Lessons from the 2012 gTLD Round

## The 2012 Round in Numbers The first new gTLD application round accepted by ICANN ran from January to May 2012. When the window closed, ICANN had received: - **1,930 applications** from entities in 60 countries - Covering **1,409 unique strings** - With **751 string contention sets** (multiple applicants for the same string) - Total application fee revenue: approximately **$357 million** By the end of 2015, 1,000+ new gTLDs had been delegated. By 2020, almost all applications had reached final resolution — either delegation, withdrawal, or rejection. The 2026 round builds directly on what was learned from that experience. Use the TLD Comparison Tool to explore the performance data for new TLDs that launched from the 2012 round. ## What Went Well ### The Domain Ecosystem Expanded Dramatically The most straightforward success: the internet's namespace genuinely expanded. Extensions like .io (repurposed from British Indian Ocean Territory into a tech startup favourite), .co (Colombia's ccTLD, marketed globally), .app, .dev, .blog, .shop, and .xyz became commercially significant. Millions of businesses chose new gTLD (Generic Top-Level Domain) extensions for their primary web presence. ### Brand TLDs Delivered Control and Security Companies that applied for and received their own Brand TLD (.brand) — Google (.google, .app, .dev), Amazon (.amazon, .audible), BMW (.bmw), Barclays (.barclays), and hundreds of others — gained a controlled namespace free from cybersquatting. Google's use of .app as a trusted, HTTPS-enforced consumer extension demonstrated how a brand TLD could become a genuine public utility. ### Geographic TLDs Built Real Communities City TLDs like .nyc, .london, .berlin, and .tokyo developed genuine local identity. New York City's .nyc extension, for instance, grew to hundreds of thousands of registrations and was used by city government, local businesses, and community organisations. These TLDs demonstrated that geographic strings could build real cultural value — not just domain inventory. ### The Process Worked (Eventually) Despite its flaws, the 2012 round ultimately worked. Registries were delegated, contracted with ICANN, and began selling domain registrations. The root zone infrastructure — managed by IANA — handled the expansion without technical failures. The DNS (DNS (Domain Name System)) proved extensible. ## What Went Wrong ### The Timeline Was a Disaster ICANN projected the evaluation process would take 20 months from application close. It took over four years for many applicants. Some received their delegated TLD (Top-Level Domain) in 2014; others waited until 2016. This created: - Cash flow crises for applicants who had budgeted for a 2-year process - Bankruptcy and withdrawal among undercapitalised applicants - Market confusion as the application system's TAS (TLD Application System) crashed repeatedly during submission The root cause: ICANN severely underestimated the complexity of evaluating nearly 2,000 applications simultaneously. The 2026 Applicant Guidebook includes specific reforms designed to address this. ### String Contention Auctions Created Perverse Outcomes When multiple applicants competed for the same string, last-resort auctions produced enormous payouts — to ICANN, not the losing applicants. The money raised (estimated at over $200 million from contention auctions) sat in an ICANN escrow account for years while the community debated how to distribute it. The AGB (Auction Proceeds Governance) process — completed in 2023 — established that future auction proceeds will fund public interest projects, but the 2012 funds remained contested. More troubling: some applicants with genuine community or business rationale lost strings to well-capitalised financial bidders. .radio was a particularly debated case — a community string acquired through auction by a commercial entity. ### Some Registries Failed Completely Not every delegated TLD (Top-Level Domain) became a viable business. Several registry operators went bankrupt or abandoned their TLDs, triggering ICANN's EBERO provisions to maintain continuity. Notable failures included: - **.kiwi**: The registry experienced financial difficulties and was sold, disrupting registrant plans. - **.africa**: A long-running legal dispute between two applicants delayed delegation by years and cost enormous legal fees on both sides. - Multiple small, undercapitalised registries that had fewer than 1,000 registrations and could not achieve financial sustainability. The lesson: a viable TLD business requires a clear path to registration volume. Without sufficient registrations, the fixed costs of ICANN fees, back-end operations, and compliance crush the business. ### The Objection Process Was Inconsistent ICANN's formal objection panels produced some controversial decisions. The limited public interest objection in particular was applied inconsistently, and some objection outcomes appeared to favour large, well-resourced incumbents over smaller community applicants. The 2026 round includes reformed objection procedures intended to improve consistency. ### GAC Advice Was Sometimes Ignored, Sometimes Over-Weighted The Governmental Advisory Committee issued advice against several applications — particularly geographic strings. ICANN's handling of GAC advice was sometimes seen as inconsistent: in some cases, GAC early warnings were effectively dispositive; in others, applications proceeded despite opposition. See Objection Grounds and GAC Early Warnings for the reformed framework for the 2026 round. ### The .amazon Saga Perhaps the most prominent failure of the 2012 round was the decades-long battle over the .amazon string. Amazon.com applied for it as a Brand TLD (.brand). Brazil and Peru objected — they argued that "Amazon" refers to the Amazon River basin and the Amazon region, which spans their territories. The GAC issued advice against delegation. ICANN's Board declined to follow the advice, then reversed course, then was overruled by the ICANN Independent Review Process (IRP). The dispute was not resolved until 2022 — a decade after the application was filed. The lesson for 2026: geographic strings — even those that correspond to well-known brands — face extraordinary risks when governments object. ## What Changed for 2026 ICANN's SubPro policy working group spent five years addressing the 2012 failures: | 2012 Problem | 2026 Reform | |-------------|------------| | Unpredictable timeline | Published target durations; quarterly progress reporting | | Auction proceeds controversy | Pre-committed policy: proceeds to public benefit | | Inconsistent objection decisions | Revised objection standards and panel training | | GAC advice ambiguity | Clearer framework for when GAC advice is followed | | Registry failure / EBERO gaps | Strengthened EBERO requirements; enhanced financial capacity thresholds | | Developing economy exclusion | Dedicated support programme with fee reductions | | Application system failures | Rebuilt TAS with load testing | ## Key Takeaways for 2026 Applicants 1. **Budget for the worst case, not the best case.** The 2012 round took twice as long and cost twice as much as most applicants projected. 2. **Choose your back-end operator carefully.** Several 2012 registries failed in part because their back-end operator did not provide adequate support. Vet operators rigorously. 3. **Avoid contested generic strings unless you have deep pockets.** .app cost $25 million at auction. .blog cost $19 million. Entering a contention set for a popular generic string without $10M+ in reserve is not advisable. 4. **Community TLDs require genuine community.** ICANN's evaluation panels are sophisticated enough to identify manufactured community support. 5. **Plan for a 3–4 year process, even with ICANN's reforms.** Business plans that require revenue by Year 2 to survive are fragile. 6. **Geographic strings carry political risk.** Government opposition is difficult to predict and hard to overcome. See ICANN New gTLD Program 2026: Complete Overview for the full context of the 2026 round, and Post-Delegation: Running a TLD Registry for what successful 2012 registries learned about long-term operations. ## The Registries That Thrived: Common Factors Looking across the 2012 round cohort, the registries that achieved sustainable scale shared several characteristics that applicants for the 2026 round can deliberately replicate: ### Clear Market Positioning TLDs with a specific, identifiable audience outperformed those that tried to be generic alternatives to .com. .io succeeded because developers adopted it as a cultural identifier. .shop succeeded because it clearly signalled e-commerce intent to both registrants and their customers. Vague positioning — "a better .com" — almost universally failed. ### Registrar Investment The most successful new gTLD (Generic Top-Level Domain) registries invested heavily in Domain Registrar relationships. This meant not just signing Registry-Registrar Agreements but actively marketing to registrars, offering meaningful incentives, and working with registrar product teams to ensure the TLD was prominently featured in search results and recommendation flows. ### Premium Domain (Registry Premium) Programs Many thriving registries reserved their highest-value second-level domains — short strings, common words, geographic identifiers — as premium domains priced at $1,000 to $50,000+ per year. This generated significant revenue from a small number of registrations and established brand value for the extension. ### Patience and Capital The registries that survived long enough to build volume were the ones with sufficient capital to weather the slow initial adoption period. New TLDs typically experience a "hype-then-trough-then-growth" adoption curve: a burst of registrations at launch, followed by a quieter period as the novelty fades, followed by organic growth as the TLD builds genuine utility. ## What the 2026 Round Will Look Like Differently The 2026 round applicant pool will include many veterans of the 2012 process who have operated registries for over a decade. These applicants bring operational experience, established registrar relationships, and a realistic understanding of the economics. This sophistication means the average quality of 2026 applications will likely be higher than 2012 — and the bar for evaluation will reflect that. New applicants entering without 2012 experience should study the public records of successful 2012 registries — their registry agreements, their abuse reports, and their registration data — to understand what a well-run registry looks like in practice before committing to the significant financial and operational investment the 2026 round demands.