Late-stage negotiations eroded margins — CRM + TMS assistant stabilizes margin outcomes

Overview

A national freight brokerage’s business development team negotiated with shippers without a clear view of margin impacts in the moment. Reps switched between spreadsheets, the Transportation Management System, and market sites, then made concessions that looked reasonable but eroded margins when carrier costs came in. Intelligex integrated real-time cost models and historical lane data into a negotiation assistant embedded in the existing workflow, with guardrails and manager approval prompts for deeper discounts. Negotiations stayed within healthy bands, last-minute escalations tapered, and end-of-month margin surprises gave way to predictable outcomes backed by a shared audit trail.

Client Profile

• Industry: Freight brokerage and third-party logistics (3PL)
• Company size (range): Mid-market with nationwide shipper coverage
• Stage: Established brokerage modernizing pricing and deal governance
• Department owner: Sales & Business Development (BD Operations)
• Other stakeholders: Central Pricing, Carrier Procurement, Operations, Finance, Legal, RevOps, IT/Security

The Challenge

Reps negotiated spot and mini-bid opportunities against firm shipper timelines. They had partial context: a few recent loads in the lane, a static fuel surcharge sheet, and an anecdotal sense of carrier capacity. Market signals lived in separate portals, historical costs sat in the TMS, and margin targets were kept in a policy document that wasn’t at hand during calls. Concessions offered to win freight sometimes left little room after fuel, accessorials, and likely carrier buy costs. Pricing escalations arrived late and approvals were inconsistent by region and manager, leading to uneven outcomes and tension between BD and Operations.

The business could not pause growth or replatform critical systems. The TMS remained the operational source of truth, and the CRM housed account and pipeline context. Pricing analysts were stretched and could not sit in every negotiation. Leadership asked for a way to surface lane history and real-time cost drivers at the point of negotiation, guide reps to healthy margin bands, and trigger manager review for deeper concessions—without slowing down shippers or adding new systems to learn.

Why It Was Happening

Data and policy were fragmented. Historical lane data, tender acceptance patterns, and dwell information lived in the TMS. Market references were checked in browser tabs. Fuel surcharge tables were updated separately and passed around by email. None of it was joined to the quote object or presented as a unified cost model. As a result, reps negotiated by feel, applied blanket discounts, and hoped procurement could cover the spread.

Governance was reactive. Margin floors and approval thresholds were documented but not enforced in the flow of work. Reps escalated by chat or email when deals turned urgent, which left managers with little context and forced rubber-stamp approvals. When end-of-month financials landed, Finance found margin leakage that no one intended but that the process allowed.

The Solution

We embedded a negotiation assistant into the existing CRM and quoting flow that combined historical lane performance with real-time cost drivers. The assistant calculated a dynamic cost baseline for each lane, applied policy-driven margin bands, and highlighted the impact of concessions before an offer was made. If a proposed rate fell outside healthy ranges, the system prompted a manager approval with full context. The TMS, CRM, and market sources stayed in place; the assistant orchestrated them into a single, governed experience.

• CRM-side quoting panel that ingests lane details and shipment attributes from Salesforce opportunities and quotes
• TMS integration to pull recent load history, carrier buy rates, tender acceptance, and service exceptions for the same or similar lanes
• Market references via rate benchmarks and indices, such as DAT’s shipper-facing benchmarks (DAT RateView)
• Fuel surcharge modeling tied to the U.S. Energy Information Administration weekly diesel price index (EIA Diesel Prices)
• Rule-based cost model that normalizes historicals, applies seasonality factors, and incorporates accessorial defaults based on shipment profile
• Margin bands by account tier, mode, and lane characteristics with visual guardrails and target cues
• Manager approval prompts for rates outside policy, using CRM approval workflows with reason codes and audit trails (Salesforce Approvals)
• Scenario planning that previews gross margin and risk notes for negotiation alternatives
• Exception queue for policy overrides with human-in-the-loop review and clear expiration windows
• Dashboards for BD, Pricing, and Finance showing negotiated bands, approvals, and post-award performance against the model
• Role-based permissions and field masking for sensitive carrier and customer data

Implementation

• Discovery: Mapped the quote-to-award process across CRM and TMS, including how reps sourced market signals, how approvals flowed, and where data gaps created guesswork. Cataloged historical data fields needed to model costs by lane and mode, and documented current margin policies and exceptions.
• Design: Defined the cost model inputs and hierarchy: recent lane history, similar-lane analogs, market benchmarks, and fuel. Established margin bands by product, account tier, and service level. Designed approval triggers, reason codes, and audit fields. Outlined user experience for in-call negotiation with minimal clicks.
• Build: Connected CRM quotes to TMS lane history and cost data. Integrated market reference feeds and the EIA diesel index. Implemented the cost model and guardrails as a service that returns recommended ranges and notes. Configured CRM approvals with context packs and Slack notifications. Added a scenario planner and a save-to-quote action that freezes the basis for later audit.
• Testing and QA: Ran the assistant on historical opportunities to compare recommended ranges with actual outcomes. Tuned the model to reduce noise on low-volume lanes and seasonal flares. Confirmed that approval prompts triggered correctly and that audit records captured the inputs and policy version used.
• Rollout: Launched in observe-only mode where the assistant suggested ranges without blocking rep choices. After stakeholder sign-off, enabled guardrails and approvals for selected segments, then expanded across modes and regions. Kept a manual path for urgent or atypical freight with manager acknowledgment.
• Training and hand-off: Delivered concise walkthroughs for reps and managers on reading bands, using scenarios, and submitting approval requests. Provided playbooks for exceptions and guideline updates. Assigned model stewardship to Pricing and RevOps, with documented change control and review cadences.

Results

Reps negotiated with immediate visibility into the likely cost to cover and the margin impact of concessions. Healthy offers were easier to craft, and risky discounts stood out before they reached a shipper. Approvals for deeper concessions carried context, so managers could make consistent decisions without slowing the deal. The negotiation assistant became a common frame of reference for BD and Pricing, which reduced back-and-forth and late escalations.

End-of-month reviews moved from forensic reconciliation to variance discussions grounded in the same model. Finance saw steadier gross margin performance because policies were enforced in the moment rather than reconstructed later. When lanes behaved differently than the model predicted, the feedback loop updated assumptions, so guidance improved without changing the core systems reps used.

What Changed for the Team

• Before: Reps toggled between tabs and spreadsheets to guess at a rate. After: A single panel surfaced lane history, market context, and margin bands.
• Before: Approvals came through ad hoc chats with limited context. After: Guardrails prompted structured approvals with reason codes and audit trails.
• Before: Discounts were applied uniformly regardless of lane risk. After: Scenario planning showed the margin impact and guided targeted concessions.
• Before: Pricing analysts fielded urgent calls during negotiations. After: Analysts focused on exceptions and model stewardship instead of routine cases.
• Before: Month-end uncovered margin leakage by surprise. After: Deals tracked to expected bands with clear explanations for controlled exceptions.

Key Takeaways

• Put cost and margin context in the negotiation flow, not in separate reports or after-the-fact reviews.
• Lane history, market benchmarks, and fuel indices form a practical, transparent basis for dynamic cost models.
• Guardrails and approval prompts work best when they are visible, explainable, and tied to business policy—not hidden algorithms.
• Keep TMS and CRM in place; layer a negotiation assistant and governance so reps can move quickly without sacrificing margin discipline.
• Close the loop by updating model assumptions from real outcomes to keep guidance relevant through seasons and disruptions.

FAQ

What tools did this integrate with?
We embedded the assistant in Salesforce for quoting and account context, connected to the Transportation Management System for lane history and carrier buy data, and referenced market benchmarks such as DAT RateView for external context. Fuel surcharges were tied to the U.S. Energy Information Administration diesel index. Approvals and notifications ran through Salesforce and Slack, and analytics used the existing business intelligence stack.

How did you handle quality control and governance?
Margin bands and approval thresholds were expressed as policy in CRM metadata and version-controlled. Each negotiation saved the inputs used—lane history, benchmark references, and fuel assumptions—along with the policy version and approver, creating an auditable record. Exceptions flowed to a human-in-the-loop queue with expiration windows, and model changes followed a documented change process owned by Pricing and RevOps.

How did you roll this out without disruption?
We began in observe-only mode, showing recommended bands while allowing existing practices to continue. After validating accuracy with stakeholders, we enabled guardrails and approvals for a subset of accounts and modes, then expanded coverage. The manual path remained available for urgent freight, with manager acknowledgment to maintain speed.

Did this replace the pricing team?
No. The assistant automated routine guidance and made policy visible. Pricing analysts focused on complex lanes, atypical freight, and continuous improvement of the model. Their expertise shaped the guardrails, and they approved exceptions with better context and less rework.

How were market fluctuations and seasonal shifts handled?
The model blended recent lane history with market benchmarks and updated fuel indices. When conditions shifted, the model weights and assumptions were adjusted by stewards, and changes were versioned. The system highlighted lanes with low data density or unusual variance and routed them for closer review to avoid overconfidence.