast Asia”; no ANAO or departmental performance audit located across two independent search passes Grade B (outcomes) / A (audit-absence, verifiable)
services-exports.gov.au Landing Pads program page · anao.gov.au, publications index

Government program self-report + promotional press for outcomes. The absence of a national-audit-office review is itself confirmed independently and is the stronger finding.

9 · German Accelerator: homepage counters 1,800+ companies / $17B+ raised; other own/ecosystem surfaces: 500+/$12B, 1,200+/$16.7B, and a fourth figure of 290 completers/$4.9B raised (Munich Startup); no published market-entry conversion metric across two search passes; one OECD-cited external evaluation (Ramboll, 2017 & 2019, for the German federal ministry) found process quality and self-reported perception outcomes, not counterfactual impact Grade B
germanaccelerator.com · munich-startup.de

Vendor self-published marketing counters; internal inconsistency across four non-reconciling figures. The Ramboll evaluation is cited by the OECD (entry 30 below) but the underlying reports were not independently located.

10 · BLOCK71 (NUS Enterprise): 11 cities; 1,152 startups housed in its first decade (747 Singapore, 449 overseas); 100+ VC/corporate/research partners; startups raised over US$2bn in venture funding; bidirectional design (Nagoya explicitly built to bring Southeast Asian startups into Japan) Grade B
block71.co

Institution self-published counts; disclosed splits, no external audit. Funding-raised is a context metric, not conversion evidence.

11 · EIC: 6% of 1,500+ corporate–startup engagements at 120+ corporates reach a signed deal EST
European Innovation Council headline counts, 2025 — a computed ratio, not a published rate; as graded on arcshift.ventures/beyond-the-pilot

Carried over from the Beyond the Pilot appendix (disclosed sample).

12 · “In-week metrics run roughly an order of magnitude ahead of paid outcomes” EST
This report’s synthesis (principally source #1, corroborated directionally by #2, #3, #7)

Reasoned estimate across heterogeneous sources, labelled EST wherever it appears. Not a survey line.

13 · Denmark: export support raises exporting probability +3.9pp (year 1) → +5.9pp (year 2); effects concentrate on small firms and first-time exporters; partner-search/matchmaking support: +9.4pp for small firms after two years vs +6.7pp for intelligence-and-analysis support Grade A
Munch, J. & Schaur, G., “The Effect of Export Promotion on Firm-Level Performance”, American Economic Journal: Economic Policy 10(1), 2018

Peer-reviewed, Danish Trade Council administrative data, diff-in-diff with propensity matching.

14 · Belgium (Flanders): export support raises exporting probability +8.5pp on average; subsidy-type support (funded trips, trade fairs, subsidised meetings) shows 4.6–8.4pp higher probability than inquiry/communication-only support Grade A
Broocks, A. & Van Biesebroeck, J., “The Impact of Export Promotion on Export Market Entry”, Journal of International Economics 107, 2017

Peer-reviewed, Flanders Investment and Trade administrative data.

15 · Meta-regression of 479 published estimates across 19 countries (studies to 2020): export-promotion effects on market entry are systematically exaggerated in the literature; the bias traces to researchers not writing up null/negative results, not to journal rejection Grade A
Demana, B.A., “Publication bias in export promotion impact on export market entry: evidence from a meta-regression analysis”, Applied Economics Letters, 2024

Peer-reviewed meta-analysis, disclosed sample of 479 estimates, method disclosed. A meta-finding about the evidence base itself, not about any single programme.

16 · Egyptian rug producers randomly given export-market access: 16–26% higher profits; improved production quality with no increase in output per hour (“learning-by-exporting”) Grade A
Atkin, D., Khandelwal, A. & Osman, A., “Exporting and Firm Performance: Evidence from a Randomized Experiment”, Quarterly Journal of Economics 132(2), 2017

Randomized controlled trial — the cleanest causal design in this file. Different mechanism from a delegation trip (sustained buyer exposure vs a one-week visit); single sector, single country; flagged as an import from a different context, as with Spence and Head & Ries.

17 · Sweden vs. Denmark/Finland/Norway/Netherlands/UK/Germany, comparative review: most Swedish export-promotion impact evaluations are conducted by the actors themselves, “only a few by external actors”; Norway and the UK have used randomized controlled trials for this purpose, Sweden had not; Netherlands/UK found only already-exporting SMEs positively affected by some efforts Grade A
Tillväxtanalys, comparative PM, “Sveriges exportfrämjande jämfört med andra länder i Europa”

Government evaluation-agency self-critique of the evaluation ecosystem across seven countries, not of one programme. The Netherlands/UK finding is a genuine counter-note to this page’s “first contact matters most” framing and is stated as such.

18 · Vinnova’s “Innovativa Startups” grant programme evaluation flags “a need for knowledge about internationalization in the early stages” as an identified gap Grade A
Tillväxtanalys, Rapport 2025:12

National evaluation body stating, of its own accord, that internationalisation outcomes for this cohort are not well understood.

19 · A Nordic deep-tech accelerator disclosed €6 million across 15 startups (~€400K/startup) in its 2025 call, explicitly to “accelerate market entry”; no output/conversion metric published for the same cohort. A separate Nordic landing-pad network (Singapore/Hong Kong house) was funded at NOK 10.8 million over three years, with zero outcome numbers found published anywhere for the Singapore house specifically Grade B
businessfinland.fi · nordicinnovation.org

Input/grant figures only, both cases. The Singapore-house outcome silence is a genuine absence, confirmed by targeted search of the operator’s own site and the funder’s programme pages.

20 · Hub71 (Abu Dhabi): $2.7bn cumulative funding raised, $1.5bn cumulative revenue, $244m in signed corporate deals (2022–2025), 390+ startup community; Access Programme structure includes AED 250k non-cash incentives + AED 250k cash-for-equity + AED 250k performance top-up, with a disclosed founder-relocation requirement; no national-audit-authority evaluation located Grade B
hub71.com press release · economymiddleeast.com

Self-reported “Impact Report,” but notably better-denominated than most programmes in this file — revenue and signed-deal figures are disclosed, not only funding raised. Still no per-cohort breakdown or external audit.

21 · A Gulf government-entity-matching accelerator: 250+ startups from 37 countries since 2017, from 600+ applicants across 70+ countries; solutions described as “now in daily use across the emirate” with no count, percentage or named example given Grade B
dubaifuture.ae

Programme self-report; the operational-use claim is unquantified and unverifiable as published.

22 · A Gulf national technology-development programme: SAR 2.5bn (2020 launch) + a further $430m across six initiatives incl. $150m venture debt; a partner accelerator’s latest cohort selected 8 startups from 690+ applicants (~1.2%); a separate science-park programme reports 400+ companies hosted from 39 countries Grade B
techinafrica.com · wamda.com

Committed/budget and selectivity figures only; zero disclosed outcome metrics found for any of these structures.

23 · La French Tech aggregate (2025 cohort): €10bn cumulative revenue, 42,000 jobs, 93% “international presence” (undefined); French Tech Visa: 1,200+ companies hired through the programme since 2017, 12,000+ jobs created (a hiring metric, not a market-entry metric) Grade B
Sramana Mitra roundup · welcometofrance.com

“International presence” is undefined and appears only in secondary ecosystem-roundup coverage, not a located primary government report.

24 · A French public investment bank’s immersion-bootcamp programme: structure well-documented (bootcamp + market immersion trip), no outcome/conversion numbers found; a secondary press paraphrase claims “auditors found encouraging initial results” for the bank’s broader decade of activity, primary report not located B / UNV
Sifted, “Bpifrance: Inside the machine powering French tech’s rise” · bebeez.eu

The “auditors” claim is a secondary-press paraphrase with no locatable primary audit document; flagged as unverified, not cited as evidence of independent evaluation.

25 · Singapore Market Readiness Assistance grant: +5.3% revenue, +4.4% total exports per additional grant of average amount vs. similar non-recipients (statistically significant); recipients smaller and less export-oriented than non-recipients (avg. revenue S$5.8m vs S$23.8m) Grade A
Toh, B., Koh, W.J. & Feng, A., “Impact of Enterprise Singapore’s Grants on Firms’ Revenue and Exports”, Economic Survey of Singapore, 3Q2021, Ministry of Trade and Industry

Disclosed-method government study: firm-level administrative data, firm and year fixed effects, controls for other grants received, non-recipient comparison group. Authored by MTI economists evaluating an MTI-family agency’s own grants — disclosed-method, not third-party. Measures grant amount, not programme participation; period 2013–2018 predates the GIA landing-pad programme entirely.

26 · Japan Startup Visa: 716+ individuals granted residence status as of May 2024; at least 359 transitioned to or renewed “Business Manager” status (≈50% floor, our division); a national outbound acceleration programme requires participant cooperation with post-programme “follow-up questionnaires and business status surveys…at any time during or after the program” as a condition of eligibility, with no cohort result ever published Grade B (outcomes) / A (programme facts, official page)
jetro.go.jp, Japan Insight · jetro.go.jp, acceleration programme page

Immigration-status change is the hardest definition available anywhere in this research — a registered, capitalised, office-holding business, though still not revenue or survival. The follow-up-collected-but-unpublished finding is a documented eligibility condition, not an inference.

27 · A metropolitan government’s outbound/inbound accelerator: “supported 50 startups so far, successfully raising over ¥20 billion in funding” Grade B
x-hub-tokyo.metro.tokyo.lg.jp

Funding-raised metric — the same class already audited for a European national accelerator, confirmed as the house style of a second, unrelated public programme.

28 · A national bilateral-bridge programme: 20 bilateral bridges since 2017, 5 multilateral engagements, 25 further active countries, “more than 850 startup beneficiaries and counting” (“beneficiary” undefined; page footer dated three years prior to fetch) Grade B
startupindia.gov.in

Only quantity on the page is undefined and the page itself may be stale; testimonials describe leads and introductions, not contracts or market entries. ≈42 beneficiaries/bridge over ~9 years (our division).

29 · An inbound market-access bridge (South Asia): explicit no-equity, participant-paid structure; own marketing copy states “now closing your deals is all in your hands”; page written in past tense, suggesting the programme is dormant Grade B
t-hub.co

The clearest first-person admission of the corridor handover found in this research, stated as a feature in the operator’s own copy. No cohort outcomes ever published.

30 · A 2026 institutional review of internationalisation-assistance programmes across roughly ten countries: formal evaluation reported for exactly one programme (process quality and self-reported perception outcomes, no counterfactual); no outcome data for the rest; India and China absent from the review entirely; “what converts” qualitative findings: local partners in the target market (stated three times independently), sector specialisation with matched local partners, stage-gating to advanced firms, bicultural mentor networks, home-country corporates as landing infrastructure Grade A (institutional publication)
OECD, Incubation in Entrepreneurial Ecosystems: Hatching Growth, Ch. 6 “Internationalisation assistance”, OECD Publishing, 2026, DOI 10.1787/21be9d03-en

No survey statistics in this chapter — every figure is a programme-level descriptive count. The “1 of ~10” is our count of the chapter’s contents, stated as such, not an OECD claim.

31 · Qualitative case study, 4 incubators + 4 born-global startups: entrepreneur, networks, mentoring and credibility identified as influencing factors in internationalisation; “this influence is perceived differently depending on the company’s circumstances” Grade A (qualitative, n=8)
Pinto, M. & Rua, O., “Incubators’ practices influence in the born global startup’s internationalization process”, Journal of Open Innovation, 2023

Peer-reviewed but small-sample and qualitative: supports a mechanism claim (networks and mentoring matter), cannot support any rate claim.

32 · A regional startup-registry platform reports 11,000+ startups and 5,000 investors “connected” since a 2025 launch; a China–ASEAN innovation-base network announced its first four demonstration sites (incl. one in Singapore) with no participant, budget or outcome figures disclosed; a Chinese municipal soft-landing scheme offers a ¥40m+ prize pool and dedicated visa/registration assistance, with no published outcome of any kind Grade B
startup-asean.org · english.www.gov.cn

Registry/account and announcement metrics, not corridor outcomes. Intra-ASEAN and China→SEA structures in this research were consistently the best-resourced and least-measured of any region: prize pools, service centres and dedicated hubs, and not one published participant, entity or revenue figure.

33 · CEPR/VoxEU survey column summarising the Danish, Belgian and Colombian firm-level literature (entries 13, 14 above, plus a bundled-support finding from Colombian export-promotion data) Grade B (commentary, not primary)
Makioka, R., “The effectiveness of export promotion measures: A survey”, VoxEU/CEPR, 2019

A CEPR policy column, not a peer-reviewed journal article. Used only as a pointer to the underlying peer-reviewed studies, which are cited and graded independently above; not treated as independent Grade-A evidence in its own right.

Search-and-found-nothing notes (negative provenance, kept honestly)

No published follow-up study tracking a specific startup-delegation cohort’s meetings→deals conversion at 12 months was found for any programme researched, in any region. This absence is load-bearing for the core finding and is stated as an absence, not spun as a number.
No formal national-audit-office (Australia’s ANAO, Singapore’s Auditor-General’s Office, Korea’s Board of Audit and Inspection, France’s Cour des comptes, Germany’s Bundesrechnungshof, or equivalents in the UAE and Saudi Arabia) evaluation of a named landing-pad, accelerator or soft-landing programme was located, with the single exception of the German Accelerator’s ministry-commissioned Ramboll reviews (entry 9/30 above).
No study, anywhere in this research, isolates what makes a landing convert — i.e., no quantitative work tests pre-existing local customer relationships, founder relocation, local hire, or follow-up cadence as competing predictors of corridor conversion. The OECD chapter lists determinants qualitatively; Spence remains the only disclosed-sample quantitative treatment of the neighbouring trade-mission question. This gap is genuine and is the strongest thing this page can claim to be first to name.
No intra-ASEAN corridor programme with published cohort outcomes was found; ASEAN-level structures are policy frameworks and registries, and national soft-landing schemes at the member-state level publish nothing.

Delegations don’t fail.They land — and then fly home.