Add a column.
Get the page
that said no.
Zenrows is the web data infrastructure under Clay.
How Clay reaches the protected web
Clay is where go-to-market teams build enriched prospect and account lists. A good part of what those lists need lives on sources that refuse automated access. Not one of Clay’s users builds a scraper to get it.
- Sources that refuse
- The pages a GTM team needs most defend themselves hardest, and change those defenses without warning.
- A scraper per source
- Reaching one meant building and maintaining a scraper for it, or copying the data across by hand.
- Hours per list
- Preparing a single enriched list ran to three to six hours of manual work.
- Silent failure
- A blocked request can still return a page. The list looks finished when it is not.
- Under the platform, not beside it
- Clay runs Fetch inside its own scraping layer, for the targets its native scrapers came back empty on.
- And in the table, as a column
- The same capability surfaced as an action, pointed at a URL another column produced.
- The method is not a setting
- Whether a page needs a real browser or a different origin is decided per request, not configured.
- Structured output
- Markdown or JSON lands in the cell, ready for the next column to read.
- Runs on every row
- New rows enrich as they arrive, and re-running the table re-runs the column.
-
Preparing one list
3–6 hours
Seconds
-
A large-scale scrape
2–3 weeks
Minutes
The protected web
A list starts as a table of companies and ends as something a rep can act on: funding, headcount, tech stack, open roles. Most of that is on the public web. A good part of it is on the part of the public web that does not want to be read by a machine.
What four of those sources return to an unassisted request
- 403crunchbase.comForbidden. The request never reaches the page.
- 302zoominfo.comRedirected to a sign-in wall.
- 200zillow.comA page, and a challenge where the listings should be.
- 429similarweb.comToo many requests, on the second one.
The 200 is the dangerous one. The list still looks finished.
Clay describes the first of those from their own side: a Crunchbase page needs no login, a single click on one still triggers a human verification step, and what is behind it is not structured for reading.
For a team building one list, a source that starts refusing is an afternoon lost. For a platform whose whole promise is that the list builds itself, it is an architectural problem.
Zenrows has emerged as a go-to solution for many of our users, including myself, simply because it’s so powerful. What sets Zenrows apart is its exceptional ability to bypass advanced anti-scraping measures.
The layer underneath
Clay ships several ways to get a page, from a native scraper to Claygent, its own agent. Zenrows is the one their team reaches for when the source defends itself, and it does not sit beside the platform: Clay runs Fetch inside its own scraping layer, for the targets the native scrapers came back empty on.
The same capability is in the table too, as a column. A user drops it in, points it at a URL another column produced, and gets clean content back in the cell, on their own Zenrows credentials if they want them.
https://crunchbase.com/organization/northwind{ "funding_stage": "Series B", "raised": 42000000 }That is the whole interface, and it is why this scales across a customer base rather than across one team. There is no scraper to build for a new source, and none to repair when that source changes.
One of the things I love about Zenrows is that you don’t need to configure it on your own. You can simply use it through Clay.
What their users build
The column is general, so what it feeds is whatever the play needs. Four of the things Clay users put on top of it:
- Lead enrichmentCompany sites read for job title, tech stack, and product detail, straight into columns.
- Funding researchRound and investor data pulled from public pages for outbound targeting.
- Hiring signal trackingCareers pages read for open roles, used to prioritise growing accounts.
- Profile completionRecords filled in where the usual providers came back empty.
Clay’s own walkthrough puts a number on what the second of those is worth. Researching ten startups, their broad-web agent missed that one had closed a Series C. The Crunchbase page, read through Zenrows, had it. On a list, that is one row that was wrong and no one would have checked.
Tasks that previously took 2 to 3 weeks now complete in just a few minutes.
Conclusion
A Clay user who needs a protected source adds a column. The question of how to reach it never reaches them, and it does not come back the next time that source changes its mind.
Reduced data preparation from 3 to 6 hours per list to mere seconds.