Next-best-action engine for AEs (L5)

The steps

1. 01

   Build a codified Play Library

   Begin by cataloging the 'plays' your Account Executives (AEs) can actually execute. An 'action' is not just 'follow up'; it must be a specific, discrete event. Brainstorm 15-25 named plays such as 'Executive Multi-threading (VP+)', 'Send Custom ROI Calculator', 'Technical Gap Analysis', 'Request Mutual Action Plan (MAP)', or 'Discount for Quarter-End Close'. • Open a spreadsheet or Airtable. Columns should include: Play Name, Description, Channel (Email/Call/LinkedIn), Cost (in minutes of AE time), and Historical Lift (estimated impact on win rate). • For the 'Historical Lift', look at your CRM data from the past 12 months. Compare the win rate of opportunities where a specific event occurred (e.g., 'Executive Meeting') vs. those where it didn't. • Owner: RevOps Manager. • Time Estimate: 4-6 hours. • Pitfall: Being too vague. 'Follow up' is not a play. 'Send 3rd-party Case Study relevant to Industry X' is a play. • Definition of Done: A finalized library of 20 categorized plays with assigned estimated win-rate boosters.

2. 02

   Engineer the EV engine

   To rank actions, you need a formula to calculate Expected Value (EV). You should create a calculated field in your CRM or an external BI tool (like Tableau or Looker) using this logic: EV = [Probability of Close] × [Estimated Deal Value (ACV)] × [Play Lift %] - [Opportunity Cost of Play]. • Create a custom object in Salesforce called 'NBA_Recommendations'. • Use a tool like Census, Hightouch, or a Python script to sync this calculation daily. • Example Formula for 'Send ROI Calc': if ACV is $50k, Current Win Prob is 20%, and ROI Calc Lift is 5%,the EV might be ($50,000 * 0.20 * 0.05) - (30 mins AE time value). • Owner: Sales Analyst / Data Engineer. • Time Estimate: 8-10 hours. • Pitfall: Over-weighting high-ACV deals. Ensure the formula doesn't just recommend actions for the biggest deals while neglecting winnable mid-market ops. • Definition of Done: A dashboard or CRM view showing the top 3 recommended plays for every open opportunity, ranked by EV.

3. 03

   Deploy AE-facing surfaces

   The engine is useless if AEs have to hunt for it. Embed the recommendations directly into the opportunity record page in your CRM (Salesforce Lightning Component or HubSpot Sidebar). • Configure the view to display: 1) The Play Name, 2) The predicted % lift in win-probability, and 3) An 'Update' button to mark the play as 'Executed', 'Dismissed', or 'In Progress'. • Set up an automated Slack or Teams notification via Zapier or Tray.io that pings the AE at 8:00 AM every Monday with their top 5 'High EV' plays across their entire pipeline. • Owner: CRM Administrator. • Time Estimate: 3-5 hours. • Prerequisite: The EV Engine must be active and populating data. • Pitfall: Cluttering the UI. Only show the top 3 recommendations to avoid 'choice paralysis'. • Definition of Done: AEs can see recommended actions directly on their deal view without clicking through multiple tabs.

4. 04

   Operationalize play-based coaching

   Shift the culture from 'interrogation' to 'coaching' by restructuring 1:1 meetings. Managers should no longer ask 'Where does this deal stand?',the CRM already says that. Instead, the manager asks 'Why did you choose to skip the ROI Calculator play for the Acme Corp deal?' • Update the 1:1 meeting template in your performance management tool (e.g., Lattice, 15Five, or a shared Doc). • Focus the agenda on: 1) Play Execution Rate (target >80% of top recommendations), 2) Feedback on play quality, and 3) Identifying 'unwinnable' deals where no play provides high EV. • Use a prompt for Managers: 'I see the system recommended a Multi-threading play for this account. What blocker is preventing us from executing that?' • Owner: VP of Sales and Sales Managers. • Time Estimate: Ongoing (1 hour per AE per week). • Pitfall: Managers falling back into old habits of asking for 'gut-feel' updates. • Definition of Done: Manager 1:1 notes reflect play-based coaching rather than stage-based status updates.

5. 05

   Automate the feedback loop

   AI-driven systems require a feedback loop to remain accurate. You must track which recommended plays actually led to won deals. • Build a report in your BI tool that correlates 'Play Executed' status with 'Opportunity Won' status. • Every Sunday night, run a script or SQL query that re-weights the 'Lift' factor in your EV formula based on the last 30 days of performance. • SQL Logic: SELECT Play_Name, COUNT(Opp_ID) as Tries, SUM(CASE WHEN Stage = 'Closed Won' THEN 1 ELSE 0 END) as Successes FROM Sales_Data GROUP BY Play_Name. • If the 'Discount' play is no longer moving the needle, its EV should drop. If 'Peer Reference' is winning every time, its priority should rise. • Owner: RevOps Analyst. • Time Estimate: 2 hours weekly. • Pitfall: Small sample sizes making the data noisy (wait for at least 30 instances of a play before significantly changing its weight). • Definition of Done: A weekly automated update to the 'Historical Lift' variable in the EV engine.

Next playbooks

• Voice-of-Customer → Product & Enablement Loop (L2)
• Zapier Agents for sales ops chores (L2)
• Agentic Nightly CRM Hygiene + Ownership Sync (L3)
• AI-driven price-quote acceleration with CPQ (L3)