Two people, 8.3M city pages: how a Greek weather site climbed to 180k global rank by waiting for Google to index it
What it is A free global weather site with 8.3M locations in 52 languages, one forecast page per location, monetized with ads.
Why it took off It uses public meteorological data plus its own correction engine to spread tens of millions of long-tail pages with live numbers. It's been building for nearly two years, and this autumn search engines indexed it heavily, with rankings climbing steadily.
CHow you could follow it Don't build another weather site. Apply the same playbook to a vertical big sites ignore, where the data is public and changes daily (air quality, tides, pollen, fuel prices, etc.). Put live numbers on every page, publish a full methodology and editorial policy, and let indexing take its time.
Global rank 181k · 90 days ago, top 1M+
In 2026 Jul 29, a Hacker News account called weatherfun posted a Show HN: "Free weather maps, 25 model layers, live lightning and storm cells." It got 8 points and no comments. The same day they posted another about a weather widget, which got 3 points. The poster was Eleftherios Papadopoulos, a programmer in Thessaloniki, Greece. His site wfy24.com had already been around for nearly two years by then and wasn't yet in the global top 1M. A month later, in Aug 27, it first broke into the top 1M; by Nov 5 it reached rank 180,844. What pushed it up wasn't those posts. It was the tens of millions of weather pages it had built up over two years.
What it is
WFY24 is a free global weather site: one page per location, with current conditions, hourly and 16-day forecasts, air quality, weather maps, radar, and lightning. The About page says 199 countries, 8.3M locations, 52 languages. Behind it is a two-person company, Aethra Tech: Eleftherios writes the code, and co-founder Ioanna Vergini, who studied environmental science and meteorology, holds the title of editor-in-chief.
Open any page at random, such as Maggie Valley, a small town in North Carolina, USA. The title reads "Maggie Valley (Haywood County) Weather | 16-Day Forecast"; the body opens with "It's 49°F and cloudy in Maggie Valley at the moment"; below, a line says "Light rain Thursday through Saturday, about 8.6 mm in total, a small fraction of a normal month's rainfall"; a note beside it says the nearest weather station is 24.6 km away. That's one template, multiplied by 8.3M locations and 52 languages.
The money comes from ads: the site serves Google Ad Manager / AdX. There's also a free weather widget credited "Powered by WFY24" (with a Pro tier), a weather API still in beta, and a Greek-language site, kairos.gr, built on the same engine. Revenue isn't public.
Why now
The ranking curve is very flat: Aug 27 at rank 978,906, Sep 10 at 524,804, Oct 1 at 357,377, Oct 22 at 264,134, Nov 5 at 180,844. Over ten weeks it climbed almost every day, with no spike-and-fall. Traffic from a single exposure doesn't look like this; gradually broader indexing and ranking does. Chrome's real-user data also first put it in the global top 1M in September, but it doesn't rank in any single country. Its users are spread across many countries, which is what multilingual long-tail pages tend to look like.
Why this autumn specifically? Probably three things stacking up:
- The page pool kept growing. In a dev.to post, the author, March, wrote "5M locations, 51 languages"; earlier, when the site was listed in a product directory, the figure was 2.5M; now it's 8.3M. More pages means more long-tail queries it can rank for.
- The pages aren't empty shells. Every page has live numbers, and titles and summaries pack in specific figures like "Highest of X° on Wednesday, X rainy days in 16", which naturally lifts click-through. The site also publishes a full set of pages covering methodology, data sources, the team, and editorial policy. The editorial policy page openly states that "most of our tool pages are assembled from location data, forecast data, and structured templates": it admits they're templates, and tells search engines that someone is responsible for them.
- Similar sites are declining while it rises. Several programmatic weather sites with the closest playbook fell from August to November: weather25.com dropped from about 205k to 288k, weather.town from 196k to 320k, and world-weather.info from 41k to 78k. One possibility is that a search engine quality update rewarded pages with real data; that's an inference I can't verify. Also, weather searches are naturally heavier in autumn, and several regional weather sites (meteum.ai, freemeteo.gr, okairos.gr) rose over the same period, so it caught some of that tailwind.
A caveat: the site sits behind Cloudflare, and part of the ranking data is based on DNS resolution. Widget and API calls also nudge the rank up a little. How many of the 180k are real people can't be told from the outside, but Chrome's real-user data does place it on the list, so real people are there.
Where users come from
Almost all of it comes from search. The other channels the author tried are, on paper, very small:
| Channel | What I found | Results |
|---|---|---|
| Location × language long-tail pages | Japanese, Greek, and US small-town pages are being indexed heavily | Main channel |
| Show HN | From 7 to September: 6 posts, top score 8, 0 comments | Almost zero, a few backlinks |
| Free widget | WordPress plugin listed in Apr 29, 50 active installs | Few backlinks |
| Facebook country pages | About 6,400 followers on the global page, 206 on the US page; link placed in the first comment | Small |
| X / TikTok | Reply posts on the US weather account during severe weather; TikTok with 13 followers | Basically unused |
| Weather × travel articles | Article numbering has passed 1,300, footer includes "Set as preferred source on Google" | Trying for Google News and Discover; how much traffic that brings is unknown |
I saw no ad buying. It also has no app.
Who it's competing with
This is one of the most crowded categories on the internet. weather.com and AccuWeather rank in the global top 1,000; Windy (around rank 2,000) and Ventusky are the leaders for weather maps, and WFY24's 25-layer maps are a small-scale version of that approach; timeanddate, yr.no, meteoblue, and foreca have run "one page per city" for over a decade.
To users, WFY24 doesn't feel very different from them. Its real difference is in the eyes of search engines: more languages (52), locations pushed down to villages and towns (8.3M), live data on every page, plus a full set of "someone is responsible" explainer pages. It didn't invent any new feature. It carved out a slice of search results in smaller languages and smaller places that big sites weren't covering.
The data and costs are why it can sustain itself for two years: forecasts use public data such as Germany's DWD ICON and the US GFS, and it wrote its own correction engine called HYPERION (bias-corrected using measurements from 19k+ weather stations), running on a dedicated Hetzner server in Helsinki. Adding a page costs almost nothing extra, so ads can cover the bill.
How many people know about it
Almost nobody. Its best Hacker News score was 8 points, nobody has mentioned it on X in the last 90 days, there's no media coverage, no public revenue, and I saw no one copying it.
If it were you
Verdict: building another weather site isn't worth it. What's worth learning is the playbook: public data plus a layer of your own calculations, multiplied across "locations × languages" to spread pages, then wait patiently for indexing.
Who it's for: developers who can write a data pipeline, maintain tens of millions of pages alone, and survive a year or two with no revenue; or people who already have structured data others can't easily get.
Who it's not for: anyone who wants results in three months; anyone counting on HN or Product Hunt launches to drive traffic. WFY24 posted 6 Show HNs and none of them took off. Also not for people who can't make pages with real "useful signal" and just want AI to mass-generate text: that batch of similar sites is already sliding.
How to do it: don't go after weather itself. Pick a niche no big site has covered, with public data and something that changes daily for every entity: air quality, tides, sunrise and sunset, pollen, public holidays, fuel prices, local cost of living. Every page should have live, specific numbers (in the title too), plus methodology, data sources, team, and editorial policy pages. Get indexing working in three to five languages first, then expand into smaller ones. One engine can output a website, widgets, and an API, so one dataset can be sold several times.
How to get users: search is the acquisition channel. Launch platforms are only good for backlinks, not traffic. Widgets with attribution are a way to slowly build backlinks, but the WFY24 example shows it's slow.
Window: there's no deadline, this is slow work. But the pages need to hold up before the search engine's next quality update. Build them as empty templates like weather25 did, and they're the next batch to drop.
What would show you're wrong: if WFY24's rankings fall back to beyond 300k over the next one to two months, the rise was mostly seasonal or a one-off algorithm swing, not proof that this approach works. If your pages have been live for six months, are indexed, and still get almost no clicks, the big sites in your niche have taken the search results, so switch.
Research notes: competitor table, acquisition evidence, full research report
- Competitors
Name Where Launched Scale How it differs from this one weather.com / AccuWeather Big-company features 1990s Tranco 230 / 644 Brand + app + owned data; WFY24 targets the smaller languages and local long tail they don't cover Windy / Ventusky Web 2014 / 2016 Tranco 2,070 / 14,930 The main player for multi-model weather maps. WFY24's weather maps are a smaller version of the same approach; the real entry point is city pages timeanddate / yr.no / meteoblue / foreca Web 2000s Tranco 2k–25k Also one page per city, built over more than a decade. WFY24 competes on 52 languages and 8.3M locations, down to villages weather25.com / weather.town / world-weather.info / weather-us.com Web Years ago Tranco 288k / 320k / 78k / 164k, all falling since August Same programmatic SEO approach. WFY24's pages have more live data and maps, and it also has credibility pages like methodology and editorial policy. It's rising over the same period weatherwidget.io and other weather widgets Web Years ago Tranco 38k WFY24's widgets are just a side effect of traffic acquisition. Its WordPress plugin has only 50 active installs Open-Meteo / WeatherAPI / Tomorrow.io Other 2015–2022 Tranco 7.7k / 25k / 39k Aethra's Weather API is still in beta, not yet competing head-on - Where users come from
- Verified Programmatic SEO (place × language long-tail pages): 8.3M places, 52 languages; Japanese, Greek, and US small-town pages are heavily indexed (Brave site: search); the editorial policy page admits pages are generated from templates plus datasets; the smooth ranking climb matches indexing expanding
- Inferred Content articles / Google Preferred Sources: the /en/articles pagination cursor shows article numbers above 1,300; the footer has "Add as a preferred source on Google"
- Verified Country-specific Facebook pages: wfy24.global has 6.4K followers, wfy24.us has 206, plus Wfy24 Србија and others; links are placed in the first comment to drive traffic
- Verified X: riding weather news / TikTok: @WFY24Weather replies on the US weather account; TikTok 13 followers
- Verified Show HN / dev.to / directories: 6 Show HN posts, the top one at 8 points and 0 comments; dev.to has 1 articles; listed on launchrecord and alternativeto; a self-created Baidu Baike entry
- Verified Free widgets with backlink attribution: widgets carry "Powered by WFY24" and come in HTML / WP / Joomla versions; the WordPress plugin is listed with 2026-04-29 and has 50 active installs
wfy24.com: why it works
One line
WFY24 is a free global weather site run by a two-person company, Aethra Tech, based in Thessaloniki, Greece. It has 8.3M locations, 52 languages, and 16-day forecasts, plus tool pages for multi-model weather maps, radar, lightning, AQI, earthquakes, and weather along driving routes, monetized through ads (Google Ad Manager / AdX). Its traffic growth comes mainly from programmatic SEO: places times languages, producing hundreds of millions of long-tail pages. Once Google started indexing them in large numbers, traffic kept climbing from late August in 2026. Launch channels like HN, developer communities, and WordPress widgets have brought very little.
1. Data
- Tranco (traffic library, rank_day): 2026-08-27 first entered the top 1M, rank 978,906 → 09-10 524,804 → 10-01 357,377 → 10-22 264,134 → 11-05 180,844. It rose almost every day over 10 weeks with no dips. This steady climb looks more like search indexing slowly expanding than a single exposure spike (after a single exposure, the 30-day rolling average would surge and then fall back).
- CrUX (crux_domain): 2026-09 made it into the global tier for the first time at 1M (best_origin www.wfy24.com), and it isn't in any country tier for 21 countries. That means Chrome real-user traffic just cleared the threshold, and the users are spread across several countries, which fits a multilingual long-tail SEO pattern.
- Caveat: the site sits behind Cloudflare (visiting /about brings up a Cloudflare verification page), and Tranco's five sources include Cloudflare Radar and Cisco Umbrella (DNS-based). DNS lookups from the widget.js widget and API calls may push the ranking up somewhat. On the plus side, CrUX also reached the top 1M, so real users exist; how many of the 180k ranking visits come from actual people is still uncertain.
- The domain isn't new: Wayback has captures from 2024-09-24, and the author wrote on dev.to (2026-03-18) "I've been building wfy24.com solo for the past couple of years." The card's "not on the chart for 90 days" marker means the site took a long time to gain traction, not that an old domain changed hands.
- Scale over time: 2026-03 the author described coverage as "5M+ locations / 51 languages" (dev.to); when launchrecord listed it, it said "2.5M locations"; the current about page says "199 Countries · 8.3M Locations · 52 Languages" (wfy24.com/en/about, 2026-09-18 published). The page count grew several times over in a year, which lines up with the timing of the Tranco rise.
- No app: searching the App Store for wfy24 returns nothing, and the traffic library's product_domain only matches HN. All the traffic is on the web.
2. Competitors (Tranco ranks from the traffic library; 2026-11-05 vs. 2026-08-10 compared)
| Competitors | Type | Rank now / August | How it differs from WFY24 |
|---|---|---|---|
| weather.com / accuweather.com / msn weather | big tech | 230 / 644 / 94 | Brand + app + owned data; WFY24 targets the smaller languages and small places they don't cover |
| windy.com / ventusky.com | Web + app, weather maps | 2,070 / 14,930 | Multi-model map layers are their home turf. WFY24's "25-layer weather map" follows this path, but at a scale two orders of magnitude smaller |
| timeanddate, yr.no, meteoblue, foreca | Long-established global city forecasts | 2k–25k | Same "one page per city" approach, built over more than a decade. WFY24 competes on language count (52) and place count (8.3M, down to villages) |
| weather25.com / weather.town / world-weather.info / weather-us.com / weathertab | Similar programmatic SEO weather sites | 288k (August 205k) / 320k (196k) / 78k (41k) / 164k (143k) / 175k (112k) | The closest in approach, all declining over the same period; only WFY24 is rising |
| meteum.ai, freemeteo.gr, okairos.gr | Regional | 34k(48k)/ 11k(23k)/ 26k(50k) | Also rising over the same period, possibly from seasonal demand for autumn weather searches |
| weatherwidget.io and various WP weather plugins | Widgets | 38k | WFY24's widgets are just a traffic side effect; its WP plugin has only 50 active installs |
| open-meteo.com, weatherapi.com, tomorrow.io | API | 7.7k / 25k / 39k | Aethra's Weather API is still in beta and hasn't met this segment head-on yet |
Conclusion: this category is extremely crowded, with big companies and established sites everywhere. WFY24 has no new features; it's carving out a position on long-tail pages.
3. Growth tactics (evidence)
- Programmatic SEO (the main channel, highly credible): a page is generated for each place and each language, with URLs like /en/weather/maggie-valley-us154172056, /ja/weather/san-francisco-us26819236, /el/weather/kopi-gr2861923914. A Brave search for site:wfy24.com shows many Japanese, Greek, and US small-town pages indexed. The title and description templates include specifics like "Sat high X°C, X of the next 16 days rainy," which helps click-through rates. There are also country, region, AQI, time, holiday, earthquake, hiking route, and driving route pages. The site's editorial policy page states plainly: "Most utility pages are assembled from place data, forecast data and structured templates" (/en/methodology/editorial-policy). To look credible to Google, the site has Methodology, Data Sources, Team (listing co-founders and an "Editorial Team"), and Editorial Policy pages covering E-E-A-T.
- Content / press releases: the /en/articles pagination cursor shows article numbers have passed 1,300 (topics like "European hot springs winter temperatures", weather × travel content). The footer has "Add as a preferred source on Google" (Google's Preferred Sources / Top Stories entry point), so they're clearly trying to get into Google News and Discover. How much traffic this channel actually brings is unverified.
- Facebook country pages: wfy24.global has 6.4K followers, wfy24.us has 206, plus Wfy24 Србија and others. Posts use the common FB tactic of putting the live weather map link in the first comment (see the wfy24.us video titles). Volume is small.
- X / TikTok / Pinterest: @WFY24Weather replies to posts from US weather accounts during breaking weather (for example the Texas Hwy 287 pileup post) to ride their traffic. TikTok @wfy24.com has only 13 followers and 13 videos (TikHub). Basically no impact.
- Launch channels: HN user weatherfun (karma 6) posted 2026-07 Show HNs from 6 to September, the top one at 8 points with 0 comments; dev.to has only 1 articles; listed on launchrecord and alternativeto; the author posted in the Google Search Central community (titled "My site wfy24.com is a weather forecasting platform (8.4M locations…"), which now reads 410, likely asking about indexing or quality issues. These channels work as backlink sources, not as traffic sources.
- Backlink widgets: Free widgets carry a "Powered by WFY24 / wfy24.com" credit, in HTML, WordPress and Joomla versions, with a Pro upgrade (/en/widgets/pro). But WordPress.org lists a plugin 2026-04-29, with only 50 active installs and 1 ratings, so the external link footprint is small. On the DNS side it may have contributed a little to its Tranco rank.
- Self-built encyclopedia entry: The English Baidu Baike has a "Global Weather Forecast Platform WFY24" entry, which amounts to brand endorsement.
- Ads: The library shows no paid ads matched (card ads empty). No evidence of paid acquisition.
4. Why this one took off
- Page count is big enough: 8.3M locations × 52 languages means theoretically hundreds of millions of URLs, and over the year the pool grew from 2.5M to 5M to 8.3M pages.
- Owns its data and costs: Built on open data like DWD ICON, GFS and ECMWF, with a self-built HYPERION engine (interpolation, altitude correction, 19,000+ station correction, XGBoost post-processing), running on an independent server at Hetzner Helsinki (aethra.dev). Each extra page adds almost no cost, so ad revenue can cover it.
- Pages aren't empty templates: Every page has hourly and 16-day data, AQI, map layers, and a description with live numbers. Compared with other programmatic weather sites, its pages send stronger "usefulness" signals. After August, that group of similar sites fell together while WFY24 rose. One explanation is that Google's quality update rewarded this kind of page, but this is only speculation, unverified.
- Seasonality: Weather searches rose in autumn, and meteum, freemeteo and okairos rose at the same time. WFY24 caught part of that wave.
5. Differentiation
What users see differs very little: the weather visuals resemble Windy and Ventusky, the city pages resemble timeanddate and meteoblue. Three things really differ:
- Coverage of languages × places is larger than similar programmatic sites, reaching down to villages and towns;
- A proprietary engine plus an E-E-A-T shell (methodology, editorial policy, team pages);
- One engine powers four products: wfy24, the Greek kairos.gr, the widgets, and a Weather API.
Points 1 and 2 are what matter: the difference is visible to search engines, not to users.
6. How many people know about it
Almost nobody knows about it. Its best HN post got 8 points and 0 comments; in the X corpus it was mentioned 90 times in 0 days; no media coverage; no public revenue; no copycats seen.
7. What to take from it
- Programmatic SEO built on "open data + your own added layer + entities × languages" can be moved to other verticals (air quality, tides, sunrise and sunset, public holidays, prices...);
- Pair template pages with a methodology, editorial policy and team pages to lower the risk of being treated as spam pages;
- One engine powers a consumer site, localized sites, widgets and an API, so one dataset gets sold several times;
- Counterexample: launch channels like HN, dev.to and WP plugins brought almost no traffic. The real traffic came from long-term page building and waiting for indexing.
8. Not sure
- How much of the 180k names come from DNS / widgets / API versus real people browsing;
- Traffic share by language and country (it doesn't appear in any CrUX country chart);
- Traffic from articles and Google Discover;
- How much ad revenue it makes;
- Whether the group-wide drop among similar sites was caused by a Google algorithm change.
Sources
- www.wfy24.com/en/about, /en/team, /en/widgets, /en/methodology/editorial-policy, /en/articles
- aethra.dev/
- wordpress.org/plugins/wfy24-weather-widget/(api.wordpress.org:active_installs 50,added 2026-04-29)
- dev.to/sellinios/i-built-a-66kb-embeddable-weather-widget-he…(2026-03-18)
- HN item 49096599 / 49107291 / 49098382 / 49608866 (user weatherfun)
- traffic database: rank_day, crux_domain, product_domain (query date 2026-11-07)
- Facebook wfy24.global (6.4K), wfy24.us (206); TikHub @wfy24.com (13 followers); x-radar search for wfy24,90 days 0 posts
- Wayback CDX (2024-09-24 first capture)