← all playbooks
Sourcing

Map a whole talent pool, including the half that is not looking

A talent map of the addressable market for a role: who is where, which employers are carrying a bench worth raiding, who is signalling that they are open, and a qualified shortlist pulled out of all of it.

use when
the role is hard, the obvious search is exhausted, and you need to prove to a client that you have covered the market

The prompt

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

paste this into Claude
/hyreflow-recruit

I need a full talent map for <ROLE_TITLE> in <LOCATION>, radius <KM> km.

<PASTE THE BRIEF OR SPEC>

Leave no stone unturned. Source from every angle, not just one search:
1. A structured search on titles and skills.
2. A semantic search, for the people whose profiles describe the work
   without using the standard title.
3. A competitor talent map. Work out who the real competitors are, tier
   them, and search inside each. Tell me which employers are carrying more
   than a couple of matches, because those are the ones worth raiding.
4. A lookalike expansion from the strongest five people you find.
5. Anyone signalling that they are open to work.

Merge everything, de-duplicate, and keep a column showing which angle each
person came from. Then qualify the whole pool against the brief, which is
free, and only spend on contact details for the ones that pass.

Peek the counts before each search and tell me what the full run will cost.
Deliver the map plus a shortlist of <N>.

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

What you need first

  • A role brief or job spec
  • A location or region, with a radius if the commute matters
  • A Hyreflow workspace with credits
  • Optional: seed companies or a competitor list to expand from

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

    Turn the brief into a search

    free

    Titles, skills, seniority band and location are extracted and checked against what each data source will actually accept, so a search does not silently return nothing because of an unrecognised value.

  2. 2

    Size the pool before spending

    free

    Some people databases report how many rows match a filter without charging for them. The agent reads those counts first and comes back with the size of the market and a projected cost for the full pull, flagging the sources that can only be counted by paying for a page of results.

  3. 3

    Search structurally and semantically

    credits

    The structured pass catches everyone whose title and skills match. The semantic pass catches the ones who describe the same work in different words, which is where most of the missed talent lives.

  4. 4

    Build the competitor talent map

    credits

    Competitors are discovered, tiered, and searched one by one. Where several strong matches cluster at one employer, that employer is flagged as a bench worth raiding rather than a single name to chase.

  5. 5

    Expand on lookalikes

    credits

    The strongest few people become the template. Their employers, keywords and career shapes feed a second round of searching that finds people the original filters were never going to reach.

  6. 6

    Mine the open-to-work signal

    credits

    People who have flagged that they are open get their own pass, paired with the title and location filters so it stays a targeted search rather than a dredge of the whole database.

  7. 7

    Merge, then qualify for free

    free

    All five angles collapse into one de-duplicated list that keeps the lineage of every row. Qualification against the brief costs nothing, so it runs across the entire pool before any further spend.

  8. 8

    Spend only on the survivors

    credits

    Work history, any proof-of-work research and finally personal contact details are bought for qualified candidates only, capped at the size of shortlist you asked for.

When a single search is not enough

Most sourcing is one search, run well. That is the right tool for most roles. This play is for the ones where it is not: the retained search, the role that has been open for four months, the client who wants to see the market before they believe your shortlist.

The five angles are deliberately chosen to fail differently. Structured search is precise and blind to anyone who words their profile unusually. Semantic search reads meaning and drifts. Competitor mapping is high signal and narrow. Lookalike expansion finds the shape you did not think to describe. Open-to-work finds intent and nothing else. Run one and you get a slice. Run all five and the gaps stop overlapping.

What you get back

Two artefacts. The map: pool size, the distribution across employers, the tiers, the benches worth raiding, and where in the geography the skills actually sit. And the shortlist: the qualified subset with contact details, ranked, with the reasoning kept.

Reading the map

Concentration matters more than the total. A market of nine hundred people spread evenly is a different job from nine hundred people where a hundred and twenty sit at four employers. The second one tells you where to spend your week.

The angle column is the interesting one. People who show up in three of the five passes are the obvious market and everyone is calling them. People who only surfaced semantically are the ones your competitors' Boolean strings did not reach.

A thin pool is a finding, not a failure. If the map comes back small, the honest answer to the client is that the brief is not addressable as written. That conversation is worth more than a padded list, and you now have the numbers to have it.

Where this goes wrong

Running the paid angles before qualifying. The qualification step is free and the per-person steps are not, so anything that spends on the pool before cutting it is money set on fire. The play orders it correctly, but if you rewrite the prompt, keep that order.

Treating an empty result as a zero. A provider that is unavailable returns nothing, which looks identical to a market with nobody in it. The play distinguishes the two rather than folding an outage into your market size.

Confusing coverage with reachability. Mapping a thousand people does not mean a thousand people can be contacted. Coverage falls off at every stage and always will. The map is the market; the shortlist is what you can actually work this week.

Questions

How is this different from just running a bigger search?

A bigger search on the same filters returns more of the same kind of person. The angles here fail in different directions on purpose: structured search misses the people who describe their work unusually, semantic search misses the people with sparse profiles, competitor mapping misses anyone outside the obvious set, and open-to-work misses the passive majority. Running them together is what turns a search into a map.

What is a bench, and why does it matter?

When several people who all fit your brief sit at the same employer, that company is carrying a bench in your discipline. It is a better target than any individual name on the list: one relationship, one approach angle, several possible candidates, and a strong signal about where the skills are concentrated in your market.

Does open to work mean they are actually looking?

It means they said so at some point. It is a real signal and worth a pass of its own, but it is the weakest kind of intent on the map and it skews towards people who are already being contacted by everyone else. Treat it as one angle among five, which is exactly how the play uses it.

Can I give this to a client as a market map?

That is the main reason to run it. You get pool size, distribution by employer, where the concentrations sit and what the realistic reachable subset looks like. It is the evidence behind a retained pitch, and it is the same artefact whether or not you go on to place the role.

This sounds expensive. How do I keep it under control?

Two things do most of the work. The counts are read before any search is paid for, so you approve the spend knowing the size of what you are buying. And qualification is free, so the pool gets cut down before the expensive per-person steps run. Cap the shortlist size in the prompt and the funnel narrows to fit it.