Skip to main content
Back to AI in Action

Contractor Pricing Intelligence

Eliminating Margin Erosion Through Real-Time Pricing Intelligence Across 10 Branches

Industry

HVAC & Plumbing

Scale

$195M

Duration

20 Weeks

Location

Nashville, Tennessee

Engagement

AI Consulting

Executive Summary

The VP of Commercial Sales at a 10-branch HVAC and plumbing distributor in Nashville discovered that the same 3/4" ProPress fitting sold for 11.4% less in Chattanooga than in Nashville — to comparable contractors, on comparable jobs, from reps using the same Epicor Prophet 21 pricing matrix. When she pulled transaction data across all 10 branches, the pattern was systemic: pricing on the 500 highest-velocity commercial SKUs varied by 6-14% between branches, driven by rep-level discounting habits that had developed over years with no visibility or correction mechanism. We embedded AI-driven pricing intelligence directly into the quoting workflow on Prophet 21.

Business Impact

+190bp

Gross margin improvement across commercial accounts

37%

Reduction in branch-to-branch pricing variance on comparable transactions

19%

Faster average quote response time

0

Contractor accounts lost during the transition

The Situation

The distributor served mechanical contractors across Tennessee and Northern Alabama — commercial HVAC projects for hospitals, schools, office buildings, and multi-residential, plus residential new construction and service/repair. Commercial work through 32 reps at 10 branches represented 64% of revenue.

Pricing on the 500 highest-velocity commercial SKUs varied 6-14% between branches — not because of market differences, but because of discounting habits that had developed rep by rep, branch by branch, over years without centralized visibility.

  • The Prophet 21 pricing matrix assigned customer tiers (A through E) with corresponding multipliers by product category. In theory, a Tier B mechanical contractor buying copper fittings paid the same price at every branch. In practice, 24 of the 32 reps had developed individual pricing habits that overrode the matrix — verbal commitments to contractors (“I always give you 5% off on Viega”), branch-level exceptions that were never documented, and blanket discount approvals from branch managers who prioritized volume over margin
  • The 8 most experienced reps — averaging 12+ years with the distributor — had the strongest contractor relationships and the deepest discounting habits. They also generated 55% of commercial revenue. Telling them to "stop discounting" was politically impossible and commercially dangerous
  • When a contractor worked across multiple branches — a mechanical contractor based in Nashville pulling permits in Chattanooga and Knoxville — they experienced different pricing on identical products from the same distributor. Two contractors had raised this directly with the VP, describing it as unprofessional
  • Margin erosion wasn't visible at the branch level because branch managers tracked total gross margin percentage, not transaction-level variance. A branch running 28% overall margin could have individual transactions ranging from 18% to 36% — with the low-margin transactions invisible in the average
  • Quote response time on complex commercial projects (50+ line items across pipe, fittings, valves, and equipment) averaged 2.1 days. Reps built quotes manually in Prophet 21, applying their personal pricing judgment line by line. The fastest competitor — a regional plumbing specialist — was responding in 4 hours on comparable quotes
  • No system analyzed win/loss patterns against pricing levels. The reps who discounted most aggressively believed they were winning more business. The data — once we analyzed it — showed their win rates were statistically identical to the reps who discounted least

The pricing problem wasn't that reps were making bad decisions. It was that they were making thousands of pricing decisions per week with no structured intelligence about what worked and what didn't — and no visibility into what their colleagues at other branches were charging the same contractors for the same products.

The Challenge

The VP had tried to address pricing discipline twice. In 2022, she tightened the Prophet 21 approval thresholds — any discount below a defined floor required branch manager sign-off. Within 3 months, branch managers were batch-approving exceptions because the volume of requests was unmanageable. The policy added friction without adding intelligence.

In 2023, she introduced a quarterly pricing review where each branch manager presented their margin performance. The reviews revealed the variance but couldn't fix it — by the time a branch manager saw that one rep had been underpricing ProPress fittings for 3 months, thousands of transactions had already gone through at eroded margins.

  • The approval workflow treated every below-floor transaction the same — a $200 fitting order got the same friction as a $40K project quote. Branch managers rubber-stamped the small ones to keep the queue moving, which meant the policy caught nothing
  • Quarterly reviews were retrospective — they identified margin problems months after the pricing decisions were made, when the money was already gone
  • The reps who discounted most weren't doing it maliciously. They had developed mental models of "what this contractor expects to pay" that were 3-5 years out of date. Cost structures had shifted. Competitive dynamics had changed. But the rep's pricing instinct hadn't updated because nothing in their workflow told them it should
  • The VP needed a solution that made better pricing the path of least resistance — not a policy that punished reps for doing what they'd been doing unchecked for years

The Solution

We spent 4 weeks in discovery analyzing 30 months of transaction-level pricing data across all 32 reps and 10 branches — 1.4 million commercial transactions.

The analysis revealed patterns the VP had suspected but couldn’t prove:

Discovery Findings

  • 6 of the 32 reps accounted for 61% of the total margin variance. Their pricing habits weren’t random — they consistently underpriced specific product categories (copper fittings, ProPress, and commercial valves) regardless of customer tier or job size
  • Win rates showed no statistical correlation with discount depth. Reps who priced at matrix won at the same rate as reps who discounted 8-12% below matrix — meaning the discounts were reducing margin without increasing volume
  • Contractor-level elasticity varied dramatically. Some contractors were genuinely price-sensitive and would shop competitors on a 3% difference. Others hadn’t compared pricing in years and were receiving unnecessary discounts based on rep assumptions about their price sensitivity
  • The pricing inconsistency on multi-branch contractors wasn’t just unprofessional — it was creating a race to the bottom. Contractors who discovered the Nashville branch charged more than Chattanooga demanded the lower price at both locations

The system analyzed signals including:

  • Transaction-level pricing data across 32 reps, 10 branches, and 30 months — identifying which reps, which product categories, and which customer segments showed the largest variance from optimal margin
  • ML-modeled contractor-level price elasticity — distinguishing contractors who would leave over a 3% increase from contractors who wouldn’t notice a 5% adjustment, based on purchasing behavior, competitive alternatives in their market, and historical price response patterns
  • Win/loss correlation analysis identifying the pricing levels, response times, and product categories where price actually determined the outcome versus where speed, availability, or relationship determined it
  • Competitive pricing intelligence from bid outcomes — reconstructing competitor pricing on lost quotes to build a market-level pricing map by product category and geography
  • Cross-branch contractor profiles ensuring that any contractor purchasing from multiple locations received consistent pricing regardless of which rep or branch processed the order

The data showed that the distributor’s most aggressive discounters weren’t winning more business — they were winning the same business at lower margins. The reps pricing at matrix had statistically identical win rates.

Implementation

Deployment occurred over a 01 – 05 period.

Dynamic Pricing Guidance Engine

ML model generating real-time pricing recommendations at point of quote based on contractor elasticity, product category margin targets, competitive positioning, and historical win/loss patterns.

Branch Pricing Alignment

Automated detection and flagging of cross-branch pricing inconsistencies on the same contractor, ensuring pricing coherence without requiring manual coordination between branch managers.

Contractor Elasticity Profiles

ML-scored price sensitivity classification for every active commercial account, distinguishing contractors who require competitive pricing from those receiving unnecessary discounts.

Margin Visibility at Point of Quote

Real-time margin impact displayed on every line item during quote construction — the rep sees the dollar impact of their pricing decision before submitting, not in a monthly report afterward.

Prophet 21 Integration

Pricing intelligence surfaced within the existing quoting workflow — no separate system, no additional login, guidance appears on the same screen where the rep builds the quote.

Strategic Impact

Margin Recovery Without Commercial Disruption

+190 basis points of gross margin recovered across commercial accounts in 18 weeks — without raising prices on price-sensitive contractors and without losing a single account. The improvement came from eliminating unnecessary discounting on contractors whose purchasing behavior showed no price sensitivity, and from aligning branch-to-branch pricing so that multi-location contractors received consistent, defensible pricing. The VP: "We didn't charge anyone more. We stopped giving away margin to contractors who never asked for it and wouldn't have left if they didn't get it."

Pricing Consistency as Professionalism

Branch-to-branch variance on the 500 highest-velocity SKUs dropped from 6-14% to 2-4% — the remaining variance reflecting legitimate market differences between Nashville metro and rural Tennessee. Two contractors who had previously complained about inconsistent pricing independently told their reps they'd noticed the improvement. The VP considered this the most important leading indicator: "When your contractors trust your pricing, they stop shopping every quote. That's worth more than the margin improvement."

Quote Speed as Competitive Weapon

Average quote response time on complex commercial projects dropped from 2.1 days to 17 hours. The ML-generated pricing recommendations eliminated the line-by-line mental calculation reps performed on every quote — they reviewed the system's recommendations, adjusted where their judgment dictated, and submitted. The 19% speed improvement was a byproduct of the pricing intelligence, not a separate initiative.

Key Takeaway

In HVAC distribution, pricing variance between branches isn't a pricing strategy — it's a pricing accident that compounds across thousands of transactions until the margin damage shows up in the quarterly P&L, months after the decisions that caused it. This distributor's reps weren't underpricing because they were bad at their jobs. They were underpricing because nothing in their workflow told them what the same product was selling for at the branch 90 miles away, what the contractor's actual price sensitivity was, or whether the discount they'd been giving for 5 years was still earning them anything. The system didn't take their discretion away. It gave them the information to use it well.