Pricing experimentation with AI (L4)

The steps

1. 01

   Audit Historical Deal Performance

   Begin by extracting a comprehensive dataset from your CRM (Salesforce, HubSpot) covering the last 24 months of closed-won and closed-lost opportunities. Use a SQL query in your data warehouse or a custom CRM report builder. Required fields include: 'Deal Amount (Gross)', 'Final Discount %', 'Competitor Name', 'Outcome (Won/Lost/DQ)', 'Lead Source', and 'Company Industry/Segment'. • Tooling: Use Salesforce Data Loader or HubSpot's 'Export' feature to CSV. • Setting: Filter for 'Close Date' within the last 730 days. • Data Cleaning: Ensure you map 'Competitor' flags correctly (e.g., if the field is empty but it was against a known rival, label it 'Competitor: Unknown'). Owner: Sales Ops or Data Analyst. Time: 4-6 hours for extraction and cleaning. Prerequisite: Admin access to CRM. Pitfall: Ignoring 'No Decision' outcomes; filter those out as they skew win rates unrelated to pricing. Definition of Done: A clean CSV with at least 500 records, where every row has a discount percentage and a competitor tag.

2. 02

   Generate Discount Bands via AI Analysis

   Feed your cleaned CSV into an AI tool like ChatGPT (with Advanced Data Analysis), Claude 3.5 Sonnet, or a dedicated ML platform like Pecan.ai. Your goal is to identify how discounts correlate with win rates when competitors are present versus absent. • Example Prompt: 'I am uploading a CSV of 24 months of deals. Analyze the relationship between Discount % and Win Rate, segmented by Competitor Presence. Identify the "Sweet Spot" discount range where win probability plateaus or increases significantly without sacrificing more than 5% margin.' • Analysis Goal: Do not look for a single number. You are looking for 'Discount Bands' (e.g., 'Against Competitor X, a 12-18% discount is optimal; Solo deals should stay under 5%'). Owner: RevOps Manager. Time: 2 hours. Pitfall: Relying on averages. Averages hide the fact that a 50% discount might have a lower win rate than a 10% discount due to perceived low product value. Definition of Done: A documented table showing recommended discount bands per competitor or per segment.

3. 03

   Embed Guidance into the Sales Workflow

   Instead of a static '10% max' rule, implement logic in your CRM or CPQ (Configure Price Quote) tool that serves these AI-driven recommendations to reps in real-time. • Implementation: Create a custom field in Salesforce/HubSpot called 'Target Discount Band'. • Logic: Use an IF/THEN formula or a simple automation (Flow/Workflow) that looks at the 'Competitor' field. If Competitor = 'Rival A', set the Target Discount Band to '12% to 18%'. • User Interface: Ensure this field is visible on the Opportunity page layout right next to the 'Amount' field. This provides 'Just-in-Time' guidance to the rep before they draft a proposal. Owner: CRM Admin. Time: 3 hours. Prerequisite: Discount bands from the previous step. Pitfall: Making the recommendation a hard limit too early; keep it as 'Guidance' first to ensure rep adoption. Definition of Done: A rep can select a competitor on a deal and immediately see a suggested discount percentage range on their screen.

4. 04

   Formalize the Deal Desk Exceptions

   Establish a formal approval workflow for any discount that falls outside the AI-recommended band. This prevents 'discount creep' while allowing flexibility for strategic accounts. • Ruleset: In your CRM Automation (e.g., Salesforce Flow), trigger an 'Approval Process' if [Proposed Discount] > [Range Max]. • Approval Chains: If the discount is 1-5% above the band, route to Sales Manager. If 5%+, route to Deal Desk or CFO. • Requirement: Force the rep to input a 'Reason for Variance' in a text field (e.g., 'Multi-year upfront payment' or 'Strategic logo'). Owner: Deal Desk or Finance. Time: 4 hours. Pitfall: Over-complicating the approval chain, which slows down deals. Keep it to max 2 levels of approval. Definition of Done: An automated notification (Slack/Email) is sent to a manager whenever a rep requests a discount higher than the AI-suggested band.

5. 05

   Monitor Combined ASP and Win Rate Metrics

   Track the impact of this new pricing strategy using a combined dashboard. You must measure Average Selling Price (ASP) and Win Rate in the same view,measuring them in isolation is dangerous. • Dashboard Setup: Create a scatter plot where the X-axis is 'Average Discount' and the Y-axis is 'Win Rate'. Segment by 'Competitor'. • Key QA Check: If ASP goes up but Win Rate drops drastically, your AI bands are too aggressive. If Win Rate is 100% but ASP is flat, your bands are too generous. • Cadence: Review this monthly. Feed the new data back into the AI (Step 2) to 'retrain' the bands every quarter. Owner: RevOps / VP of Sales. Time: 2 hours per month. Pitfall: Declaring victory too early. Pricing changes often take 2-3 months to show true impact on the sales cycle. Definition of Done: A live dashboard showing 'Effective Discount Rate' (Total Discount / Total Gross Value) trending downward while 'Win Rate' remains stable or improves.

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)