Revenue AI Strategy
Open the platform
All posts
L5 Maturityrevops-data 6 min read

Next-Best-Action Engine for AEs: The L5 Playbook for RevOps

Transform AE performance from gut-feel to algorithmic precision with a Next-Best-Action engine. Learn to build an EV-based play library and automate coaching.

Run the playbook

Next-best-action engine for AEs (L5)

For every open opp, the system recommends and ranks the next action by expected value. Managers manage to the plays, not to dials.

Why this matters

The "gut-feel" sales culture is a liability. In most scaling B2B organizations, Account Executives (AEs) spend their days intuitively choosing which opportunities to prioritize, often falling into the trap of the "squeaky wheel" (the loudest prospect) or the "comfort zone" (the easiest, least impactful activities).

When you manage by gut, your forecasting suffers from an average of 25-30% "slippage"—deals that were supposed to close this month but push to the next. The cost of this uncertainty is massive: stalled hiring plans, missed quarterly targets, and inefficient CAC.

A Next-Best-Action (NBA) Engine replaces intuition with Expected Value (EV). It shifts the management paradigm from "What's the status of this deal?" to "Why haven't we executed the highest-value play yet?" Companies that adopt this level of algorithmic rigor typically see a 15-20% increase in win rates and a significant reduction in sales cycle length by eliminating low-value activities.

How it works

The transition to an L5 NBA Engine requires moving from descriptive analytics (what happened) to prescriptive analytics (what to do next).

Step 1: Build a Codified Play Library

Stop calling everything "follow-up." A play must be a discrete, measurable event. Using a tool like Airtable, catalog 15–25 specific actions. For example:

  • The Multi-thread: Getting a VP-level stakeholder on a call.
  • The ROI Build: Sending a custom ROI calculator (validated via a tool like ValueCloud).
  • The Gap Analysis: A technical deep-dive with an SE.

Look at your last 12 months of CRM data. If opportunities that feature a "VP Meeting" win 45% of the time, while those that don't win 15%, that play has a 30% lift factor.

  • Time Estimate: 4–6 hours.
  • Definition of Done: A library of 20 plays with assigned win-rate boosters.

Step 2: Engineer the EV Engine

Calculate the Expected Value (EV) for every potential action. The formula is: EV = [Close Prob %] × [ACV] × [Play Lift %] – [AE Opportunity Cost]

If a $100k deal has a 20% baseline probability and an "Executive Multi-thread" play has a 10% lift, the raw EV is $2,000. Subtract the "cost" of the AE's time (e.g., $150 for 2 hours of work) to get the net EV.

Use Census or Hightouch to sync these calculations from your data warehouse (Snowflake/BigQuery) back into a custom "NBA_Recommendations" object in Salesforce.

  • Time Estimate: 8–10 hours.

Step 3: Deploy AE-Facing Surfaces

Don't make AEs log into Tableau. Embed the top 3 recommendations directly into the Salesforce Lightning record page or the HubSpot sidebar.

Use Zapier or Tray.io to push a daily "Daily Top 5 Deals" digest into Slack at 8:00 AM. Each recommendation should include the "Why"—e.g., "Executing the 'Technical Gap Analysis' play will increase this $50k deal’s win probability by 12%."

  • Tool Callout: Tools like Momentum.io can take these recommendations and automatically create Slack channels for high-EV deals.

Step 4: Operationalize Play-Based Coaching

In 1:1s, Managers must stop asking for status updates. The CRM already shows the stage. Instead, use an Office Hours or Lattice template focused on play execution.

  • Manager Prompt: "The engine recommended a 'Mutual Action Plan' play for the Acme Corp deal four days ago. It has an EV of $4,500. Why hasn't it been initiated?"
  • Culture Shift: You are now managing to the "Play Execution Rate" (target >80%) rather than just "activity" or "dials."

Step 5: Automate the Feedback Loop

The engine must learn. Every Sunday, run a SQL query to see which plays actually correlated with "Closed Won" in the last 30 days. If the "Discount for Close" play is no longer working due to a market shift, its Lift % should drop automatically.

  • Advanced Tip: Use Fathom or Gong transcripts processed through Claude Code to verify if a play was actually executed (e.g., did the AE actually mention the ROI calculator on the call?).

Tools you need

  • Data Layer: Snowflake, BigQuery.
  • Reverse ETL: Census or Hightouch (to push EV data to CRM).
  • CRM: Salesforce or HubSpot.
  • Automation: Zapier, Tray.io, or Momentum.io.
  • Intelligence: Claude 3.5 Sonnet (for calculating lift from historical unstructured data).

KPIs to track

  • Win Rate: This should be your primary North Star.
  • Slippage Rate: The % of deals that move out of the expected close month.
  • Plays-per-opp Executed: The volume of high-EV actions taken per deal.
  • Play Execution Rate: Percentage of "Top 3" recommendations actually completed by AEs.

Common pitfalls

  • The "Jumbo Deal" Bias: If not calibrated, the formula will always tell AEs to work on their biggest deal, even if it’s a long shot. Ensure your formula accounts for "time-to-close" to keep the volume flowing.
  • Vague Plays: If a play is "Build Relationship," it cannot be measured. If it’s "Connect with 3 Stakeholders on LinkedIn," it can.
  • UI Overload: Never show more than 3 recommendations. More than that creates choice paralysis and AEs will ignore the widget entirely.

When to graduate to the next level

You are ready for the next level (L6) when your engine suggests the specific content to send within the play (e.g., using Clay to find a specific mention in an annual report and Lindy to draft the email) and moves from manual "execution" to "auto-pilot" for low-stake plays.

nbaforecastingev

Ready to ship it? Open the playbook

Next-best-action engine for AEs (L5)

Step-by-step instructions, the tools to use, and the KPIs to watch — already wired into the Revenue AI Strategy workspace.

Open playbook