Deep research on a prospect
A deeper pass reads further into a candidate that a first pass left undecided. It is routed rather than offered to everything: the rule is that a deeper read runs only where it can change a decision. A candidate already clearly in or clearly out gains nothing from more reading, so the money goes to the ones sitting on a line, and to the ones whose only unknown is how to reach them.
The routing rule, and why it is not a setting#
The tempting design is a depth slider — shallow, normal, thorough — applied to every candidate in a run. It performs badly for an obvious reason: most candidates in any run are decided by the first pass, and paying for depth on them buys a more detailed route to the same answer.
So depth is decided per candidate, by whether more reading could move the verdict. The consequence is that two prospects from one run legitimately carry different amounts of research, and the thinner record is often the one the router was most confident about.
| Situation | Deeper pass? | What it is looking for |
|---|---|---|
| Sitting on the qualification line | Yes | A signal that settles fit either way |
| Qualified, no route to reach it | Yes | A published contact or enquiry route |
| Qualified with a route already found | No | Nothing further gets used |
| Clearly outside the criteria | No | The verdict will not move |
| A person asked for it | Yes | Whatever that person is about to decide |
What the deeper pass adds#
- More sources per claim, which turns a single attributed statement into something corroborated — or exposes that two sources disagree, which is itself useful.
- Dated material, where the first pass found only undated pages. A dated announcement is what makes an opening specific rather than generic.
- Contact routes the first pass did not reach: a published enquiry form, a branch page, a number on a page nobody links to from the front.
- Named people with sourced roles, recorded as inferences — see Finding the right person.
- Disqualifying facts, which are as valuable as qualifying ones and are found at the same cost.
It does not add certainty about things public material cannot settle. It cannot decide whether a business will buy, cannot produce a contact nobody published, and cannot promote an attributed claim into a fact. A deeper pass that finds nothing writes nothing, and the record stays thin.
Cost, and where it comes from#
The pass consumes the workspace's research allowance. A customer workspace is bounded by its plan; the Owner has no plan and no gates, which is the only difference between the two audiences here. When the allowance is spent, research stops and says so rather than accumulating a silent backlog — The prospecting allowance is spent describes what that looks like on screen.
There is a second cost that is easy to miss: time. A deeper pass reads more, so a prospect you requested one for is not ready in the same moment. On a list you are working interactively that is usually fine; on a list you are about to approve in bulk it is worth requesting before you sit down rather than during.
Reading the result#
- The verdict changed
- The pass did its job. The fit reason is rebuilt from the newer claims and shows what moved it.
- The verdict held, with more behind it
- Also success, and worth more than it looks: an approval built on corroborated claims is one you can defend later.
- Sources now disagree
- Both are kept, both visible. This is the signal to read rather than to approve, because the business presents itself two ways.
- Nothing new at all
- The population publishes little. Repeating the pass will not change that, and the honest next move is a different channel or a different search.
Whatever it returns, the freshness clock restarts on what it read. A record that has just been re-read is not queued for re-reading, which is how a workspace avoids paying twice for the same page in a week.
Questions#
Can I force a deeper pass on a specific prospect?
Yes — a person asking is itself a reason the router accepts, because you are about to decide something. That is the intended way to use it: one row you care about, rather than a whole list on principle.
Does a deeper pass ever remove a claim?
It can. If the source behind a claim no longer supports it, the claim stops being supported rather than quietly persisting. A record getting smaller after re-reading is the rule working.
Why did a prospect I care about get a shallow record?
Because the first pass answered the question. If you disagree with the verdict, asking for a deeper pass is the right move — the router optimises for decisions it can predict, and your disagreement is information it did not have.