Frequently asked questions.

NeighborhoodTreasure is an automated, public index of the live web. We discover websites from open signal sources, enrich them with public metadata, rank them by sustained traffic and freshness, and present the result as a continuously updated feed. There are no accounts and no votes, and organic ranking is independent — not for sale.

Domains come from open public sources only: the Tranco top-1M list, Hacker News, GitHub trending and curated awesome-lists, public RSS feeds, public web directories, and the open Common Crawl web index. We never scrape private platforms, bypass paywalls, or use any data that requires authentication.

Continuously. Ingestion crons run on short intervals throughout the day to pick up new domains and refresh signals; ranking, screenshots, and metadata refresh on their own schedules. The Hot, New, and Rising feeds typically reflect activity from the last few hours.

Either we have not discovered it yet through our public sources, the domain failed a basic liveness check, or it has been removed by request. A site is never delisted because enrichment failed — if we can confirm the domain exists, it stays in the index, even if all we have is a favicon and a hostname.

We make plain HTTP requests to read each site's public metadata (title, description, favicon, OpenGraph image), and fall back to a headless renderer for sites that require JavaScript. We respect robots.txt directives that apply to general crawlers and we do not bypass any access controls.

After fetching public text content, an AI pass infers a category, a region, and a brand color from what's on the page. Heuristics handle obvious cases (TLDs, language, well-known platforms); the AI handles the rest. Misclassifications can be reported via the contact page.

Hot reflects current activity over a short rolling window. It combines outbound clicks logged on this site, fresh inbound mentions on Hacker News and GitHub, and engagement velocity. The score decays quickly — a site falls out of Hot within a few days unless new signals keep arriving.

New is weighted entirely by first-detection recency. A domain we discovered today ranks above one we discovered yesterday. After about 14 days the New score effectively reaches zero and the site moves into the long-term feeds.

Rising measures acceleration relative to a site's own baseline. A small site with a sudden 10× jump in signal can outrank a much larger site with steady traffic. It's designed to surface things on the way up, not things that are already large.

Roughly 50% Tranco global popularity, 25% outbound clicks logged on this site, 15% freshness signals, and 10% quality and category coverage. All inputs decay with time so the index doesn't ossify around what was popular a year ago.

It's an estimate, not a measurement. We derive it from the site's Tranco rank using a power-law curve calibrated against publicly known traffic data. Treat it as an order-of-magnitude signal: a site ranked around 100 sees billions of visits per month, a site ranked around 100,000 sees hundreds of thousands.

You cannot pay to influence organic ranking. The Hot, New, Rising, and Overall scores come entirely from the public-signal formulas above — there is no editorial override and no way to buy a higher score. We do sell two clearly-labeled Sponsored placements: a slot in the home page hero billboard, and the hero banner at the top of any individual category page. Both rotate alongside organic picks, carry a visible 'Sponsored' badge, and do not change any organic score. Details and contact info are on the advertise page.

No. Coverage is the goal. A site never disappears because its score is low — it just sits further down the feed. The only reasons a site is removed are owner removal requests, confirmed malware, or legal obligation.

Around 60 fixed categories spanning consumer, business, civic, and industrial sites — AI, Developer, Design, News, Media, E-commerce, SaaS, Finance, Health, Travel, Crypto, Science, Politics, Podcasts, Photography, Books, Weather, Maps, Forums, Insurance, Charity, Agriculture, Construction, Manufacturing, Energy, Transportation, Telecom, Hardware, Museums, Parenting, and more. The list is fixed in code rather than user-generated, so URLs and per-category counts stay stable over time.

Classification runs heuristics first (TLD, language, well-known platform patterns) and falls back to an AI pass over your homepage's public text. If your homepage talks more about one topic than your actual product, the classifier follows the homepage. Tighten your title tag, meta description, and hero copy — or report the misclassification via the contact page.

No — each site has exactly one primary category. Keeping it one-to-one makes category pages clean and ranking comparable across sites. To slice further, use the cross-cut pages like /categories/{category}/regions/{region}.

Sixteen regions: Global, United States, Europe, United Kingdom, Canada, Australia, India, Brazil, Japan, South Korea, Germany, France, Africa, Latin America, MENA, and Southeast Asia. Region is inferred from TLD, language, and on-page signals; sites without a clear regional focus default to Global.

Yes. The Search page accepts country names directly — try "japan", "germany", or "brazil". The home feed also re-ranks based on your visitor country, which is derived from the request and not stored against any persistent identifier.

The category list is intentionally compact so each page has enough sites to rank meaningfully. Niche topics live inside their closest parent — board games under Gaming, knitting under Art & creative, mortgages under Finance. If a topic genuinely has hundreds of independent sites and no good parent, suggest it via the contact page.

The /search page runs over the full live index — currently 60,000+ sites — using full-text matching on each site's title and description, plus structured filters for country, region, category, language, and domain. Results are ranked by the same Overall score used elsewhere on the site.

Free text works ("design tools", "japan media"). Country and region names are auto-detected as filters, so "germany ai" and "southeast asia" do what you'd expect. For precision, you can also use operator syntax.

category:ai or cat:design filter by category. region:eu filters by region slug. country:japan filters by country (full names work too). lang:en or language:de takes an ISO-2 language code. domain:vercel.com or site:nytimes.com matches an exact domain. Operators combine freely with each other and with free text — for example: ai country:france lang:fr.

Yes. Each category and region page has its own filter UI, and /categories/{slug}/regions/{region} gives you the cross-cut directly. You can also stack operators on the search page, e.g. category:ai region:us.

Just public metadata we capture during ingestion: title, description, language, and category. We don't full-text index the body of every site, so very specific phrases that only appear deep inside a site won't surface here.

Two common reasons: the classifier may have placed those sites under a different category than your free-text guess, or your keyword only appears in body text we don't index. Try the operator form (category:, domain:) or browse the relevant category page directly.

Search result pages are marked noindex on purpose — they're built for exploration, not for ranking. The category, region, and individual site detail pages are fully indexable and shareable instead.

Most listings refresh automatically when we re-fetch your site, so updating the metadata on your homepage (title tag, meta description, OpenGraph image) is the fastest fix. For a manual refresh or a category correction, use the contact page and include the domain.

Use the removal page. Verified domain owners can have their listing taken down — we do not require a legal threat or justification. The domain will be excluded from future ingestion as well.

Screenshots are captured periodically, not on every visit, and some sites block the screenshot renderer with a bot challenge. If the image is stale or missing, request a refresh via the contact page or update your OpenGraph image — we'll prefer that over a fresh screenshot when one is available.

No. Listings are automated and based on public signals. Inclusion is not a recommendation, an endorsement, or any kind of relationship between NeighborhoodTreasure and the listed site.

No. NeighborhoodTreasure is an automated directory built from public signals. We don't audit, verify, or vouch for any listed site, its operators, its content, or its business practices. Descriptions and summaries are produced automatically and may be inaccurate. Treat listings as a discovery tool, not a safety endorsement — you are responsible for evaluating the legitimacy of any third-party site before transacting with it. See our Terms for the full disclaimer.

No. There are no accounts, no logins, and no profiles. Everything you can do here works without signing in.

We log aggregate, non-identifying signals: outbound link clicks, impressions, share events, and the visitor's country (derived from the request, not from any persistent identifier). We do not use third-party analytics trackers, advertising cookies, or fingerprinting.

No. We do not sell, rent, or share visitor data. The aggregate signals we collect are used only to rank and improve the index.

An outbound click logs the target website, the surface it was clicked from (Hot, New, etc.), and the visitor's country. A share event logs the same plus the channel (X, Facebook, copy link, etc.). No personal identifiers are stored and the events are rate-limited and de-duplicated.

Tranco (a research-grade aggregate of public domain rankings), Hacker News, GitHub trending and curated awesome-lists, public RSS feeds from technology and design publications, the Common Crawl open web index, and a small set of public web directories. The full list is on the Data sources page.

Yes. We use OpenAI's gpt-4o-mini for two narrow tasks: classifying a site into a category and inferring a region from its public text content. The AI never decides whether a site is listed or how it ranks — it only labels. There is a hard daily budget cap to keep costs predictable.

There is no public API or bulk export today. The site itself, including category, region, and cross-cut pages, is fully crawlable, and an XML sitemap is published at /sitemap.xml. If you have a research use case, get in touch via the contact page.

Yes — two clearly-labeled Sponsored units: (1) a slot in the home page hero billboard, and (2) the full-bleed hero banner at the top of any individual category page (e.g. /categories/fashion). Both rotate alongside organic picks, carry a visible 'Sponsored' badge, and show a disclosure linking to the advertise page. Organic Hot, New, Rising, and Overall scores are not for sale and are not affected by sponsored placements.

Email starnestsocial@gmail.com with your site, how long you'd like to run, and your budget range. We reply with availability and a flat rate. Placement is manual; sponsored slots are always visibly labeled 'Sponsored'. Full details on the advertise page.

Pricing isn't published yet — rates depend on inventory at the time of inquiry. Email us for a quote.

Scams, MLM recruitment, deceptive offers, adult content, gambling, unregulated financial schemes, malware or phishing, anything illegal in the United States, and anything that would embarrass the index. We reserve sole discretion to decline or pull any sponsored placement at any time.

Yes. Every card has a share button (X, Facebook, LinkedIn, Reddit, WhatsApp, copy link) that points to the listing's detail page on this site. You're also welcome to deep-link directly to /site/{domain} from anywhere.

NeighborhoodTreasure is operated by StarNest LLC. There is no editorial team and no community moderation layer — the index is fully automated.

Use the contact page for general questions, classification corrections, or technical issues. Use the removal page for delisting requests. We respond to legitimate requests promptly.

No. We have no commercial, editorial, or partnership relationship with the sites in the index. Listings are purely the output of an automated pipeline over public signals — separate from any paid sponsored placements, which are clearly labeled.