How to Scrape Tiki.vn Product Prices in 2026

Tiki.vn is one of Vietnam’s top e-commerce marketplaces, and like most modern marketplaces its front-end is just a thin shell over a REST API that returns clean JSON. That’s good news for anyone who needs Vietnamese product data: you don’t need a headless browser, you don’t need residential proxies, and you don’t need to parse messy HTML. You hit Tiki’s search endpoints, page through results in parallel, and get back richly structured product records. This guide covers what Tiki exposes, how the HTTP-only architecture keeps costs near the floor, the product schema, and the pitfalls of scraping a high-throughput marketplace.

What’s worth extracting

Tiki returns a dense product record — 23+ fields — per listing:

• Identity — product ID, title, canonical URL.
• Pricing — current price, original (list) price, discount amount and percentage, all in VND.
• Reputation — average rating and aggregate review count.
• Demand signals — sales volume / quantity sold, availability/stock signals.
• Brand & seller — brand name, seller/store name and metadata.
• Classification — category taxonomy, badge information (official store, freeship, etc.).

Sales volume is the field that makes Vietnamese (and broader SE-Asia) marketplace data special — Tiki surfaces how many units a listing has sold, which is a far stronger demand signal than rating count alone. Pair price with sales volume and you can estimate revenue per listing.

HTTP-only, parallel, and cheap

Tiki’s search results come from REST endpoints that return JSON directly, so the scraper is HTTP-only — no browser. That single architectural choice cascades into everything that makes Tiki cheap to scrape:

• Parallel requests. Without Chromium memory per worker, the scraper fires many concurrent requests against Tiki’s search endpoints, so a large catalog sweep finishes fast.
• Deduplication. The same product surfaces across multiple search queries and category pages; the scraper dedups by product ID so each item appears once.
• Automatic retry/backoff. Tiki rate-limits aggressive clients; the scraper backs off and retries transient failures instead of dropping rows.
• Cheap datacenter proxies. Because Tiki’s anti-bot is light, you don’t need an expensive residential pool — datacenter IPs are enough, which is the single biggest cost lever in marketplace scraping.

You target by search query or category, choose a sort order (relevance, top-seller, newest), and optionally constrain a price range. The scraper walks the paginated results and emits structured records.

Conceptually:

GET https://tiki.vn/api/v2/products
  ?q=laptop
  &sort=top_seller        # relevance | top_seller | newest
  &price=5000000,20000000 # VND range
  &page=1

▶ Run the Tiki.vn Scraper — search or category targeting, 23+ fields per product including price, discount, rating, reviews, sales volume, brand and seller. Parallel HTTP-only, deduplication, retry/backoff, cheap datacenter proxies. No browser.

Schema design for downstream use

A clean per-product row:

{
  "product_id": 274558912,
  "title": "Laptop Dell Inspiron 15",
  "url": "https://tiki.vn/...",
  "price_vnd": 14990000,
  "original_price_vnd": 17990000,
  "discount_pct": 17,
  "rating": 4.6,
  "review_count": 312,
  "quantity_sold": 1840,
  "brand": "Dell",
  "seller_name": "Dell Official Store",
  "is_official_store": true,
  "category": "Laptop",
  "badges": ["freeship", "official"],
  "scraped_at": "2026-05-29T10:00:00Z"
}

Choices worth making early:

• Keep quantity_sold prominent. It’s Tiki’s standout demand signal — multiply by price for an approximate revenue estimate per listing.
• Store both price_vnd and original_price_vnd. Discount-driven pricing is the norm; you need the list price to measure real markdown depth.
• Flag is_official_store. Brand-monitoring use cases hinge on separating official-store listings from third-party resellers.
• Stamp scraped_at. Marketplace prices and stock move constantly; a price without a timestamp is noise.

Typical use cases

• Vietnam market research — track price trends across categories over time; size category demand using sales-volume aggregates.
• Competitor pricing and discount monitoring — watch a rival’s listings for price changes and promotion launches on a schedule.
• Automated price monitoring — set price-range queries per category and re-run to catch shifts.
• E-commerce analytics — analyze ratings, reviews, and sales volume distributions by category to find under-served niches.
• Brand monitoring — compare official-store listings against third-party seller listings of the same brand to spot grey-market pricing.
• Product sourcing — surface trending products and reliable suppliers by sorting on top-seller within a category.
• Academic research — model Vietnamese e-commerce dynamics with structured, repeatable pulls.

The value is breadth plus the sales-volume signal: a single product’s price is trivia, but a category’s full price-and-sales distribution refreshed weekly is a market dataset.

Cost math

This actor is pay-per-event with results priced at $0.003 each plus the small per-run start fee. Cost scales directly with rows: a 5,000-product category sweep is about $15 in result fees. Because it’s HTTP-only on cheap datacenter proxies, there’s no residential-bandwidth bill or Chromium compute hiding underneath — the per-result price is essentially the full cost, and it’s predictable. Budget simply: expected products × $0.003. Compared to building your own concurrent crawler with retry/backoff and dedup on a VPS, you skip the build week and the maintenance whenever Tiki tweaks its search payload.

Common pitfalls

• Prices are in VND, and they’re large numbers. Don’t accidentally treat them as USD or truncate them — a 14,990,000 VND laptop is ~$590, not a typo.
• Dedup across queries. Products appear under multiple searches and categories; without product-ID dedup you’ll over-count and skew sales aggregates.
• Respect the backoff. Tiki rate-limits hard hitters; the actor’s retry/backoff keeps you under the threshold — don’t crank concurrency past what the proxies can absorb.
• Sales volume is cumulative, not recent. quantity_sold is lifetime-ish, not last-30-days — useful for ranking, less so for short-term trend unless you diff snapshots over time.
• Official vs third-party matters for brand work. The same brand sells through many sellers; always check is_official_store before drawing brand conclusions.
• Datacenter proxies are usually enough — until they aren’t. If a high-frequency run starts seeing blocks, throttle concurrency before reaching for pricier IPs.

Wrapping up

Tiki.vn is a near-ideal scraping target — a JSON REST API, light anti-bot, and a rich product schema with a genuine sales-volume signal. The work is in parallelizing safely, deduping across queries, handling VND normalization, and backing off cleanly under rate limits, all of which a managed HTTP-only actor already does. If you need fresh Vietnamese marketplace data without a browser farm or a residential-proxy bill, run the actor and go straight to analysis.

▶ Open the Tiki.vn Scraper on Apify — 23+ fields per product, parallel HTTP-only, cheap datacenter proxies, dedup and retry built in. Pay per result. Start with Apify’s free monthly credit.