← all playbooks

Turn any web page that lists companies or people into a prospect list

The page as a structured list: every name extracted, companies resolved to their own domains, the list qualified against your desk, and the decision-makers at the ones that fit, each with a work email. The source link stays on every row.

use when
you have found a page that names the companies or people you want, such as an awards shortlist, a top 100 ranking, a membership directory or an accelerator cohort, and it is only a web page
starts from
A list

The prompt

Paste it into Claude Code or the Claude desktop app with Hyreflow connected. The first line tells your agent to use Hyreflow, so it reads the play, asks before it spends anything, and hands the work back to you.

paste this into Claude
Use hyreflow and turn this web page into a prospect list I can work.

The page: <PAGE_URL>
It lists: <COMPANIES_OR_PEOPLE>, as far as I can tell.
The people I sell to: <DECISION_MAKER_TITLES>. My specialism: <SPECIALISM>.

- Read the page first and tell me what it holds: companies or people, how
  many, and whether it runs over several pages. Public pages only. If it
  sits behind a login, or the list is an image or a PDF, tell me and stop.
  Do not work around it.
- Extract every row, keep whatever the page says about each one, and give
  me the count before you spend on anything else.
- Resolve each company to its real website and drop the duplicates.
- Qualify against my ideal client before contacts. Here it is:
  <PASTE YOUR ICP, OR NAME YOUR BEST CLIENTS>.
- For the ones that pass, find the decision-makers and a work email. If
  the page lists people and the person named is senior enough, that
  person is the target. This is business development, so work email.

Run five rows end to end first and tell me what the full run costs before
you do the rest. Do not contact anyone.

Replace every <PLACEHOLDER> with your own detail. Everything else can stay as written.

What you need first

  • The link to a public page that lists companies or people. No login, no paywall
  • A description of your ideal client, so the list is cut before contacts are bought
  • The titles you sell to, for pages that list companies
  • A Hyreflow workspace with credits

Tools it can reach for

The agent picks per step from what your workspace has. Nothing here is required by name.

What happens when you run it

Free steps are marked free. Anything that spends credits is marked, and the agent asks before the first paid run of any size.

  1. 1

    Read the page and say what it is

    credits

    The page is rendered and read once. You are told whether it lists companies or people, roughly how many, which details each entry carries and whether the list runs over several pages. A page behind a login, or a list published as an image or a PDF, is reported at this point and the play stops.

  2. 2

    Extract every row

    credits

    Entries are pulled into rows against a fixed shape: name, website or profile link where the page gives one, and the page's own words about each, such as the award category, the rank or the cohort. Lists that run over several pages are mapped first, then read page by page.

  3. 3

    Count and remove duplicates

    free

    Rankings and directories repeat entries across categories. Duplicates collapse to one row that keeps every category it appeared under, and you see the real size of the list before any further spend.

  4. 4

    Resolve companies to their own websites

    credits

    Pages link to profile pages and social accounts as often as to the company's site, and many give a name only. Each company is resolved to its official domain by web search and checked, because every later step keys off the domain. For a page of people, the employer printed beside each name is resolved the same way.

  5. 5

    Qualify against your desk

    free

    Size, sector, geography and your exclusions are applied to what is already in hand, then the rest is judged against your description in the conversation. It costs nothing, so it runs before a single contact is looked up.

  6. 6

    Find the decision-makers

    credits

    For companies, a company-scoped search on each domain for the titles you sell to, with the title filter kept on. For a page of people, a named person who is senior enough is the target already, and the search is skipped.

  7. 7

    Get the work email and check it

    credits

    The work-email waterfall runs in a set order and stops at the first hit, and a miss usually costs nothing. Addresses can be deliverability-checked before the list is handed over. Nobody is contacted.

What a run costs

Credits are spent per company the play actually works, and a lookup that finds nothing usually costs nothing. The two figures are the run where the first provider answers and the run where every lookup walks its full chain.

companiesif the first provider answersif every lookup walks the chain
25$2.121 credits$11110 credits
100$8.181 credits$42420 credits
500$40400 credits$2102100 credits
1,000$80800 credits$4204200 credits

Free before anything is charged

  • Count and remove duplicates
  • Qualify against your desk

What moves the number

  • The channel. This play buys work email, because these are people you are approaching about work you want to win. A work address is the cheaper of the two to find.
  • Coverage on people search and work email. The chain stops at the first provider that answers, and only that provider bills.
  • How many companies survive the free filters. Everything dropped before the paid steps costs nothing.
  • Providers you connect with your own key. Those calls bill your account, not your credits.

An estimate, not a quote, priced at the volume credit rate. Your agent sizes the run against your own workspace and tells you what it will cost before it spends anything.

Somebody already built your list, and published it as a web page

Every week you scroll past a list that is better targeted than anything a database filter returns. The shortlist for your sector's awards. A ranking of the fastest-growing firms in your region. The member directory of a trade body. This year's cohort at an accelerator. The guest archive of the podcast your buyers listen to.

Each one was compiled by somebody with a reason to get it right, and each one carries a reason to call that a database row never has: they were shortlisted, they were ranked, they were admitted. What the page lacks is everything a recruiter needs next. There is no domain, no size, and nobody named who owns hiring.

This play closes that gap for any page. Two plays are the same method applied to one kind of page, and they go further on their own ground: work a conference exhibitor list adds the timing and the pre-event approach, and map a market that lives on a public register covers whole markets and the structured registries behind them. Use this one for everything else.

What you get back

A CSV with one row per entry: the name as printed, what the page said about it, the official domain, the tier it landed in and why, and the link to the page it came from. For the entries that qualified, the decision-makers with title, LinkedIn where there is one, and work email.

Above it, three numbers: how many entries the page held, how many fit your desk, and how many came back with a named contact.

Variations worth knowing

A page of people. Speakers, guests, judges and award winners are named individuals. A senior one is the target, and what the page says about them is the opening line. The employer still gets qualified.

The page lists candidates, not buyers. A rising-stars list can be a sourcing list. Say so at the start, because the channel changes to personal email and LinkedIn, never the work address.

Last year's page too. Rankings and cohorts are annual. Reading two editions shows who is climbing, and who has been on the list long enough that every supplier has already called.

Where this goes wrong

The list cannot be read. Logins, images, PDFs and pages that forbid collection all stop the extraction. The play says so and does not invent rows.

Name to domain is ambiguous. Two companies share a name, or the page uses a brand and not the legal entity. Doubtful matches are listed for you before anything is spent on them.

Treating the list as qualified. It was qualified for the publisher's purpose. Skip your own cut and you pay for contacts at companies you would never call.

Writing to everyone on a famous list. Every supplier has seen the same ranking. "Congratulations on making the list" is the line they have all received.

Questions

What kinds of page does this work on?

Any public page that prints a list as text: an awards shortlist, a ranking, a trade body's member directory, an accelerator's cohort page, a podcast's guest archive, a supplier list, a portfolio page. The test is simple. If you can select the names with your cursor without signing in, it can probably be read.

What happens with a page behind a login, or a list that is a PDF or an image?

It is reported and the play stops. A login means the publisher did not intend the list to be public, and that puts it out of scope. Image-only lists and PDFs read unreliably, and an unreliable extraction produces a confident-looking list with wrong names in it. Paste the names into the conversation yourself and everything after the extraction runs as normal.

Is it allowed to read the page?

The play reads public pages and respects what a site says about automated access. Where a page's terms forbid collection it is flagged and left alone. Nothing is sent by this play in any case. It ends at a list.

Why qualify before finding contacts?

Because somebody else chose this list for their own reasons. A ranking of fast-growing companies was not compiled for your desk, and most rows will not fit it. Qualifying is free and finding contacts is not, so the cut comes first and you only pay for the companies you would call.

Does reading the page cost anything if it turns out to be unusable?

Yes, a little. A page read is charged per page whether or not the page holds a usable list, and a blocked page can still be charged. That is why the first read is a single page, and why a blocked page is not retried.