← back to Vendor Price Requests

CRAWL-FINDINGS.md

19 lines

# Cost-crawl viability (2026-06-17)

Tested whether vendor retail/cost is scrapable cheaply (static HTML curl) for the Tier-B set.
Result: **NOT viable via static HTML.** Sampled product pages:
- Designers Guild — price JS-rendered (Knockout.js `ko.unwrap(price)`); not in HTML.
- Arte International — no price in HTML (`price:0` / JS).
- Scalamandre — empty `PRICE` data-attrs (JS / trade-login).
- Elitis — `"price":0` (JS / not public).

Most Tier-B vendors are **trade-only** and publish NO public price. So:
1. **No cheap crawl path** — would require Browserbase (paid, per-vendor adapters) or trade logins.
2. For most, there's **no public price to get at all** → the vendor LETTER is the only path.

**Decision:** skip the crawl as uneconomical; route everything to the letter pipeline
(approve/send web page already built). A Browserbase crawl for the handful of public-retail
vendors is a separate, Steve-gated (spend) sub-project — not worth building blind.
The price_sourcing schema already supports a future crawl (crawled_cost/crawl_status):
if a crawl ever runs and finds a price, those SKUs auto-drop off the letter list.