Case Study Skeleton: Lead Qualification Agent for a B2B Services Firm
Status: SKELETON. Sections marked [PILOT DATA] are populated only with measured results from an actual engagement — never estimated.
Dry-run evidence (simulated). Our deterministic pilot simulator already exercises this exact agent stack end-to-end: over a 14-day synthetic window it scored 42 inbound leads → 14 hot / 14 warm / 14 cold, wrote an auditable outcome record for every decision, and made zero outbound contact — the no-contact rule is enforced in code, not in the prompt (full dry-run report available on request). No real prospects were involved.
The client
B2B professional-services firm, ~40 inbound inquiries/week across web form, referral emails, and event lists. One founder doing all triage between billable work. [CLIENT NAME PENDING CONTRACT]
The problem
- Hot leads wait 2–3 days for a first reply while the founder is delivering work — many go cold
- Cold/no-fit inquiries consume the same attention as serious buyers
- No record of why a lead was pursued or dropped, so follow-up is inconsistent
What we deployed
A Lead Qualification Agent running our standard loop, wired through:
- a lead-source adapter (forms, inboxes, exports) — swappable without core agent changes
- a transparent scoring model: every lead gets a hot / warm / cold tier plus an explicit reason, not a black-box number
- no-contact enforcement in code: the agent's action space physically cannot include outreach — it qualifies and routes, humans decide who to contact
- an auditable outcome record per lead (score, reason, routing decision, loop cycle) so nothing is dropped silently
- instrumented lessons: repeated no-fit traffic from one channel surfaces channel-quality recommendations instead of being silently absorbed
Results framework (fill during/after pilot)
| Metric | Before | After | Source |
|---|---|---|---|
| Median time-to-first-response on hot leads | [PILOT DATA] | [PILOT DATA] | lead timestamps |
| % leads triaged without founder touch | — | [PILOT DATA] | outcome records |
| Hot-lead conversion to conversation | [PILOT DATA] | [PILOT DATA] | CRM / inbox |
| No-fit leads consuming sales time | [PILOT DATA] | [PILOT DATA] | outcome records |
| Cost per qualified lead | [PILOT DATA] | [PILOT DATA] | token ledger |
Why the no-contact guarantee matters
Most "AI SDR" tools automate outreach — and burn your domain reputation and relationships with it. Ours does the opposite half of the funnel: it reads, scores, explains, and routes. The decision to contact a human being stays with a human being, by construction.
Lessons learned section
[POST-PILOT: list top 3 lessons from data/loops/<project>/ — real ones, verbatim.]