{"@context":"https://schema.org","@type":"BlogPosting","headline":"ChatGPT 5.3 Halved Its Hotel Sources — March 5, 2026 Cutover","description":"GPT-5.3 rollout cut URLs/answer 49% (24→12). Wikipedia coverage 31%→4%; Reddit 14%→2%. TripAdvisor and Expedia actually gained share vs halving.","datePublished":"2026-04-17","dateModified":"2026-04-17","url":"https://nicolassitter.com/research/chatgpt-hotel-source-shift-2026","category":"research","keywords":["ChatGPT 5.3","GPT-5.3 hotel","ChatGPT source shift","Stockholm listicle"],"articleSection":"Research","wordCount":6800,"readTime":"27 min","articleBody":"April 2026GPT-5.3 · Source Shift\n\n# ChatGPT 5.3 Halved Its Hotel SourcesWhat changed on March 5, 2026\n\n**TL;DR:** Across 9,722 hotel responses from 4 country locales, the day ChatGPT's UI switched to GPT-5.3 is a cliff. URLs per answer fell 49% (24 → 12), unique domains per answer fell 46% (18 → 10), and the **pool of domains per prompt fell 54%** (132 → 61). Answer coverage collapsed for Wikipedia (31% → 4%), Booking (44% → 20%), Expedia (56% → 33%) and Reddit (14% → 2%). A Stockholm-operated network of ~17 Booking-affiliate listicle sites gained the freed-up citation slots.\n\nNS\n\nNicolas Sitter\n\nPublished April 17, 2026\n\n140\n\nWorld prompts\n\n24 → 12\n\nURLs / answer\n\n132 → 61\n\nDomain pool / prompt\n\n18 → 10\n\nDomains / answer\n\n[Read the Report](#key-findings)\n\n[Key Findings](#key-findings)[1\\. The Halving](#the-halving)[2\\. By Country](#by-country)[3\\. Category Shift](#category-shift)[4\\. Winners & Losers](#winners-losers)[5\\. Stockholm Listicle Syndrome](#listicle-network)[6\\. By Query Type](#by-tier)[7\\. By City](#by-city)[8\\. Timeline](#timeline)[Why this happened](#why)[Methodology](#methodology)[FAQ](#faq)\n\n## Executive Summary\n\nA model change, not a geography rollout.\n\nSince late December 2025 we've run the same 140 world-wide hotel discovery prompts on ChatGPT's UI every day, from 4 country locales (US, GB, DE, ES). On 2026-03-05 OpenAI quietly switched the UI from GPT-5.2 to GPT-5.3 (GPT-5.4 in the API). The same prompts, unchanged, now produce a very different answer: half as many URLs, half as many domains, and a radically different source mix.\n\nThe drop is uniform across all 4 locales (−47% to −53%), so this is a model/pipeline change — not a rollout that hit one country first. The breadth of the source pool per prompt shrank from 132 distinct domains to 61, and the source mix tilted away from UGC and chain brands toward a handful of programmatic- SEO hotel directories.\n\n−49%\n\nURLs per answer\n\n24 → 12\n\n−46%\n\nUnique domains\n\n18 → 10\n\n−54%\n\nDomain pool per prompt\n\n132 → 61\n\n−27pp\n\nWikipedia answer coverage\n\n31% → 4%\n\nFinding 1\n\n## The halving\n\nOn **March 5, 2026 — the day OpenAI rolled GPT-5.3 out to the ChatGPT UI** — the same 140 prompts started returning roughly half the sources overnight. Every headline metric moves together on that date, which is why we treat the UI model swap as the cause.\n\nSource metrics — before vs after the Mar 5, 2026 cutover\n\nMetric\n\nPre (GPT-5.2)\n\nPost (GPT-5.3)\n\nΔ\n\nCaptures\n\n6,414\n\n3,308\n\n—\n\nURLs per capture (mean)\n\n23.99\n\n12.21\n\n−49.1%\n\nUnique domains per capture (mean)\n\n18.43\n\n10.03\n\n−45.6%\n\nURLs per unique domain\n\n1.31\n\n1.24\n\n−5%\n\nUnique domains pooled per prompt\n\n132\n\n61\n\n−54%\n\nThe ratio **URLs ÷ domains** barely moved (1.31 → 1.24). ChatGPT didn't cite fewer pages per site — it cited **fewer sites**. The breadth of the source pool was halved.\n\nFinding 2\n\n## Uniform across locales\n\nIf the drop were a geographic A/B rollout, we'd expect country deltas to differ. They don't.\n\nURLs per answer — pre vs post cutover by country\n\nBy country: pre vs post URLs/answer and domains/answer\n\nCountry\n\nCaptures (pre / post)\n\nURLs/answer\n\nDomains/answer\n\nΔ URLs\n\nGermany\n\n1,249 / 739\n\n24.67 → 12.12\n\n18.75 → 9.82\n\n\\-50.9%\n\nSpain\n\n1,232 / 944\n\n26.24 → 12.44\n\n19.25 → 10.03\n\n\\-52.6%\n\nUnited Kingdom\n\n631 / 641\n\n23.11 → 12.10\n\n17.99 → 10.06\n\n\\-47.6%\n\nUnited States\n\n3,302 / 984\n\n23.07 → 12.13\n\n18.09 → 10.17\n\n\\-47.4%\n\nThe drop lands everywhere at once — a model/pipeline shift, not a regional rollout. Country-level deltas range from −47% (US, GB) to −53% (ES), a narrow band given the sample-size differences.\n\nFinding 3\n\n## The source mix tilted\n\nEvery category shrank in absolute terms, but some shrank far faster than others — and the tiny `seo_directory` bucket held flat, nearly tripling its share of URLs.\n\nCategory coverage — % of answers where category appears at least once\n\nCategory coverage — % of captures (answers) with ≥1 URL in that category, counted once per answer\n\nCategory\n\nPre %\n\nPost %\n\nΔ pp\n\nOTA\n\n83.2%\n\n60.4%\n\n\\-22.8pp\n\nMeta\n\n81.9%\n\n65.8%\n\n\\-16.1pp\n\nIndependent\n\n99.9%\n\n99.0%\n\n\\-0.9pp\n\nEditorial\n\n47.4%\n\n35.9%\n\n\\-11.5pp\n\nChain\n\n34.2%\n\n11.5%\n\n\\-22.7pp\n\nUGC\n\n21.1%\n\n2.4%\n\n\\-18.7pp\n\nSEO Directory\n\n20.9%\n\n22.8%\n\n1.9pp\n\nGoogle\n\n2.3%\n\n0.1%\n\n\\-2.2pp\n\nUGC coverage (Reddit, Instagram, YouTube, Facebook) collapsed from 21% of answers to 2%. Chain-brand coverage fell by two-thirds (34% → 11%). Google Maps/Search citations effectively vanished (2.3% → 0.1%). Meanwhile, **seo\\_directory** was the only category whose coverage _rose_ (21% → 23%) — in a market where every other source shrank.\n\nFinding 4\n\n## Winners & losers\n\nRanked by change in **answer coverage** — the share of ChatGPT answers in which a brand appeared at least once (counted once per answer, no double-counting for repeated mentions).\n\n**Why two delta columns?** Every answer now carries roughly **half as many URLs**(24 → 12), so every brand would lose coverage even if nothing else changed.\n\n-   **Δ %** — relative change: `(post − pre) ÷ pre`. Wikipedia 31% → 4% reads as −88%.\n-   **Δ vs halving %** — `(post − pre × 0.509) ÷ (pre × 0.509)`, where 0.509 is the URLs/answer ratio (12.21 / 23.99). This separates _“squeezed by the halving”_ from _“actively deprioritised”_. Green = gained share net of the halving, red = lost share on top of it. E.g. Wikipedia went from 31% to 4% — a halving would have predicted 16%, so the relative shortfall is −76%.\n\nTop losing brands — % of answers citing the brand, pre vs post cutover\n\nΔ vs halving — who genuinely gained (green) or lost (red) once you factor out the halving\n\nChart filtered to brands with ≥2% pre-coverage. Emerging SEO directories (pre near zero) are shown in the dedicated SEO panel below where the % blow-ups reflect real 10–30× growth.\n\nTop losers by Δ vs halving (actual post vs the halving baseline pre × 0.509)\n\nBrand\n\nCategory\n\nPre %\n\nPost %\n\nExpected post %\n\nΔ %\n\nΔ vs halving %\n\nwikipedia.org\n\nIndependent\n\n30.9%\n\n3.7%\n\n15.7%\n\n\\-88.0%\n\n\\-76%\n\nreddit.com\n\nUGC\n\n14.2%\n\n2.1%\n\n7.2%\n\n\\-85.2%\n\n\\-71%\n\nmichelin.com\n\nEditorial\n\n14.4%\n\n4.5%\n\n7.3%\n\n\\-68.8%\n\n\\-39%\n\nbooking.com\n\nOTA\n\n44.3%\n\n20.3%\n\n22.6%\n\n\\-54.2%\n\n\\-10%\n\nmarriott.com\n\nChain\n\n10.0%\n\n2.9%\n\n5.1%\n\n\\-71.0%\n\n\\-43%\n\ntrivago.com\n\nMeta\n\n17.0%\n\n6.6%\n\n8.7%\n\n\\-61.2%\n\n\\-24%\n\nTop gainers by Δ vs halving (brands that held more shelf than the halving predicted)\n\nBrand\n\nCategory\n\nPre %\n\nPost %\n\nExpected post %\n\nΔ %\n\nΔ vs halving %\n\ntripadvisor.com\n\nMeta\n\n43.8%\n\n33.5%\n\n22.3%\n\n\\-23.5%\n\n+50%\n\ntheluxuryeditor.com\n\nEditorial\n\n17.1%\n\n15.1%\n\n8.7%\n\n\\-11.7%\n\n+73%\n\noyster.com\n\nIndependent\n\n21.5%\n\n16.1%\n\n11.0%\n\n\\-25.1%\n\n+47%\n\nexpedia.com\n\nOTA\n\n56.2%\n\n33.3%\n\n28.6%\n\n\\-40.7%\n\n+16%\n\ntravelmyth.com\n\nMeta\n\n12.5%\n\n10.3%\n\n6.4%\n\n\\-17.6%\n\n+62%\n\ntrip.com\n\nOTA\n\n16.5%\n\n11.9%\n\n8.4%\n\n\\-27.9%\n\n+42%\n\nAdjusting for the halving flips the story for the biggest OTA/meta names. **theluxuryeditor.com held +74% above halving, TripAdvisor +50%, Oyster +46%, Travelmyth +61%, Trip.com +42% and Expedia +16%** — all genuinely won shelf. Even Booking.com (−10% vs halving) is close to neutral. The genuine losers are UGC and editorial: **Wikipedia −76% below halving**is the clearest deprioritisation, Reddit −71%, Four Seasons −68%, Agoda −69% and Hilton −54% all collapsed far below a uniform halving. Every luxury chain brand landed in the −40% to −68% range. In the long tail, the yonder network (yonder-society.com + yonder.fr, combined pre 10.1% → post 6.1%) held +20% above halving. The other above-halving gainers are programmatic-SEO listicle sites — the subject of Finding 5.\n\n### Δ vs halving, broken down by source type\n\nSame metric expressed as a **relative percentage**: `(post − pre × 0.509) ÷ (pre × 0.509)`. Within each type the story is self-contained: OTAs and metas have a clear internal ranking, editorial is mixed, chains are universally down, UGC is a bloodbath, and the SEO directories have no losers at all.\n\n#### OTAs\n\nBooking, Expedia, Hotels.com, Agoda, Trip.com and peers\n\n**Expedia (+16%) and Trip.com (+42%)** genuinely won shelf. Agoda (−69%) and Booking.com (−10%) landed below halving.\n\n#### Metasearch & curated directories\n\nTripAdvisor, Kayak, Trivago, The Hotel Guru, Travelmyth\n\n**TripAdvisor (+50%)** is the single biggest winner of the whole cutover (in absolute pp). Kayak, Trivago and Mr & Mrs Smith all lost share on top of halving.\n\n#### Editorial & publishers\n\nCondé Nast, Michelin, Forbes, Time Out, The Luxury Editor\n\nEditorial is mixed: **theluxuryeditor.com (+74%)** and Time Out (+46%) gained, while Forbes Travel Guide (−71%) and Guardian (−27%) were demoted.\n\n#### Hotel chains\n\nMarriott, Hilton, Four Seasons, Ritz-Carlton, IHG and peers\n\nEvery single chain brand is red. **LHW −83%, Ritz-Carlton −69%, IHG −64%, Four Seasons −68%, Hilton −54%, Marriott −43%**. GPT-5.3 appears to actively deprioritise brand .com sites in favour of meta/editorial.\n\n#### User-generated (UGC)\n\nReddit, Facebook, Instagram, YouTube\n\nUniversal collapse. Reddit −71% below halving, and **Facebook, Instagram, YouTube all lost 90-95% of their expected shelf** (effectively zero post-cutover).\n\n#### SEO directories (programmatic listicles)\n\nluxuryhotel.guide, all-boutique-hotels, hotels-with-\\*, etc.\n\nNo losers. Every brand in this bucket gained share. Percentages are massive because most of these sites had near-zero pre-coverage — **all-boutique-hotels.com +3,000%, luxuryhotel.guide +1,170%, boutiquehotel.guru +780%** — i.e. 10-30× growth vs expected.\n\nFinding 5\n\n## The Stockholm listicle network syndrome\n\nThe biggest surprise is who took ChatGPT's freed-up citation slots. Not Michelin, not Condé Nast, not Marriott — a cluster of thin-content, single-operator Booking-affiliate directories.\n\n17\n\nConfirmed network domains\n\n14 primary + 3 secondary\n\n1\n\nShared image host\n\nimages.luxuryhotel.guru\n\n1\n\nShared registrar\n\nGandi SAS\n\n2\n\nCloudflare NS pairs\n\nANDY/RITA, BRENDA/GRAHAM\n\n### Sample sites\n\nAll share a hero search backed by Booking.com's affiliate widget, a templated destination grid, a Scandinavian curator persona (Ted Valentin, Maja Holm, Elain Olsson, David Bachmann), and byte-identical footer boilerplate.\n\nall-boutique-hotels.com\n\nhotels-with-balcony.com\n\nsmall-luxury-hotels.net\n\nadults-only-hotels.net\n\nspahotelbreak.com\n\nhotels-with-pool.com\n\nhotels-with-private-pool.com\n\nhotels-with-tennis.com\n\nhotels-with-sauna.com\n\nhotels-with-great-views.com\n\nhotels-with-rooftop.com\n\naward-winning-hotels.com\n\namazing-hotels.com\n\nromantichotels.com\n\nluxuryhotel.guide\n\nluxuryhotel.guru\n\nboutiquehotel.guru\n\npoolhotels.guide\n\n5starhotels.guide\n\nbeachhotels.guide\n\ncouples-hotels.com\n\nsmall-hotels-guide.com\n\nromantic-getaway.me\n\n### Infrastructure fingerprint\n\nWHOIS / DNS evidence\n\nFingerprint\n\nValue\n\nRegistrar\n\nGandi SAS\n\nCloudflare NS (primary cluster)\n\nANDY.NS.CLOUDFLARE.COM + RITA.NS.CLOUDFLARE.COM\n\nCloudflare NS (secondary cluster)\n\nBRENDA.NS.CLOUDFLARE.COM + GRAHAM.NS.CLOUDFLARE.COM\n\nImage host (shared across sites)\n\nimages.luxuryhotel.guru\n\nBulk .guide registrations\n\n5starhotels / poolhotels / beachhotels / luxuryhotel — all 2014-05-08\n\nAbout-page slug\n\n/about-us/ identical across inspected members\n\n### The operator\n\n**Ted Valentin**, named as curator on all-boutique-hotels.com, is a Stockholm-based entrepreneur publicly known for directory sites (hitta.se among others). The other curator names — Elain Olsson, Maja Holm, David Bachmann — do not correspond to verifiable individuals and appear to be personas reused across the templated sites.\n\nThe absolute shelf these sites hold is still small: post-cutover, the biggest of them (luxuryhotel.guide) appears in 2.6% of answers. It's a _spike, not a takeover_ — several of these domains had near-zero pre-cutover presence, so the percentage gains are large off a small base. What's notable is the direction: every single brand in this bucket gained, and the only category whose coverage rose while everything else shrank is this one. For a model update that otherwise tilted toward established meta/editorial names (TripAdvisor, Oyster, Expedia), the appearance of a Stockholm- operated affiliate network — not editorial authority — worth tracking.\n\nFinding 6\n\n## Luxury queries hit hardest\n\nWe classified each prompt into an intent tier by keyword. Luxury and high-amenity queries — which previously leaned on OTAs, chains and editorial — took the biggest hit.\n\nΔ URLs per answer by query tier\n\nBy query tier\n\nTier\n\nPrompts\n\nPre URLs\n\nPost URLs\n\nΔ\n\nRooftop / Pool\n\n5\n\n26.32\n\n11.30\n\n\\-57.1%\n\nLuxury (5-star)\n\n30\n\n27.74\n\n12.64\n\n\\-54.5%\n\nBeachfront\n\n7\n\n27.98\n\n12.91\n\n\\-53.9%\n\nFamily\n\n14\n\n23.34\n\n11.90\n\n\\-49.0%\n\nBoutique / Design\n\n28\n\n23.02\n\n11.89\n\n\\-48.3%\n\nOther\n\n30\n\n23.50\n\n12.42\n\n\\-47.2%\n\nAffordable\n\n12\n\n19.83\n\n11.69\n\n\\-41.0%\n\nRomantic\n\n14\n\n20.34\n\n12.15\n\n\\-40.3%\n\nFinding 7\n\n## City hotspots\n\nSt Barts and the Alpine ski markets were hit hardest — both previously pulled 30+ URLs per answer because the destinations are OTA-thin and required more sources to triangulate. Post-cutover every city lands in the same ~12-URL band.\n\nURLs per answer — pre vs post cutover by city\n\nBy city\n\nCity\n\nPre URLs\n\nPost URLs\n\nΔ\n\nSt Barts\n\n33.59\n\n12.28\n\n\\-63.5%\n\nCourchevel\n\n26.67\n\n12.16\n\n\\-54.4%\n\nIstanbul\n\n25.67\n\n12.02\n\n\\-53.2%\n\nBarcelona\n\n25.27\n\n12.02\n\n\\-52.4%\n\nAmsterdam\n\n23.78\n\n11.47\n\n\\-51.8%\n\nLos Angeles\n\n23.68\n\n11.42\n\n\\-51.8%\n\nRome\n\n23.46\n\n11.67\n\n\\-50.2%\n\nSaint-Tropez\n\n25.81\n\n13.27\n\n\\-48.6%\n\nBerlin\n\n23.65\n\n12.28\n\n\\-48.1%\n\nSan Francisco\n\n22.61\n\n11.84\n\n\\-47.7%\n\nMegeve\n\n23.82\n\n12.45\n\n\\-47.7%\n\nDubai\n\n23.30\n\n12.45\n\n\\-46.5%\n\nNew York\n\n23.45\n\n12.64\n\n\\-46.1%\n\nLas Vegas\n\n23.24\n\n13.05\n\n\\-43.8%\n\nTimeline\n\n## Weekly view of the cliff\n\nThe drop isn't drift — it's a step change. Weekly means sit in a ~23-URL band for all of January and February, then collapse to ~12 in the week of March 9 and stay there.\n\nWeekly mean URLs and domains per ChatGPT hotel answer\n\nWeekly means weighted by capture count across all 4 country locales. Cutover week is 2026-03-02. The tough week of 2026-03-23 (mean 8.95) is rollout noise — the following weeks return to a stable ~12-URL equilibrium.\n\nWhy this happened\n\n## Two plausible drivers\n\nThe data is what it is. The reason ChatGPT now consults half as many hotel sources isn't in the data — OpenAI didn't publish a rationale for the 5.3 rollout. But two hypotheses fit the shape of what we see.\n\nHypothesis 1\n\n### Cutting serving cost\n\nEvery extra URL pulled into context is a web-fetch request, extra tokens in the context window, and more reasoning time. Going from 24 URLs / answer to 12 roughly halves the web-search + context- assembly cost for every hotel query. ChatGPT handles millions of travel queries a day, so a straightforward **50% cost cut** on a whole vertical is a meaningful margin lever.\n\nConsistent with this: the halving is uniform across all 4 country locales and all query tiers. That looks more like a product/config knob being turned than an organic quality improvement.\n\nHypothesis 2\n\n### Reverting a GPT-5.2 spike\n\nIt's possible 5.2 was the anomaly, not 5.3. GPT-5.2 may have shipped with an unusually high retrieval budget (24 URLs / answer is quite a lot of citations for a 300-word answer), and 5.3 simply pulls that back toward a pre-5.2 baseline.\n\nWe don't have pre-5.2 data at the same resolution, so this is speculation — but if it's right, 5.3's 12 URLs isn't a cut, it's a **return to normal**. Implication for marketers: don't over-index on the pre-March 2026 citation counts; they may have been the outlier.\n\nBoth hypotheses predict the same data, and they aren't mutually exclusive — 5.3 could be both cheaper to serve _and_ closer to a long-run equilibrium. What they have in common is that neither implies a _quality_ reason for the cut. If OpenAI thought 24 URLs gave materially better answers, they'd have kept them.\n\nMethodology\n\n## How we ran the study\n\n### Data collection\n\n-   Same 140 world hotel discovery prompts, run daily from 4 country locales (US, GB, DE, ES)\n-   Capture the ChatGPT UI (not the API) — including the full sources panel per response\n-   Per URL: domain, position, cited=true/false (was it linked inside the answer text?)\n-   Coverage window 2025-12-25 → 2026-04-17 (87 distinct capture days, 9,722 responses)\n-   Residual gpt-5-2 rows after Mar 5 are dropped as rollout noise\n\n### Processing\n\n-   **Brand rollup.** Localised TLDs (tripadvisor.es, expedia.de, hoteles.com, fr.wikipedia.org, maps.google.com) are canonicalised to a single brand.\n-   **Taxonomy.** Every domain is bucketed into OTA, meta, chain, editorial, UGC, Google, independent, or the new `seo_directory` bucket for programmatic-SEO hotel listicles.\n-   **Two views.** `citation_count` = every URL in the sources panel. `cited=true` = URLs also linked inline in the answer text.\n-   **Listicle forensics.** WHOIS + Cloudflare nameserver lookups + image-host and template inspection, not just scrape signals.\n\n### Example prompts\n\nThe 140 prompts span 14 destinations × 10 intent tiers (luxury, beachfront, rooftop, family, romantic, affordable, design/boutique, ski, other). A representative sample:\n\n“best luxury hotels in Dubai”\n\n“5 star hotels in Rome with rooftop terrace”\n\n“romantic hotels in Saint-Tropez for couples”\n\n“family friendly hotels near Brandenburg Gate”\n\n“boutique hotels near Place des Lices Saint-Tropez”\n\n“best beachfront hotels in Los Angeles”\n\n“design alpine hotels in Courchevel village”\n\n“affordable hotels in Amsterdam with good reviews”\n\n“5 star hotels in Las Vegas with spa”\n\n“luxury hotels in Barcelona with pool”\n\n“family friendly hotels near Pampelonne beach”\n\n“affordable ski hotels in Megeve”\n\nFull list of 140 prompts with tier + city classification: [/data/chatgpt-hotel-source-shift-2026/prompts.csv](/data/chatgpt-hotel-source-shift-2026/prompts.csv). Aggregate headline stats: [summary.csv](/data/chatgpt-hotel-source-shift-2026/summary.csv).\n\n### Caveats\n\n-   Post-cutover sample is lighter (3,308 vs 6,414) because of the shorter post window. Country-level deltas are consistent, which mitigates this.\n-   The ChatGPT UI may have product-level changes stacked on the model change. We treat them as a single “March 5 intervention” because they shipped simultaneously.\n-   WebFetch 403'd on several listicle homepages; WHOIS and DNS evidence confirmed the network irrespective of that.\n-   140 prompts is a finite set chosen for breadth of destinations and intent tiers, not statistical exhaustiveness.\n\nFAQ\n\n## Frequently asked questions\n\nCitations per answer dropped 49%. Before Mar 5, 2026 each hotel answer consulted 24 URLs from 18 distinct domains on average. After Mar 5 the same prompts returned just 12 URLs from 10 domains. The pool of unique domains across all 140 world prompts also halved: 132 → 61. The drop is consistent across all 4 country locales we tested (US −47%, GB −48%, DE −51%, ES −53%).\n\n### Summarize with AI\n\n## Continue reading\n\nOther studies on ChatGPT hotel search.\n\n[ChatGPT Index vs Live Web](/research/chatgpt-hotel-index-vs-live-web-2026)[ChatGPT Hotel Ads Are Live](/research/chatgpt-hotel-ads-live-2026)[Anatomy of a ChatGPT Hotel Search](/research/anatomy-chatgpt-hotel-search-2026)\n\n[All Research](/research)","author":{"@type":"Person","name":"Nicolas Sitter","url":"https://nicolassitter.com/about","sameAs":["https://www.linkedin.com/in/nicolassitternolleau/","https://github.com/Nicositter88","https://hotelrank.ai"]},"publisher":{"@type":"Person","name":"Nicolas Sitter","url":"https://nicolassitter.com"},"image":"https://nicolassitter.com/api/og/chatgpt-hotel-source-shift-2026","mainEntityOfPage":{"@type":"WebPage","@id":"https://nicolassitter.com/research/chatgpt-hotel-source-shift-2026"},"tags":["ChatGPT","GPT-5.3","AI Visibility","Source Analysis"],"sameAs":["https://hotelrank.ai/research/chatgpt-hotel-source-shift-2026"],"alternateFormat":{"html":"https://nicolassitter.com/research/chatgpt-hotel-source-shift-2026","json":"https://nicolassitter.com/api/post/chatgpt-hotel-source-shift-2026","rss":"https://nicolassitter.com/rss.xml"},"datasets":[{"name":"summary","contentUrl":"https://nicolassitter.com/data/chatgpt-hotel-source-shift-2026/summary.csv","encodingFormat":"text/csv"},{"name":"prompts","contentUrl":"https://nicolassitter.com/data/chatgpt-hotel-source-shift-2026/prompts.csv","encodingFormat":"text/csv"}]}