On-demand webinar

Introducing Specter 1.0: Account scoring without the black box

Your ABM platform says an account is in the Decision stage, and sales asks, “Based on what?” 81.6% of marketers can’t reliably explain why an account’s score or stage changed. Josh and Andrew walk through Specter 1.0, which stages accounts with a machine learning model trained on real buyer journeys, then uses a language model to factor in sales calls and explain which signals set the stage.

Added Nov 5, 2026 · 22 min watch

What we cover

What you’ll take away

Reps can’t do much with an account score alone. This session covers where surges and DIY scoring break down, and how Specter tells sales why an account is in its stage and who to contact.

  • Why account-level surges leave reps guessing

    A surge is often one person doing research, and reps get the account without knowing who that person is.

  • Where DIY scoring hits its ceiling

    Hand-weighted points and LLM grading aren’t tied to outcomes, and LLMs need heavy prompt work to keep labels stable.

  • Why ABM scoring needed two models

    One model stages accounts by their odds of opening an opportunity. The other reads sales calls to refine that stage.

  • How Specter compares

    Against 6sense, Demandbase, and LLM-only scoring. Even an optimally tuned hand-weighted model scored worse.

0:00
-0:00