What prospects want. Where calls break. What to say next.

The funnel shows exactly what was available, what contained real dialogue, what remained after deduplication, and what the model actually analyzed.

Call-level outcomes describe what happened on the phone. CRM outcomes describe the later commercial deal state. Both matter; they answer different questions.

Prioritized from the full corpus. Confidence labels reflect evidence coverage and the difference between direct observation and inference.

Clusters are counted by unique call, with a representative verbatim quote that survived exact matching.

Frequency alone is not enough. Each cluster shows how often the objection was handled, mishandled, or left unanswered.

Question clusters separate answered, partial, and unanswered moments. The answer bank below uses only approved knowledge; gaps remain visibly unresolved.

These are evidence-led associations, not causal claims. Use them as coaching hypotheses to test.

Each priority is defined as an observable behavior, a practice drill, and a measure—so this can become operating cadence, not report decoration.

Natural, short responses grounded in the current PTD canon. “Needs owner” means the evidence was insufficient and the system refused to guess.

Drafts for ads, landing pages, WhatsApp, and email. Each carries its evidence and any proof or approval still required.

Open a sequence to copy the subject, body, and CTA for each step. Review brand voice and live offer facts before sending.

Coverage, evidence gates, and known limits are part of the result—not footnotes to hide.