Scrapes That Don't Break on Page Two

Scrapes That Don't Break on Page Two

Build the scraper template on a row with a missing field, cap the first run, and confirm page two actually differs from page one.

bardeen web-scraping data-extraction browser-automation lead-lists quality-control

What it is for

A build-and-verify routine for Bardeen scraper templates, covering the split between list pages and record pages, capturing links alongside text, and the three checks that catch a template matching the wrong container. Reach for it before your first real scrape of a directory, search result or listing page. It prevents the quiet failure where columns shift on rows with missing fields and nobody notices until the data is already in the CRM.

The skill

Build the scraper template against the hardest row, not the first one.

Before you capture anything:
- Decide whether this is a LIST page (many rows) or a DETAIL page (one record).
  A list template and a detail template are different templates. Do not try
  to make one do both.
- Scroll to a row that is missing something - no location, no price, no
  subtitle. Build against THAT row. Templates built on a complete first row
  silently shift columns when a field is absent further down.

Capture:
- Name every field the way you want the destination column named.
- Take the link as well as the text wherever a row links somewhere. You will
  want it later, and re-capturing means rebuilding.
- If the fields you need only exist on each record's own page, capture the
  row URLs here, then run a second scraper over that list in the background.
  One template cannot reach through to a page it never opens.

Before you trust it:
- Run it and stop at page one. Count the rows returned against the rows you
  can see. A different number means the template matched the wrong container.
- Turn on pagination and run again with a small limit. Confirm page two rows
  differ from page one. Repeated page-one data means pagination clicked nothing.
- Open the last row of the output. Trailing junk means the selector is
  catching a footer, an ad slot, or a "load more" element.

Never let the first run go unbounded. Cap it, read the output, then raise the
cap. An unbounded first run is how you spend credits proving the template was
wrong.

The failure this prevents: eight hundred rows that look fine in the sheet and
have the wrong company against half the names.

How to install it

1. Open the tool and start a new conversation, project or template.
2. Paste the text below wherever the tool keeps standing instructions
   — usually called system prompt, instructions, or persona.
3. Save it so it applies to every run rather than a single message.

Did it work for you?

Where it came from

Skills app Original