Digital Companion

Vignettes

The book's vignettes show how recurring decisions and failure modes play out in practice. All are composite and anonymized; the patterns are real. Each one introduces a problem that the chapter and section named above it then work through: how it is overcome, how the opportunity is taken, and what value follows. The book labels each one simply "Vignette", so the companion names them by chapter and gives each a short title for finding it.

How you may use these vignettes

Free to use within your organization with a citation to the book and this site; please do not republish them. Using a vignette is your agreement to these terms.

The vignettes on this page are from CTRM for Bulk Liquids: A Practitioner's Guide to Strategy, Selection, Implementation, and Operation by Kent B. Landrum (Thin Leaf Press, 2026). They are provided for readers of the book and for practitioners working on CTRM strategy, selection, implementation, and operation.

You may use a vignette, in whole or in part, in training material, workshops, presentations, and working documents within your own organization, provided you cite the book and this site. You may link to any vignette; each title is a permanent link to it.

Please do not republish the vignettes, post them in full on public sites, or collect them into another work. Using a vignette is your agreement to these terms and to the site's terms of use.

45 of 45 vignettes

Chapter 2Strategy First: Commercial Objectives

Chapter 2, Vignette 1Level 1: Maximizing Asset Utilization

Two crude units on spreadsheets

A Gulf Coast refiner operates two crude units with different quality tolerances. Its commercial objective is to minimize crude input costs while maintaining unit stability. Traders are focused on sourcing barrels at favorable differentials, not on speculative positioning. The CTRM must track crude slates, quality attributes, and inventory positions accurately enough to inform procurement decisions. Risk reporting focuses on inventory exposure and hedge performance against the cost of crude. A spreadsheet environment is technically manageable at low volumes, but as nominations increase and pricing complexity grows, the risk of error in position tracking becomes the limiting constraint.

Chapter 2, Vignette 2Level 2: Optimizing Assets

Swapping barrels between pipelines

A Midcontinent crude producer has committed to space on two pipeline systems moving barrels to Cushing. When one system runs at a discount due to temporary congestion, the producer’s trader identifies the opportunity to swap barrels between lines, capturing the differential. The CTRM must support this activity with daily position reporting, exchange account integration for hedging, and enough flexibility to handle multiple deal types across physical and financial instruments. Reporting has moved from monthly to daily; risk management now requires an experienced mid-office analyst rather than a junior accountant.

Chapter 2, Vignette 3Level 3: Optimizing Margin

Storage plays in contango

A large independent refiner uses owned and leased tank capacity to execute storage plays when market structure shifts into contango. When forward prices rise above spot by more than the cost of carry, the trading desk purchases crude, stores it, and simultaneously sells it forward, locking in the spread. The CTRM must support multi-leg deal structures, track storage costs against forward sales, and provide daily P&L attribution that distinguishes storage-play returns from core refining margins. The risk manager monitors drawdown exposure daily. Trade surveillance is now in place. Implementation cost for the CTRM at this level begins at $1,000,000 and can scale significantly with geographic and functional scope.

Chapter 2, Vignette 41. The Asset-Backed Trading Refiner

Adjusting the crude slate for margin

Consider a Gulf Coast refiner that runs both light sweet and heavy sour crudes. The commercial strategy is to continually adjust crude slates to maximize margins while maintaining unit stability. When heavy crudes trade at a wide discount, the refinery may increase the proportion of those barrels in the run plan, provided the system can track resulting yields and account for higher sulfur content in refined products. Success is measured by refining margin visibility and by aligning physical operations with market opportunities. For this refiner, the business model is driven first by asset optimization, with trading activity serving as a hedge and a complement to physical operations.

Chapter 2, Vignette 52. The Rack Marketer and Blender

Pennies per gallon at the rack

A Midwest fuel marketer supplies both branded and unbranded product from multiple pipeline-connected terminals. Each day, the marketer balances index-based purchases with rack sales to fleets, municipalities, and retail operators. Small blending opportunities add incremental value, such as using ethanol to meet oxygenation requirements while improving netbacks. The business strategy is to capture pennies per gallon at scale. Profitability comes less from market timing and more from discipline in execution, ensuring that every rack sale reflects the correct tax treatment and blend composition. Success is defined by throughput volume and consistent, reliable margin capture rather than speculative trading gains.

Chapter 2, Vignette 63. The Logistics-Oriented Midstream Operator

Throughput, storage, and custody

Imagine a midstream company that owns a series of pipeline laterals feeding into a major distribution hub. Its revenue model is based on throughput fees and storage leases rather than market speculation. Each month, the operator must balance nominations, allocate space, and reconcile deliveries with customers. The commercial strategy is to maximize asset utilization, avoid imbalances, and maintain customer confidence in custody reporting. While the operator may occasionally engage in trading to balance exposures or optimize space, the model's heart is logistical efficiency. The firm succeeds when assets are full, contracts are honored, and custody reporting is accurate.

Chapter 2, Vignette 74. The Renewables-Compliance Hybrid

Credits held through the compliance period

A renewable diesel producer on the West Coast sells fuel across multiple jurisdictions, each with its own carbon-intensity scoring and credit requirements. Some credits must be retained for compliance; others can be sold in the open market to generate incremental revenue. The commercial strategy blends traditional trading with careful compliance management. The producer may choose to hold LCFS credits through the end of a compliance period if prices are trending upward, even if doing so limits short-term liquidity. Profitability is defined by both the margin on fuel sales and the effectiveness of compliance credit management. In this model, the CTRM must reflect compliance positions with the same clarity as trading positions, since both are integral to the business strategy.

Chapter 3Risk Appetite and Regulatory Exposure

Chapter 3, Vignette 1Inventory Risk and the Realities of Volatility

A loss the books saw a month late

A Gulf Coast fuel marketer holds roughly three million barrels of refined product across a network of owned and leased tanks at any given time. When crude prices dropped sharply during a market dislocation, the firm’s inventory positions resulted in a significant mark-to-market loss that did not appear in its accounting statements until the following month because the CTRM was configured to use weighted-average cost for all inventory valuation. Leadership had no real-time visibility into the economic exposure. The correction required a parallel revaluation process run in spreadsheets for several months while the system was reconfigured, during which the risk of error was material. The lesson: the inventory valuation methodology is a risk-appetite decision that belongs in the design conversation from day one, not a back-office detail to be resolved after implementation.

Chapter 3, Vignette 2Transfer Pricing and Intercompany Regulatory Exposure

Transfer pricing by legal entity

A North American downstream company establishes a trading affiliate in a favorable tax jurisdiction to consolidate its crude oil purchasing and product marketing activities. The affiliate buys crude, arranges processing at the parent company’s refinery under a tolling agreement, and sells refined products into multiple markets. Each of these intercompany arrangements—the crude supply agreement, the tolling fee, and the product offtake pricing—must be documented and priced at arm’s length. The CTRM must be able to produce, by legal entity, the volumes transacted, the prices applied, and the resulting margin attributed to each party. If the system records activity only at the consolidated trading book level, the organization cannot produce the legal-entity-level data that transfer pricing documentation requires, and any audit becomes a reconstruction exercise with significant exposure.

Chapter 4Assessing Business Capabilities and Organizational Readiness

Chapter 4, Vignette 1The Role of Peer Benchmarking

The assessment that reframed the business case

An independent refiner with a growing crude supply and products trading operation commissioned a capability assessment before beginning a CTRM selection process. The assessment revealed that the organization executed asset-backed trading scenarios daily but lacked daily position reporting, a formal trader checkout process, and a middle office function independent of the commercial team. Scheduling was managed in non-standardized spreadsheets, and inventory outside the refinery gate had no real-time visibility. Against the peer benchmark, the organization sat in the bottom quartile on risk management capability while operating at a Level 2 to Level 3 commercial sophistication. The assessment estimated a gap in margin capture of $50 million to over $100 million annually relative to what its peer group was achieving despite having better infrastructure. That finding reframed the CTRM business case entirely: it was no longer viewed as a technology cost to be minimized but as an investment in commercial capability with a concrete return. It also produced a clear sequencing recommendation: establish the risk governance foundation and daily reporting capability first, then select and implement the CTRM against a set of requirements that the organization now understood clearly.

Chapter 5Evaluating IT Strategy and Execution Capabilities

Chapter 5, Vignette 1Business-IT Alignment as the Foundational Condition

A Level 1 architecture for a Level 3 ambition

A mid-size refiner began a CTRM selection process with a clear commercial strategy: advance from Level 1 asset utilization to Level 2 asset optimization over three years, with an eventual path to Level 3. The commercial team had done the strategy work. The IT organization had not been part of that conversation. When the selection process began, IT produced an architecture recommendation that emphasized on-premises deployment, deep SAP integration at the deal capture layer, and a custom-built scheduling interface. Each of those choices was defensible in isolation, but together they reflected a Level 1 CTRM architecture applied to a Level 2-to-Level 3 commercial ambition. The mismatch became apparent only when the requirements document, written by the commercial team, described daily position reporting, derivative hedging workflows, and logistics optimization that the proposed architecture could not support without significant additional investment. A time-bound alignment exercise at the start of the program would have cost far less than the scope revision it ultimately required.

Chapter 5, Vignette 2Synthesizing the Readiness Picture: Closing Part I

Two assessments, one program design

A downstream marketer preparing for a CTRM selection completed both the business capability assessment described in Chapter 4 and the IT assessment described in this chapter. The business assessment revealed strong front-office trading capability but significant gaps in middle-office risk governance and in the maturity of back-office scheduling processes. The IT assessment revealed solid ERP integration experience but limited cloud governance capability, no formal master data management program, and an SDLC that had never been tested on a program of CTRM complexity. The combined picture was clear: the organization was commercially ready to benefit from a Level 2 CTRM capability but organizationally and technically ready to implement only a well-scoped, structured program with significant parallel investment in process and governance. The assessment directly shaped the vendor selection criteria, eliminating vendors whose implementation models assumed more organizational maturity than existed, and producing a program design that phased the cloud governance and master data investments ahead of the system go-live rather than in parallel. Eighteen months later, the organization went live on schedule, on budget, with daily position reporting and a functioning middle-office risk function that had not existed when the assessment began.

Chapter 6Understanding the Vendor Landscape

Chapter 6, Vignette 1The ERP-native option and its limits

The ERP-native path and the rack

A regional refined products marketer operating six terminals across two states ran its commercial operations on a combination of spreadsheets and a legacy CTRM that had not received meaningful development investment in nearly a decade. When the organization initiated a CTRM replacement project, the project sponsor, the CFO, had previously worked at an SAP-native manufacturing company and was drawn to SAP Commodity Management for its integration with the existing SAP ERP. The evaluation team spent three months building a business case for the SAP path before a functional assessment workshop revealed the gap. SAP Commodity Management could cleanly handle the organization's forward purchasing and hedge accounting needs. It could not handle rack pricing across multiple carrier types, terminal inventory reconciliation against actual meter tickets, or the excise tax determination logic required for fuel tax filings across two state jurisdictions. The organization ultimately selected a purpose-built CTRM and built a standard SAP integration for financial settlement and accounts payable, the outcome the CFO had wanted, reached by a different path.

Chapter 6, Vignette 2Vendor health, ownership, and the installed base question

A platform acquired mid-implementation

A mid-sized crude marketer selected a well-regarded CTRM platform roughly eight months before the platform's parent company was acquired by a private equity firm. The selection process had included strong reference calls, a competitive demo, and a well-structured implementation timeline. Eighteen months into the implementation, the client noticed delayed responses to support tickets and a product roadmap that had not been updated since the acquisition. Key development items the client had been promised as near-term releases, including a pipeline nomination module the business had been counting on, were de-prioritized without explanation. The implementation was completed, but the organization entered go-live on a platform whose development investment had effectively stalled. Within three years, it was evaluating replacements. The total cost of the implementation, support, and subsequent re-evaluation substantially exceeded the initial business case.

Chapter 7Requirements Definition for Bulk Liquids and Renewables

Chapter 7, Vignette 1The anatomy of a requirements document

The commodity matrix exposes the gap

A large integrated refiner and marketer spent three months building a requirements document before engaging vendors. The document was thorough in its functional requirements and included a commodity applicability matrix spanning 17 commodity types. During the initial vendor screening call, one prominent platform claimed full coverage across all functional areas. When the evaluation team applied the commodity matrix to those responses, specifically asking about production-deployed capabilities for ethanol blending, RIN generation, and leased crude, the answers changed materially. Three functional areas that had been marked as full coverage were reclassified to partial coverage or planned. The commodity matrix had exposed the gap that the flat functional questions had obscured. The platform was not eliminated from the evaluation, but its relative position shifted, and the demo was structured to test the specific commodity-function combinations where its coverage was least certain.

Chapter 7, Vignette 2Why requirements lists alone are not enough

The transmix scenario changed the selection

A midsize, refined products marketer structured its vendor evaluation around a requirements document and a standard one-day demo agenda. Three vendors responded with comparable requirements scores, and all three performed well in the demo on deal capture, basic scheduling, and reporting. The evaluation team was prepared to make a selection when an operations manager asked whether any of the vendors had demonstrated the transmix handling workflow, a niche but operationally significant process in which interface material of mixed product grade is accumulated, valued, and disposed of. None of the three demos had covered it. When the evaluation team returned to each vendor with a transmix scenario— purchase interface from a pipeline operator, assess quality and value at transmix pricing, sell to a transmix processor, and account for the inventory loss—only one of the three vendors was able to demonstrate the workflow end-to-end without significant qualification. The selection changed. The requirements document had listed quality management and inventory reconciliation as requirements, and all three vendors had claimed coverage. The scenario revealed what the list could not.

Chapter 8Integration with External Systems

Chapter 8, Vignette 1Market data and price curves

A silent market data failure

A refined products trading desk ran its end-of-day mark-to-market process on a scheduled job that pulled the afternoon Platts assessments at 5:45 PM and ran the valuation at 6:00 PM. On a Thursday in October, the Platts feed failed silently at 4:30 PM when the data vendor's API returned a timeout that the integration logged but did not escalate. The 6:00 PM valuation ran using Wednesday's prices. Crude had moved materially overnight. The risk committee received a position report that morning, understating the desk's net exposure by a figure large enough that it would have influenced two hedging decisions made before the error was caught at end-of-day reconciliation. The monitoring gap had existed for 11 months. No one had thought to specify a freshness alert as a requirement. It was added the following week.

Chapter 8, Vignette 2Terminal automation systems

The terminal interface that was never built

A mid-sized refined products marketer implemented a new CTRM and deferred the terminal automation system integration to a second phase, scheduled 90 days after go-live. The integration was technically specified and ready for development, but the project ran over budget, the vendor's integration team was unavailable, and Phase 2 was pushed to six months. Six months became twelve. Twelve became permanent. The TAS interface was never built. For the next three years, three schedulers manually entered meter ticket data into the CTRM each day from terminal reports received via email. The CTRM's inventory position was routinely 18 to 36 hours stale. Month-end inventory reconciliation required two full days of manual work because the CTRM records and the TAS records had to be matched transaction by transaction. The original project plan had projected that the TAS integration would reduce the month-end close time from two days to two hours. That projection was accurate. The integration never happened.

Chapter 8, Vignette 3Financial and settlement data

Eleven counterparty names

A large integrated marketer went live on a new CTRM with the ERP integration built and tested. On day three of production operations, the back office began receiving calls from counterparties reporting that invoices were not matching their purchase orders. By day five, the volume of invoice-matching failures had consumed the entire back-office team's capacity and delayed payment on 43 invoices, resulting in significant outstanding receivables. The root cause was a counterparty master data mismatch identified during the post-go-live investigation: 11 counterparties had been entered in the CTRM under slightly different legal entity names than they appeared in the ERP. The CTRM-generated invoices referenced the CTRM version of the name; the ERP's matching logic required the ERP version. The mismatch had existed in the test environment as well, but testing had used a simplified transaction set that did not exercise all 11 counterparties. Resolving the master data alignment required eight weeks of cross-system remediation. The project team had identified master data alignment as a pre-go-live risk during planning and had assigned it a medium priority. The correct priority was critical.

Chapter 9Cybersecurity and Risk in System Architecture

Chapter 9, Vignette 1The CTRM is a high-value target

Thirty-seven open vulnerabilities

A mid-sized, refined products marketer operating a CTRM on-premises had accumulated 16 years of custom integration code connecting the system to pipeline portals, a rack pricing service, an ERP, and a market data feed. A routine vulnerability scan, the first the organization had run that explicitly included the CTRM environment, identified 37 open vulnerabilities across the application and database servers, 14 of which were rated high or critical severity. Three of the custom integration components were found to be transmitting data over unencrypted connections using SSL protocols that had been deprecated for security reasons six years earlier. The CTRM vendor's compatibility requirements prevented the database from being upgraded to a version that would have addressed 11 vulnerabilities. The organization had not been aware of the scan exclusion. The CTRM had been producing audit findings on access controls for two years and was about to begin a replacement project. The vulnerability scan became an additional exhibit in the replacement business case and helped accelerate the project timeline by six months.

Further readingThe Cybersecurity Risk Posed By Your Outdated ETRM System (Article, JD Supra, 2021)

Chapter 9, Vignette 2Data residency, multi-tenancy, and supply chain risk

A patch that moved the nominations

An energy marketer migrated its CTRM to a hosted cloud environment as part of a broader cloud adoption initiative and configured automatic updates to be installed under the vendor's standard managed service agreement. Eight months after migration, the vendor released a patch that inadvertently modified the behavior of the automated pipeline nomination submission process, causing nominations to be submitted with incorrect date offsets for one pipeline's cycle-based schedule. The error was not caught during the vendor's internal testing because the specific pipeline configuration was not represented in the test environment. The marketer's schedulers identified the problem two days after the patch was applied, when pipeline confirmation data showed volumes nominated into the wrong cycle. Correcting the affected nominations required manual intervention by the pipeline operator and led to a scheduling adjustment that impacted 14 downstream movements. The marketer subsequently negotiated a contractual provision requiring advance notice of updates affecting nomination or scheduling functionality and 72-hour review window before production deployment. The vendor's standard managed service agreement had not included this provision.

Chapter 10RFP and Evaluation Framework

Chapter 10, Vignette 1Demonstration and bake-off design

The demo that skipped pipeline cycles

An evaluation team at a refined products marketer sat through two days of vendor demonstrations featuring polished dashboards, slick deal-entry interfaces, and impressive reporting visualizations. All three vendors demonstrated confidently. The scoring after the demos was close, with two vendors within four points of each other on a hundred-point scale. The team selected the higher-scoring vendor and began contract negotiations. Six months into implementation, the project hit a wall on pipeline scheduling. The vendor's system handled simple pipeline nominations adequately but had no native support for cycle-based pricing and scheduling on Colonial and Plantation. The demo had never covered a pipeline cycle scenario. The evaluation team had required vendors to claim coverage on pipeline scheduling in the structured requirements, and all three had claimed full coverage. What the team had not done was require any vendor to demonstrate coverage against an actual operational scenario. The implementation required 14 weeks of custom development to build the cycle management logic that the demo had implicitly represented as native. The implementation budget was exceeded by 40%. The evaluation design had produced the wrong answer.

Chapter 10, Vignette 2Reference checks

The unofficial reference call

A refined products marketer had completed its RFP scoring and demonstration process and was prepared to select the vendor that had scored highest across both dimensions. A project team member, following a suggestion from a former colleague at a peer company, reached out directly to a scheduler who had been running the same platform for three years. The call was not arranged by the vendor and was not on any official reference list. The scheduler's account was candid in ways that vendor-supplied references rarely are. The system performed adequately for basic refined products supply operations. The pipeline scheduling functionality had gaps for cycle-based operations that the organization had never been able to resolve through configuration; schedulers had maintained a parallel spreadsheet for two years to manage the workaround. The vendor's support team had been responsive in the first year after go-live but had deteriorated noticeably following a private equity acquisition 18 months later. Three development commitments made during the sales process had not appeared in any subsequent release. The scheduler concluded: the system does the job we bought it for, but we probably would not select it again today if we were starting over. The marketer extended its evaluation, issued a targeted cycle scheduling scenario to both finalists, and ultimately selected the vendor that had placed second in the original scoring. The reference call had changed the outcome.

Chapter 10, Vignette 3Vendor diligence and contractual terms

One version behind support

A midstream operator implemented a CTRM under a traditional perpetual license agreement. The implementation was successful, and the system went live on schedule. Two years later, the organization received notice that the vendor had released two new versions and that the maintenance agreement limited support to the current release and the one immediately preceding it. The organization was one version behind the support threshold. The upgrade project was scoped at four months and $800,000 in external implementation services. The vendor's standard agreement required that upgrade services be performed by the vendor's own professional services organization, not by the independent SI the operator had used for the original implementation. The vendor's professional services team was booked out for 18 weeks. The organization operated without vendor support coverage for two months, running a production trading system on an unsupported version, while it waited for the upgrade slot. Neither the version support policy nor the upgrade services restriction had surfaced as questions during the original evaluation. Both had appeared in the contract in standard boilerplate language that the legal review had not identified as unusual. The total cost of the upgrade, including delay, forced use of premium-rate vendor services, and internal staff time, was not modeled in the original business case. It should have been.

Chapter 11Project Governance and Delivery Models

Chapter 11, Vignette 1The three-tier governance structure

Six weeks without a business sponsor

A Gulf Coast rack marketer and bulk distributor staffed a steering committee with its CFO, VP of Supply, and CIO, with the implementation partner's engagement lead attending each meeting. For the first four months, the committee met monthly as planned. Then the VP of Supply transitioned to a new role, and a replacement was not designated for six weeks. During that interval, three significant design decisions involving the interaction between the trading system and the rack pricing engine were deferred rather than made, because no one on the delivery team had the authority to commit the supply organization. When the new business sponsor arrived and reviewed the backlog, two of the three deferred decisions required rework of the configuration that had been provisionally completed under incorrect assumptions. The rework consumed three weeks and required a schedule adjustment that pushed the derivatives go-live by a month. The steering committee had not failed dramatically. The gap had been an ordinary personnel transition. But the program had no governance mechanism to maintain business decision authority during leadership changes, and the cost was concrete and measurable.

Chapter 11, Vignette 2Internal ownership and client resource commitment

A quarter of the expert's time

An independent refiner implementing a new CTRM to replace a legacy trading system allocated roughly a quarter of its primary back-office subject-matter expert's time to the project. She had nearly two decades of experience with the legacy system and knew its data structures, month-end close procedures, and edge cases better than anyone else in the organization. During the design phase, her limited availability meant that functional specifications for the accounts payable and inventory accounting workstreams were drafted primarily by the SI, with periodic review sessions rather than collaborative development. When user readiness testing began, she encountered a posting rule configuration that reflected how a different downstream company had handled LIFO inventory accounting, not how her organization had always handled it. Correcting the configuration and retesting the affected scenarios required four weeks. The go-live slipped by six weeks, the last two consumed by remediation that should have been caught during design. The fractional allocation had saved approximately 40 person-weeks of internal time. The schedule delay cost more than that in SI fees, extended project overhead, and the operational risk of continuing to run on a system everyone had already mentally decommissioned.

Chapter 12Configuration vs Customization

Chapter 12, Vignette 1How technical debt accumulates in CTRM environments

Two hundred forty custom objects

A large independent refiner had implemented a CTRM for its crude, products, and derivatives business approximately six years before undertaking what was initially scoped as a straightforward version upgrade. Over the intervening years, the implementation team had added custom interfaces for two acquired terminal networks, a bespoke P&L attribution report that joined tables across three schemas, a custom approval workflow for deal entry that the platform's native authorization model had not supported at the time of implementation, and a collection of stored procedures that handled product quality differential adjustments the core pricing engine did not natively accommodate. None of these additions had been tracked in a central inventory. When the upgrade project team began its planning assessment, it took three weeks simply to enumerate the custom objects in the environment. The final count was over 240 items. The upgrade was rescoped from a six-month version migration to a 14-month program that included a parallel assessment of which custom objects the new platform version could replace with native configuration. Several items that the original development team had built as workarounds for known platform limitations had been addressed in standard product releases in the intervening years and could be retired. But the retirement process itself, mapping each custom object against its native replacement, validating equivalence, obtaining business sign-off, and regression testing the result, added months to the program. The project was completed successfully, but at a cost and timeline that bore little resemblance to the version upgrade it had been originally described as.

Further readingThe (Technical) Debt Harvest in an ETRM Upgrade (Article, LinkedIn, 2026)

Chapter 12, Vignette 2The customization decision framework

A methodology nobody needed anymore

A midstream marketer implementing a CTRM to support its pipeline scheduling and rack distribution business encountered a requirement in the fifth week of the design phase. The scheduling team needed the system to calculate a blended average cost for multi-grade batches that had been commingled in a common carrier pipeline segment, applying a specific allocation methodology the company had developed internally and used in its legacy spreadsheet environment for over a decade. The implementation team's initial assessment was that the platform's standard inventory valuation model did not align with the methodology and that a custom calculation engine would be required. Before approving the development, the program manager applied the decision framework. Question one revealed that the platform did have a configurable weighted-average-cost methodology that met the general requirement, but not the specific allocation sequence. Question two, directed to the scheduling manager and CFO together, established that the proprietary methodology had been developed to address a specific regulatory reporting requirement that had since been superseded. The organization continued to use it out of habit rather than necessity. The business sponsor authorized a process change to the platform's standard methodology. No custom object was created. The scheduling team adopted the new approach during user readiness testing with minimal friction. The implementation avoided a custom calculation engine that would have required dedicated maintenance at every subsequent upgrade.

Chapter 13Bridging the Gaps: When Core CTRM Isn’t Enough

Chapter 13, Vignette 1Authorization and approval workflows

The custom workflow that failed silently

A downstream marketing company implementing a CTRM as part of an asset-backed trading initiative built a custom approval workflow inside the platform during the initial implementation. The workflow handled deal escalation above the standard trader limit, routing deals to desk heads through a custom notification mechanism and recording approval status in a custom database table. The workflow functioned correctly from go-live through the first 18 months of operation. At the first major version upgrade, the vendor had restructured the deal header table that the custom notification mechanism read from. The workflow continued to function in most cases but silently failed to trigger approval routing for a specific deal type that used a new transaction subtype introduced in the upgraded version. The failure was discovered six weeks after the upgrade, during an internal audit that found 17 deals of that subtype had been executed without desk head approval, as required by the organization's limits of authority policy. None of the deals represented a commercial loss; all had been within reasonable market parameters. The audit finding nevertheless required a remediation report, an update to controls certification, and a commitment to replace the custom workflow with a governed solution. The replacement was a purpose-built workflow application integrated with the CTRM via its published API. It took four months to implement, test, and deploy. The total cost of the custom workflow, including the original build, the upgrade remediation, the audit finding response, and the replacement, substantially exceeded what a purpose-built solution would have cost if the correct architectural decision had been made at the outset.

Chapter 13, Vignette 2AP invoice matching and payables automation

The payables spreadsheet only one person understood

A large refined products marketer operating crude and products books across multiple refineries went live on a new CTRM with payables automation deferred to Phase 2. The first month-end close after go-live took 11 days to complete, driven primarily by the volume of unmatched payable items the back-office team had to investigate manually. By the third month, the team had established an informal spreadsheet-based matching process that partially automated the volume tolerance checks but required manual exception routing and lacked integration with either the CTRM or the ERP. The spreadsheet was maintained by the most experienced payables analyst on the team, who had built it herself over three months and was the only person who fully understood how it worked. Eighteen months after go-live, that analyst accepted a position at another company. Her replacement spent six weeks reconstructing the matching logic from the spreadsheet formulas and the informal documentation left behind. During that period, the month-end close was extended to 14 days, two invoices were paid incorrectly, and one carrier relationship was strained by a payment dispute that the investigation ultimately traced to a volume tolerance that the spreadsheet had handled correctly but that the replacement did not initially understand. The organization eventually deployed the purpose-built payables tool that had been in the Phase 2 backlog since go-live. Month-end payables processing stabilized at under two days. The cost of the 18-month gap was measured not just in the direct labor of the extended close cycles but in the institutional knowledge concentrated in a single person and the organizational exposure created when that person left.

Chapter 14Data Management, Testing, and Cutover

Chapter 14, Vignette 1The data management challenge: harmonization, not loading

A location hierarchy built on the wrong keys

A regional rack marketer and bulk distributor implementing a new CTRM to support its pipeline and terminal operations completed its location master data design in the design phase and loaded an initial set of locations into the development environment without a structured review against the terminal management system identifiers. The implementation team proceeded through configuration and into system integration testing before the scheduling workstream attempted to run end-to-end movement tests using real terminal feeds. At that point, a systematic cross-reference failure appeared: the CTRM location hierarchy had been organized around the company's internal regional structure, with terminals grouped under region nodes that did not correspond to any identifier in the terminal management system or the pipeline nomination interface. Every automated movement record arriving via the interface referenced a terminal identifier that did not exist in the CTRM hierarchy. The cross-reference tables that mapped external identifiers to internal location records had been built against the wrong hierarchy structure. Rebuilding the location hierarchy, updating the cross-reference tables, retesting all affected interfaces, and revalidating the configuration objects referencing location nodes required four weeks. Two integration test cycles were consumed in remediation work that should have been resolved in the design phase. The implementation team had correctly identified location hierarchy as a design decision. It had not identified the alignment between the internal hierarchy and external system identifiers as a validation requirement before loading master data into the development environment. That gap cost a month.

Chapter 14, Vignette 2The go/no-go decision

Twelve thousand barrels and a no-go

A large integrated marketer executing a year-end go-live on a new CTRM reached the go/no-go checkpoint with three of the four readiness dimensions confirmed. System testing was complete with all critical defects resolved. Process readiness was confirmed. People readiness was confirmed through user readiness testing. Data readiness had one amber condition: the opening inventory balances for two terminal locations did not reconcile with the legacy system's closing balances. The discrepancy was approximately 12,000 barrels across the two locations, within a portfolio of several million barrels. The implementation team assessed the discrepancy as attributable to a timing difference in the terminal inventory feed and recommended proceeding with a post-go-live correction. The cutover plan's rollback criteria specified that any unreconciled inventory balance at go-live was a no-go condition. The program manager invoked the criterion, recommended against proceeding, and escalated to the steering committee. The steering committee deferred the go-live by 10 days. In that window, the team identified the root cause: a cross-reference mapping error in the terminal inventory interface that had swapped two location identifiers in the opening balance calculation. Correcting the mapping and rerunning the opening balance load resolved the discrepancy completely. The corrected go-live executed cleanly. The 10-day delay was absorbed without material commercial consequence. Had the team proceeded with the original schedule and the unreconciled balances, the mapping error would have continued to corrupt inventory positions at both production locations, resulting in daily discrepancies that compounded until the root cause was identified. The rollback criterion had done exactly what it was designed to do: convert a schedule pressure into a governed decision with a traceable rationale.

Further readingLeading Practices for ETRM System Conversion, Cutover & Go-Live (Article, Energies Media, 2021; with Jason Wilton)

Chapter 15Change Management and User Adoption

Chapter 15, Vignette 1The super-user and change agent model

Super-users with no time to be super-users

A downstream marketing company implementing a CTRM for its pipeline scheduling and rack distribution operations identified its super-user candidates in the scheduling group by working with the scheduling manager to name the three or four people others consistently came to with questions. All four were fully committed to daily operational responsibilities and available for only a fraction of their time during the implementation. The program team accepted this constraint and trained them as planned. At go-live, all four super-users were present and engaged. Within three weeks, two of them had reverted almost entirely to their previous roles under operational pressure, and the remaining two were fielding questions they could not answer because their training had not covered the specific exception scenarios users were encountering in production. The scheduling group, lacking effective peer support, developed a set of informal workarounds that persisted for months after go-live and required a dedicated stabilization effort to remediate. The post-project review identified the super-user selection process as the primary contributing factor. The candidates had been chosen for their subject-matter expertise and operational standing, both valid criteria, but no adjustment had been made to their operational workloads to protect their availability for super-user responsibilities during the critical post-go-live period. The selection was right. The organizational support for the role was not.

Chapter 15, Vignette 2Operational health: establishing baselines and measuring real adoption

Phantom positions from late confirmations

A Gulf Coast refiner went live on a new CTRM covering crude, products, and derivatives in the fourth quarter of the fiscal year. The go-live itself was technically clean: interfaces ran, master data loaded correctly, and the cutover checklist was completed on schedule. The program team remained on-site for a four-week hypercare period and reported positive adoption signals: users were logging in, deal entry was occurring, and no critical system defects had been identified. In the sixth week after go-live, the risk control group escalated an issue: P&L variance reports showed significant unexplained swings in the final week of the month, making month-end close difficult and resulting in financial statements that did not reconcile cleanly with physical inventory. The investigation took three days. The root cause was a scheduling practice that had not been identified during training: the scheduling team had been confirming pipeline nominations in the system when the invoices arrived, rather than when the carrier confirmations were received, resulting in a delay of three to seven days per movement. The system's position calculations assumed same-day confirmation. The lag was producing phantom open positions that resolved at invoice receipt, creating the P&L swings the risk group was seeing. No one had defined a nomination confirmation timeliness metric before go-live. No operational health monitoring was in place during hypercare. The behavior had been invisible until it surfaced as a financial reporting problem. The remediation required reprocessing six weeks of nominations, retraining the scheduling team on the confirmation workflow, and a conversation between the scheduling manager and the risk control group that should have happened during the design phase. The financial impact was contained. The cost in time, credibility, and organizational frustration was not.

Chapter 16Renewable Fuels as a Trading and Compliance Domain

Chapter 16, Vignette 1The commercial dimension: compliance as margin, not just reporting

Compliance and trading optimizing separately

A mid-continent refiner with LCFS exposure in California and Washington managed its renewable fuels compliance through a dedicated compliance team that operated separately from the commercial trading desk. The compliance team tracked obligations and credits in a monthly spreadsheet. The trading desk managed the physical renewable fuel inventory in the CTRM without visibility into the credit position. In the fourth quarter of one compliance year, the trading desk sold a substantial volume of renewable diesel into a non-LCFS jurisdiction due to an attractive physical margin. The compliance team, unaware of the sale until the monthly reconciliation, discovered that the organization was short the LCFS credits needed to satisfy its California obligation. Purchasing the required credits at quarter-end, when every other participant in similar situations was also buying, cost roughly $4 million more than the credits would have cost three months earlier, when the trading decision was made. The physical margin on the trades that created the shortfall was approximately $1.8 million. The compliance and commercial functions had optimized their respective objectives independently and produced a combined result worse than either had intended.

Chapter 16, Vignette 2The multi-jurisdictional complexity problem: proliferation, stacking, and the architecture it requires

A compliance application outgrown by the rules

A large independent refiner with production assets across several western states built a custom compliance tracking application to manage its California LCFS and federal RFS obligations. The application was well-designed for the two programs it was built to support: it consumed daily transaction data from the CTRM, accurately calculated obligations and credit positions, and produced the required regulatory reports on schedule. When Oregon enacted its Clean Fuels Program and Washington enacted its Clean Fuels Standard in successive years, the compliance team requested enhancements to the application to add the new programs. Each enhancement required a four- to six-month development cycle, a testing period, and a deployment window. During the gap between program enactment and system readiness, the compliance team managed the new obligations in spreadsheets. The Washington program launched during a period of unusual market tightness in renewable diesel supply, creating simultaneous compliance pressure and spreadsheet-management risk. The organization's subsequent investment in a redesigned compliance platform with a configurable obligation logic engine was driven less by the cost of the two development cycles than by the recognition that the next three programs on the regulatory horizon would each require the same cycle, and the development cycle was longer than the lead time the new programs allowed.

Chapter 17Fuel and Excise Tax Compliance

Chapter 17, Vignette 1Why it is harder than it looks: rate changes, state variation, and the volume problem

Missed rate changes in eight states

A regional petroleum distributor operating terminals across eight states managed its motor fuel excise tax through a combination of manual rate tables maintained in spreadsheets and a CTRM configured to apply state tax rates as fixed-price components on invoices. The configuration worked adequately when the business operated in three states, where the tax team could monitor rate changes and update the system within a day or two of each change. As the business expanded through acquisition into five additional states, the manual process became unsustainable. In a 12-month period, the tax team missed rate-change effective dates in three jurisdictions: two because the notice periods were shorter than the team's internal update cycle, and one because the acquisition had brought in a state the team had not previously monitored. The resulting under-collection on affected transactions, discovered during a routine internal audit, required the company to absorb the tax difference rather than pass it on to customers whose invoices had already been paid. The absorbed amount was just under $400,000. The process improvement that followed replaced the manual rate table with an automated tax determination engine, which eliminated the rate monitoring burden entirely and reduced the tax team's role from calculation to exception review.

Chapter 17, Vignette 2License management, exemption certificates, and the audit exposure they represent

Expired certificates and an audit

A fuel marketer that had grown through a series of acquisitions over a decade operated with a license-and-exemption portfolio that reflected its acquisition history rather than its current business structure. Each acquired entity had maintained its own license files and certificate records, and integrating those records after each acquisition had been treated as a lower priority than the operational integration of the businesses. When a state tax authority initiated a routine audit of the marketer's combined operations, the audit team requested documentation for all exempt sales in three states over a two-year period. The tax compliance team spent three weeks locating and organizing the relevant certificates, ultimately producing documentation for approximately 70% of the transactions in question. The remaining 30% had been covered by certificates that had expired and been renewed by the acquired entity's original tax team but not transferred to the acquirer's records during the integration. The assessed tax on the undocumented transactions, plus interest and penalties, totaled just over $1.1 million. The subsequent investment in a centralized certificate management system with automated renewal workflows and transaction-level certificate linking was roughly $180,000, a number that put the true cost of informal certificate management in perspective.

Chapter 18Risk Management and Analytics

Chapter 18, Vignette 1The risk reporting hierarchy: assembling the stack

A precise VaR from imprecise inputs

A mid-sized products marketer implemented a CTRM with a full risk module and began producing daily VaR reports within three months of go-live. The reports were reviewed in weekly risk committee meetings and used to set position limits for the trading desk. Eighteen months later, during a period of significant market volatility, the desk experienced a loss roughly four times the reported one-day VaR. The post-mortem identified three contributing factors. Approximately 30% of physical inventory in transit was not captured in the position system due to a timing gap between nomination and trade entry. Pricing terms on a subset of term purchase contracts had been configured as month-average, whereas the actual contracts used a five-day pricing window around the bill of lading, resulting in a systematic understatement of flat price exposure. The price curves used for mark-to-market had not been updated to reflect an illiquid condition in a regional differential, causing the basis component of several positions to be carried at stale values. None of these were model failures. All were data and configuration failures that the model had no mechanism to detect or correct for. The VaR figure was precisely calculated from imprecise inputs.

Chapter 18, Vignette 2Priced inventory as a risk management discipline

Priced inventory treated as a suggestion

An asset-backed trading refiner with significant pipeline and terminal assets built a priced inventory management program as part of a CTRM implementation. The system was configured to track pricing status across all physical categories: inventory in tank, in transit, purchase trades priced but not ticketed, and sales trades priced but not ticketed. In the first months of operation, the risk function produced and reviewed the daily hedge recommendation, but implementation of the hedge was discretionary. The trading desk treated the recommendation as a starting point, frequently choosing not to execute the full hedge when their market view was directional. Within a year, the program had evolved into a culture in which the priced-inventory target was treated as a suggestion rather than a control. When a sharp market move produced a loss that exceeded the board-authorized VaR limit, the post-mortem revealed that the trading desk had been consistently running priced inventory above target for three months, citing a bullish market view that was not authorized under the firm's risk mandate. The priced inventory framework had been correctly implemented in the CTRM. The governance structure around it had not been.

Chapter 18, Vignette 3Where VaR breaks: stress testing, CVaR, and the limits of historical data

The stress scenario the model could not see

A refined products trading organization at Level 3 sophistication operated under VaR limits set by the board three years earlier, during a period of moderate market volatility. The risk control function produced daily VaR reports against those limits, which were rarely approached. When a period of unusual volatility in regional gasoline differentials began, the VaR figures remained within limits because the 60-day historical window had not yet fully captured the new volatility regime. The risk manager, who understood the limitation, supplemented the daily VaR report with an informal scenario analysis showing what the portfolio would look like if regional differentials moved to levels last seen during the 2012 supply disruption. The scenario showed a potential loss roughly three times the reported VaR. The analysis was shared with the trading desk and the trading risk committee. The desk voluntarily reduced its basis exposure before the scenario materialized. When the differential move occurred two weeks later, the organization reported a modest gain on the reduced position. The VaR model had not predicted the event. The stress scenario analysis, run by a risk manager who understood what the model could not tell her, had provided the warning that prompted a change in behavior.

Chapter 19Leveraging Emerging Technologies

Chapter 19, Vignette 1Generative AI in the trading and operations environment

Inland waterway logistics through the maturity phases

Consider how a purpose-built platform for inland waterway logistics would develop through the maturity phases. No refiner has yet deployed such a system at scale; as Chapter 8 noted, this remains the most fragmented data environment in the hydrocarbon supply chain, though purpose-built products are beginning to emerge from vendors targeting this space. The design case is nonetheless worth walking through, because it shows how the phases build on one another in a domain most readers know well. The foundation would be a structured data model that ties every movement to its route, vessel assignment, consist configuration, and external events that affect it: lock delays, weather alerts, and river gauge readings. Phase 1 delivers situational awareness from that model alone: schedulers see where the product is, what is in its path, and which movements carry elevated delivery risk. Phase 2 adds planning support, using accumulated movement history to produce calibrated transit-time estimates that replace the informal assumptions currently living in schedulers' spreadsheets. Phase 3 adds a natural-language interface through which schedulers ask operational questions in plain language and receive structured answers, with the model's reasoning visible alongside the output. The lesson the sequence teaches is the one this chapter keeps returning to: the value of the Phase 3 interface would be directly proportional to the quality of the Phase 1 data model. Queries touching well-structured, consistently attributed data would produce answers schedulers trust and act on. Queries touching thin or inconsistently populated domains would yield answers that require manual verification. Two years of structured movement records, patiently accumulated before the intelligent layer arrives, would outperform institutional knowledge at every layer of analysis the system eventually supports. The organization that wants the Phase 3 capability in 2028 is making the Phase 1 data model decision now.

Further readingSupply Chain Technology and Vessel Tracking: A Better Approach to Barge Logistics (Article, Supply & Demand Chain Executive, 2025)

Chapter 19, Vignette 2The organizational prerequisites: why most AI implementations fail before the model runs

Forecasting a recording lag

A refined products marketer with aspirations to expand into asset-backed trading invested in a machine learning demand forecasting platform after observing competitor organizations reporting material efficiency gains from similar implementations. The platform was deployed against the organization's CTRM data. The initial results were disappointing: the model's forecasts were less accurate than the planning team's existing methods, and the planning team stopped using the system's outputs within three months of go-live. A post-implementation review identified the cause. The CTRM's inventory data had a systematic timing gap: movements were entered after delivery confirmation rather than at the time of scheduling, so the historical record the model was trained on reflected when transactions were recorded, not when they occurred. The model had learned the recording lag as a feature of the data rather than an artifact of entry practice. Correcting the model required correcting the data, which in turn required a change in operational discipline around trade entry, taking eight months to establish and another six months to accumulate enough clean historical data to retrain the model. The platform eventually produced the forecast accuracy the organization had targeted. The timeline from investment to value was 22 months rather than the six months projected at the outset, and the delay was entirely attributable to data quality issues that were known but underestimated before implementation began.

Chapter 20The Evolving CTRM Operating Model

Chapter 20, Vignette 1What the organization inherits at go-live

Three people, then one

A Gulf Coast refiner completed a well-run 18-month CTRM implementation and went live on schedule. The implementation partner transitioned support to an internal IT team of three in the week following go-live. Within 60 days, two of the three had been pulled to other projects. The one remaining team member was fielding 40 to 60 tickets per week, had no documentation beyond the implementation team’s project files, and had no authority to make configuration changes without a monthly change advisory board review. Six months after go-live, the commercial team had rebuilt the freight accrual spreadsheet that the implementation was supposed to replace. The configuration was sound. The operating model was not.

Further readingKey Considerations For Supporting Your ETRM System (Article, SHALE Magazine, 2021)

Chapter 20, Vignette 2The internal-versus-managed-services decision

Managed services without business context

A midstream marketing organization outsourced full CTRM application support to a managed services provider two years after go-live, retaining one internal system owner to manage the relationship. The first year was difficult: ticket response times were adequate, but the provider’s analysts did not understand the business well enough to distinguish configuration questions from process questions, and the internal owner spent more time translating than governing. In year two, the organization restructured the engagement to require that the provider’s lead analyst attend weekly commercial operations calls. The change in context led to a measurable reduction in escalations and to a managed services team that began identifying issues before users reported them. The model worked. The original scope definition had not.