Asian forum scrapability & cost table

Thirteen major Asian forums, after actually running them: what the data looks like, which exit IP it takes, and what one item costs. Not guesswork — 9 rows are read out of actors that run every day, the rest out of probe records.

Snapshot 2026-09-04 · 9 green / 3 amber / 1 red · 5 markets: TW, HK, SG, TH, MY

The next question after “can I reach it” is “what columns do I get” — the site coverage table lists the modes and the exact output fields every shipped actor returns.

Four things this table says

1. Not one of the 13 needs a browser

All 9 shipped forum actors run CheerioCrawler — plain HTTP, no browser, no Playwright. The single row on this table that truly requires a browser is BabyHome, and requiring one is exactly why we did not build it. "Forums need a headless browser" does not hold in Asia, and believing it makes your per-item cost 20× higher for nothing.

2. What blocks you is the CDN, not the site

Three of the four non-green rows (Dcard, Uwants, BabyHome) are stopped at Cloudflare, not by anything the site itself wrote; the fourth (KiasuParents) is an nginx UA allowlist. In practice that means your scraper is less likely to die because the markup changed than because a CDN layer appeared in front of it.

3. Residential solves a 403, not a JS challenge — and that line is a 5× cost difference

Dcard and Uwants get a flat 403 from Cloudflare — no waiting page, no JS challenge — and a residential IP turns that into a 200 at $0.0012 per item. BabyHome serves a managed challenge (the "Just a moment…" page); residential-TW does not get through either, leaving only a browser at $0.006 per item. So a 403 in a probe is good news; "Just a moment" is the bad one.

4. More of these sites expose a JSON API than you would expect

LIHKG (/api_v2), HKGolden (api.hkgolden.com/v1) and KiasuParents (NodeBB /api/recent) all serve open JSON endpoints, and so does Dcard — it is just blocked in front. A JSON endpoint is worth more than easy parsing: it does not break wholesale when the front end is redesigned.

How the cost works, and the 50% margin line

Against the common listing price of $0.002 per item, the three verdicts look like this:

So an amber site is not a question of whether it can be done — it is a question of whether you price it at $0.003 or above. One exception is worth knowing: residential bills by bandwidth, so cost per item scales with page weight. Uwants' archiver is a plain-text view, and our probe estimated it at roughly $0.0001 per item — an order of magnitude under a normal HTML site. The $0.0012 in the table is the upper bound for heavy HTML, not a flat rate for every amber row.

Separately, the platform charges $0.005 for apify-actor-start once per run, which the table does not include. Across 1,000 items it flattens to nothing; across 10 items it is your single biggest cost. To run your own numbers, use the Apify cost calculator.

Where this table comes from

• The 9 shipped rows: data shape, proxy requirement and endpoint type are read from each actor's own source — crawler class, the proxy prefill in its input schema, and the URLs it actually requests. Those actors run daily under canary monitoring.
• The 4 unshipped rows (Dcard, Uwants, KiasuParents, BabyHome): from probe records, each run over both a datacenter and a residential exit IP, recording the real HTTP status and the kind of block.
• Costs are estimates: Apify compute plus proxy bandwidth, dependent on page weight. Labelled an estimate because that is what it is.
• This is a 2026-09-04 snapshot. Anti-bot posture moves: a green site today can be behind Cloudflare tomorrow. If you cite this table, the date matters more than the verdict.

Three ways to start

From least effort to most freedom

I run it for you Managed, daily to your sheet You do nothing. From US$299/mo. Use it here Run the free tool Instant, no setup, on this site. Run my actor On Apify, your account Same data, you run it and pay per result. Build your own Sign up for Apify Free account — build the pipeline yourself.

This post contains Apify affiliate links (no extra cost to you). Costs and comparisons are stated honestly, including where Apify is not the best fit.

FAQ

How was "scrapable" decided? By actually running it, not by looking at the site. The 9 shipped rows come from actors that run every day — data shape, proxy requirement and endpoint type are read straight out of their source. The 4 unshipped rows come from probe records, each tried over both a datacenter and a residential exit IP.

Why is there no Dcard actor? Both routes into Dcard — the JSON API and the HTML page — return 403 from Cloudflare WAF, so it needs a residential IP. That is technically fine, but per-item cost jumps from $0.0003 to $0.0012, which leaves a 40% margin at $0.002 listing pricing and fails our 50% margin rule.

Can a green site turn red overnight? Yes, which is why the table is dated. The usual cause is not a site redesign but a new CDN layer appearing in front of it. Every shipped actor runs a daily canary so we find out immediately.

Why is a residential proxy more expensive? It bills by bandwidth, not by request, so cost per item scales with page weight. A heavy HTML page approaches the $0.0012 in the table; a plain-text endpoint like Uwants’ archiver can be an order of magnitude cheaper.

Do these numbers include the Apify start fee? No. The table shows per-item data cost. The platform charges a separate $0.005 apify-actor-start once per run. Across 1,000 items it disappears; across 10 items it is your largest single cost.

Is scraping these sites legal? We only scrape public content, never login-walled data, and we follow each site’s terms of service and local law. Where the table says we do not build a site, the reason is cost, not evading a block.

Chad runs 40+ published Apify actors, 9 of which are the forum scrapers on this table. Every verdict here maps to a request that was actually made; anything we could not run or could not source is not on the table.