NexSpark Research data files https://nexspark.com/research Publisher: NexSpark LLC (nexspark.com). Author: Matthew Firth. License for every file below: Creative Commons Attribution 4.0 (CC BY 4.0), https://creativecommons.org/licenses/by/4.0/ Free to use with a link. Suggested credit: "Source: NexSpark Research, nexspark.com/research, checked November 19, 2026". All files were generated on November 19, 2026 (US Eastern). Each CSV has one header row and no comment lines. Corrections are noted on the page each file supports. reverse-image-test-2026-11-19.csv Supports: https://nexspark.com/blog/reverse-image-search-dating-profile https://nexspark.com/blog/how-to-tell-if-a-dating-photo-is-ai-generated Date: November 19, 2026. Method: 15 portraits licensed under the Unsplash License. For each photo we searched the original image and an edited copy (mirrored horizontally, cropped to the central 75 percent, brightened 8 percent) in four engines (Google Lens, TinEye, Bing Visual Search, Yandex), using an image URL in a US Chrome session. That is 15 photos x 2 versions x 4 engines = 120 rows. A hit (source_found = yes) means the engine returned the Unsplash photo page or another page showing the same photo. Tag or listing pages count only when the photo itself appears on them. Lookalike photos don't count. Yandex blocked the test with a bot check on the first search (source_found = blocked); we did not try to get past it. Rescore: on November 19, 2026 two results were rescored to apply the rule strictly (Bing on the original of photo 13 and Google Lens on the edited copy of photo 14 returned only tag pages), so they are recorded as misses. Both rows say so in notes. Columns: id (photo number), version (original or edited), engine, source_found (yes, no or blocked), exact_matches (number of results or exact-match items the engine reported, blank when none was reported), notes (what the engine returned). Results change from day to day, so this is a snapshot of one date. matchmaker-price-transparency-2026-11-19.csv Supports: https://nexspark.com/blog/how-much-does-a-matchmaker-cost Date: November 19, 2026. Method: we read each firm's own website. A firm counts as publishing a price only when a dollar figure for the client fee appears on its own site; third-party reports don't count. 19 rows: the 18 firms counted, plus One on One Matchmaking, which blocks automated page reads, is marked not determined and is left out of the 18. Columns: firm, url_checked (pages read), price_published (yes or no), figure (what the site states), intros_stated (whether the site says how many introductions the price buys), cost_per_intro (only where both a price and an introduction count are on the firm's own site), notes. month-theory-trends-2026-11-19.csv Supports: https://nexspark.com/blog/august-theory-dating, https://nexspark.com/blog/november-theory-dating, https://nexspark.com/blog/december-theory-dating Date pulled: November 19, 2026. Source: Google Trends via DataForSEO Google Trends explore. United States, web search, weekly (Sunday to Saturday), past five years, 2021-11-14 to 2026-11-19. The final week (week_start 2026-11-15) is partial, covering November 15 to 19. Columns: week_start (first day of the week), term, value (Google Trends index, 0 to 100), pull. Method note: a Trends index is relative within one request. Each pull below was one request, scaled so its highest point is 100, so values are comparable only inside one pull. The same term can read differently across pulls (for example "december theory" peaks at 88 in dec-theory-3term and 82 in nov-theory-3term for the week of 2025-12-07). The pages say which pull they use. When the source returned no value for a term in a week (interest too low to report), it is recorded as 0, so a 0 means below the reporting floor, not a measured zero. The October theory page uses a separate single-term pull (January 1, 2024 to November 19, 2026) that is not in this file. Pulls: nov-theory-3term (november theory, december theory, cuffing season); aug-theory-3term (august theory, september theory, cuffing season); dec-theory-3term (december theory, november theory, cuffing season); dec-theory-4term (december theory, cuffing season, dating app, breakup). 262 weeks per term. Index values are not search counts. breakup-calendar-2026-11-19.csv Supports: https://nexspark.com/blog/breakup-season-data Date pulled: November 19, 2026. Source: Google Trends via DataForSEO. United States, web, weekly, 2021-11-14 to 2026-11-19. "breakup" and "divorce lawyer" were pulled in one request together with "how to get over a breakup", so they share one 0 to 100 scale. Method: weekly index values averaged by calendar month over the four full years 2022 to 2025. Columns: month, term, index_average_2022_2025. Index values are not search counts. fingerprint-checks-2026-11-20-double-texting.csv Supports: https://nexspark.com/blog/double-texting Date: November 20, 2026. Google US organic results for "double texting" via DataForSEO live SERP (desktop, 7 organic; mobile, 8 organic), each opened with a page reader. Columns: device, rank, site, title, url_opened, cited_named_study_survey_or_dataset, notes. 3 pages opened, none cited a named study, survey or dataset; 12 could not be opened. fingerprint-checks-2026-11-20-black-cat-energy.csv Supports: https://nexspark.com/blog/black-cat-energy-dating Date: November 20, 2026. Google US organic results for "black cat energy" via DataForSEO live SERP (desktop, 7 organic; mobile, 7 organic). The claim uses the desktop list only, excluding nexspark.com. Columns: device, rank, site, title, url_opened, gives_way_to_tell_apart_from_emotional_avoidance, notes. fingerprint-checks-2026-11-20-introvert-apps.csv Supports: https://nexspark.com/blog/dating-apps-for-introverts Date: November 20, 2026. US App Store rating and rating count per listed app (iTunes lookup API, cross-read on apps.apple.com). Columns: app, us_app_store_url, average_rating, rating_count, rating_count_as_shown_on_apple_page, checked, notes.