Battery manufacturing and dealer networksFractional AI + RevOps LeadershipShippedAI

Commissions That Close Rep-for-Rep, and a Chatbot That Stopped Leaking Prices

A lithium-battery manufacturer selling through about 3,000 dealers and about 100 OEM partners, with inside and outside sales teams across six territories · Five-year engagement, ongoing

Salesforce
QuickBooks
Shopify
Chatbase
+1

15% → 90%of revenue attributed to a rep in one day

15% → 90%of revenue attributed to a rep, in one day

Rep-for-repcommission close tie-out to the sales head's sheet

77% → 3%chatbot dead links in three weeks

89territory rules caught before zeroing a rep's book

The Challenge

Commissions lived in the head of sales' spreadsheet. Lead routing was hardcoded by state. 92% of 2,401 leads had never been touched. There was no dealer lifecycle status, so "dormant" lists were full of accounts that had never ordered. The public support chatbot was leaking dealer pricing and sending customers to dead links. Forty-plus AI seats on the CRM had been licensed and never turned on.

The CEO's goal was blunt: "I need everyone to learn how to use AI more."

Our Approach

Run as a fractional operator: a dated handoff after every session, with a "corrections" section for anything we got wrong
Mockup first, then build, with mandatory pauses after schema discovery and field mapping
Parallel-run every financial calculation against the owner's sheet until it tied
Audit the bot the client's team had built rather than replace it

The Solution

A commissions engine: nightly invoice sync, rep and territory stamping, monthly accrual freeze, four plan types, and a one-page close runbook
A ZIP-based territory engine with gated, idempotent cutovers and a live map
Dealer lifecycle status as a zero-maintenance formula
A shadow-account sweep with a snapshot before every merge, and a daily job that flags new duplicates
The public support chatbot audited and hardened: source cleanup, pricing guardrails, a regression set, and a weekly audit
A company-wide AI adoption program alongside the build work

Results

The first commission close inside Salesforce tied to the sales head's sheet rep-for-rep.
Invoice attribution went from 15 percent to 90 percent of the year's revenue in one day.
A per-rep overpayment defect and a six-figure account misclassification caught before payout.
1,795 of 3,077 dealers revealed as never having ordered.
986 of 988 duplicate contact clusters and 187 of 188 retail groups merged, each with a snapshot taken first.
Chatbot dead links fell from 77 percent to 3 percent in three weeks and pricing leaks went to zero.

This story continues in One Loop a Week: Field Service Dispatch and Command Centers Inside the CRM.

Where AI does the work

The public support bot (built by the client's team, audited and hardened by us) answers product and warranty questions. The adoption program puts AI tools in every department's hands with guardrails.

Where rules do the work

Every dollar. Commission math, territory stamping, accrual freezes, lifecycle status, duplicate merges, cutover gates, and the watchdog are deterministic and reconciled against the owner's numbers.

The mechanisms behind it

Data You Can TrustShipped

Duplicate accounts merged with a snapshot, then caught daily

Duplicate accounts and contacts a sync kept creating, grouped by address and ID keys, merged with snapshots and human approval, and flagged daily after that.