Understanding Account Propensities

Prev Next

AI-Based Account Prioritization Model

An AI-driven solution that surfaces the accounts most likely to convert, enabling sales and marketing teams to focus resources on high-value prospects and accelerate revenue generation.


1. Model Overview

The Account Prioritization Model assigns each account two complementary propensity scores—Base Fit Propensity and 3-Month Propensity—that quantify long-term and short-term conversion likelihood, respectively. By ranking accounts on these dimensions, teams can:

  • Quickly identify which accounts best match your Ideal Customer Profile (ICP).

  • Forecast near-term pipeline and bookings, enabling agile, data-driven outreach.

  • Allocate resources to high-impact opportunities and deprioritize lower-potential accounts.


2. Data Inputs & Feature Sets

  1. Demographic & Firmographic Attributes
    • Industry, region, company size, existing customer status, and any custom ICP fields.

  2. Historical Performance
    • Past conversion rates, stage-by-stage pipeline volumes, and booking outcomes.

  3. Engagement & Activity Metrics
    • Website visits, campaign interactions, content downloads, email/call logs, number of leads per account.

  4. Temporal Features
    • Date components (day, month, quarter, year) derived from account and activity timestamps.


3. Propensity Scores

3.1 Base Fit Propensity

A measure of an account’s inherent compatibility with your ICP over the long term.

  • Heuristic Mode (low-data scenarios): Applies rule-based scoring when < 2 years of history are available.

  • Machine Learning Mode (sufficient history): Trains an XGBoost classifier on ≥ 2 years of closed-won vs. closed-lost data, using firmographic and demographic features to predict an account’s baseline conversion likelihood.

3.2 3-Month Propensity

A short-term forecast of an account’s likelihood to generate pipeline or bookings in the next quarter.

  • Model: XGBoost classifier trained on recent activity and pipeline metrics.

  • Key Inputs:

    • Current pipeline counts and stage volumes

    • Activity frequencies and diversity (e.g. emails, calls, event touches)

    • Lead-to-account ratios and campaign touch counts

    • Temporal signals indicating seasonality or recent momentum


4. Scoring & Categorization

Each account receives four core scores:

Score Type

Definition

Pipeline Fit Propensity

Long-term likelihood of generating new pipeline

Booking Fit Propensity

Long-term likelihood of closing bookings

3-Month Pipeline Propensity

Short-term (next 3 months) likelihood of generating pipeline

3-Month Booking Propensity

Short-term (next 3 months) likelihood of closing bookings

Scores are normalized to a 0–1 range and then bucketed into:

  • High – Top quintile (best fit and strongest momentum)

  • Medium – Middle three quintiles

  • Low – Bottom quintile

  • Deprioritize – Significantly below average or actively declining


5. Structured Output

The model produces a ranked list of accounts with associated metadata:

Account Name

Industry

Region

Pipeline Fit

Booking Fit

3-Month Pipeline

3-Month Booking

Bucket

Acme Corp

Manufacturing

EMEA

0.82 (High)

0.75 (High)

0.68 (Medium)

0.61 (Medium)

High

Beta LLC

Financial Svcs

APAC

0.54 (Medium)

0.47 (Low)

0.35 (Low)

0.28 (Low)

Medium

Gamma Inc

Technology

Americas

0.22 (Low)

0.18 (Low)

0.12 (Deprior.)

0.09 (Deprior.)

Deprioritize


6. Operationalizing Insights

  • Immediate Follow-Up: Assign high-bucket accounts to SDRs/AEs for outreach within 24–48 hours.

  • Nurture Campaigns: Develop targeted content and drip sequences for medium-bucket accounts.

  • Resource Reallocation: Shift effort away from deprioritized accounts—use capacity to expand into new market segments or support high-potential territories.