assess

A market map is being built for {{anchor}} — {{sells}} Its buyer: {{buyer}}

The markets it sells into, which are the only legal values for a query's market field: {{capabilities}}

Copy one of those strings character for character onto every query you write. A query tagged with a market that is not on that list detaches everything it finds, because the graph hangs each host on the market whose query surfaced it.

So far, after {{waves}}: {{hosts}} distinct hosts from {{asked}} queries. Saturation: {{saturation}} — past roughly 70%, more of the same shape buys corroboration, not coverage; steer the next round at ground the sample below does not already hold, or say enough.

Angles already worked (intent · platform): {{angles}}

A sample of what came back: {{sample}}

What each family has bought so far, and what remains unreleased:

{{families}}

Every query opens reading two pages of results. Where a product has bought both, here is what page 2 added beyond page 1:

{{pageYield}}

A product whose page 2 kept finding hosts page 1 did not is a market this run has not finished reading at this depth — name it in deepen and its queries read four pages instead of two from here on. A product whose page 2 mostly repeated page 1 is already saturated at this depth; naming it would spend real money on rows already seen.

Template queries still held in reserve, per product:

{{reserve}}

Widening is per product, per family. A product whose queries keep returning the same hosts is a door already walked through — draw a DIFFERENT family or its next term from reserve. A family returning fresh hosts on a product is paying — release more of its reserve. A product whose whole hand came back thin gets more, taken from nobody. Ask for reserve releases in draw (product + how many), and write fresh debranded queries only for gaps no template can reach.

Is this enough to show someone as a map of this market, or is something obviously missing? Think about the facets a complete map holds: companies going head-on, things that solve the same problem a different way, what this is built on and what plugs into it, and where these buyers actually gather and talk.

If something is missing, write queries aimed at THAT — not more of what has already been asked. Every rule from before still binds: never name the anchor, describe what things do, and each query must ask something the others do not.

You are deciding whether to spend more of the run's money. Saying enough when the map has an obvious hole wastes the whole run; saying it is not enough when the last wave taught you nothing spends real money buying corroboration of what you already know. Judge the map you have, not the number of queries you have asked.