A tanker pulls away from a retail site with several hundred gallons still aboard because the underground tank had less ullage than the dispatcher assumed. The driver’s options both cost money: find another site that can take the remainder, or return it to the terminal and account for the retain. Neither shows up as a line item on the P&L.
Fuel hauling margin lives in that gap between the planned drop and what the tank could actually take. Vendors in this space, Wezom among them, tend to find recoverable cost in dispatch and inventory decisions rather than in fuel economy or rate negotiation, which is why serious evaluations of oil and gas transportation management software begin with how a delivery was planned rather than how the truck was driven.
The distinction matters at board level because the two paths lead to different investments. One leads to telematics, driver coaching and equipment specification. The other leads to inventory forecasting, load construction and reconciliation, which is where the larger and less visible numbers usually sit.
The Cost Structure Executives Are Usually Shown, and the One That Matters
Fuel haul operations are commonly reported on cost per mile or cost per gallon delivered. Both are useful for benchmarking and both obscure where controllable cost sits.
The controllable unit is the truck shift. A tractor-tanker with a qualified hazmat driver is a fixed cost for the shift regardless of how many drops it makes. What determines profitability is how many productive loads that shift produces and how full each was.
Once the shift is the unit, loss categories become visible. Terminal wait consumes shift capacity without producing a drop. Partial loads consume a full round trip to deliver less than a full compartment set. Short drops produce a retain that must be handled later. Dead miles between a delivery and the next lift consume the same hours as loaded miles. Detention at receiving sites consumes the shift and is frequently not billed because nobody recorded arrival time precisely enough to support the charge.
None of these appear in cost per mile in a way that suggests a remedy. A fleet with high cost per mile might be paying too much for fuel, or running trucks half empty because dispatch cannot see tank levels. The two have nothing in common, and only one is a software problem.
The diagnostic question is simple to ask and usually hard to answer: for the last thirty days, what percentage of available shift hours were spent loaded and moving, waiting at terminals, waiting at delivery sites, and running empty. Operations that cannot produce that breakdown are managing the cost they can see rather than the cost they have.
Replacing the Phone Call With a Forecast
The largest single change available to most distributors is moving from customer-initiated orders to inventory-driven replenishment.
In a call-in model, a site manager notices the tank is low and calls. The dispatcher fits the order into a route, and the truck delivers whatever the tank will take when it arrives. The distributor has no visibility between calls, so every delivery is planned against an estimate and every route is built from whatever orders happened to arrive.
In a managed inventory model, the distributor knows the tank level and forecasts when it reaches the reorder threshold, using either tank monitoring hardware or consumption modelling based on historical draw rates, day of week, seasonality and known events. Most operations use both, because monitoring hardware is not economic at every site and consumption modelling degrades where demand is irregular.
Delivery timing then becomes a planning variable rather than a constraint. A site needing fuel Thursday can be served Wednesday if a truck is passing with capacity, decided against real cost difference. Route density improves because deliveries group geographically rather than chronologically, and dedicated trips to single sites fall.
Forecast quality determines whether this works. A model that under-predicts consumption produces run-outs, which in retail fuel means lost sales for the customer and, under service level terms, a financial consequence for the distributor. Over-prediction produces early deliveries with low drop volumes. Both failure modes are visible in delivered volume per drop and in the distribution of tank levels at delivery, and any system implementing this should report both continuously rather than only at commissioning.
Tank monitoring data has quality problems needing handling in software rather than assumption. Gauges drift. Water ingress affects readings. Communications gaps produce stale values that look current. Deliveries in progress produce readings that are temporarily meaningless. A forecasting layer consuming gauge data without validating it produces confident predictions from bad inputs, which is worse than no prediction because dispatch will act on it.
Load Construction Is a Compartment Problem
Fuel trucks do not carry one product. A tanker has multiple compartments of fixed size, and a load is an assignment of products to compartments against a set of deliveries, constrained by weight limits, axle distribution and product compatibility.
This is where generic routing software fails. Route optimization treats a vehicle as having a capacity. A tanker has a compartment configuration, and the difference is not cosmetic: it can carry only the combinations its compartment sizes permit, and each compartment is delivered whole or leaves a retain.
Load planning involves several simultaneous decisions. Which sites are served. How much each needs, which depends on ullage forecast at expected arrival rather than at planning time. Which products go in which compartments, considering that compartments may need dedication to avoid contamination between grades or between gasoline and diesel. What the loaded weight will be, given that fuel density varies with product and temperature. And in what sequence the drops occur, which affects weight distribution through the route and which retains are possible.
Systems handling this well produce a load plan a driver executes without judgement calls at the site. Systems that do not produce a route list and leave compartment assignment to the rack operator, which works when volumes are simple and breaks when a load serves several sites with mixed product needs.
The financial effect is direct. Every gallon of unused compartment capacity on a departing truck is a round trip’s fixed cost spread over fewer gallons. Improving average load fill raises revenue per shift without adding assets or headcount.
Product compatibility and residue rules belong in the load planning logic rather than in memory. Loading gasoline into a compartment that last held diesel, or moving between grades without appropriate handling, creates a quality problem that reaches the customer’s tank.
Terminal Operations and the Cost of Waiting
Rack loading involves queuing for a bay, authenticating, entering load details, loading through a metered arm, and receiving a bill of lading recording gross and net volumes by product. Wait time depends on terminal congestion, which follows predictable daily patterns and unpredictable disruptions. A fleet dispatching without regard to terminal timing spends shift hours in a queue.
Supply allocation adds a layer. Distributors hold contracted volumes with suppliers at specific terminals, sometimes with monthly allocation limits and lifting patterns affecting pricing. Rack prices change on a schedule, and lift timing relative to a price change has a direct margin effect. A dispatcher planning purely for route efficiency, without visibility into price timing and allocation position, can make an operationally sound decision that costs money on the purchase side.
Transportation management systems built for fuel make terminal selection a planning decision rather than a default. The system should know which terminals carry which products, the current rack position, allocation status against contract, the recent wait profile by time of day, and how terminal choice affects the route. Those inputs together produce a different answer from distance alone, often enough to matter.
The bill of lading is also where the data trail begins. Inventory accounting, invoicing and volume reconciliation all depend on BOL data being captured accurately and promptly. Operations capturing BOLs as paper handed in at end of shift are running blind for the shift and reconciling days later, when discrepancies are hard to investigate.
Gross, Net and the Reconciliation Nobody Enjoys
Fuel volume is temperature dependent, which creates an accounting problem specific to this business.
Volume is measured both gross, at actual temperature, and net, corrected to a reference temperature, which in United States practice is 60 degrees Fahrenheit. A load lifted on a hot afternoon and delivered on a cold morning has a different gross volume at each end while containing the same product. Systems and contracts vary as to whether transactions use gross or net, and mixing the two anywhere produces discrepancies that look like loss.
Beyond temperature, real gains and losses occur: meter calibration drift at the rack and delivery point, retains left in compartments and lines, evaporation, and occasionally diversion. Distinguishing these requires consistent measurement and prompt reconciliation, because a discrepancy investigated three weeks later has no recoverable context.
This is the least visible area of oil and gas transportation management software and one of the more valuable. A system reconciling each load from rack BOL through delivery tickets to tank gauge readings, applying temperature correction consistently, produces a per-load gain or loss figure. Aggregated by driver, truck, terminal and site, that figure identifies calibration problems, procedural failures and anomalies worth attention. Operations without it discover the aggregate loss at month end as a number on an inventory account, with no path back to its cause.
The implementation detail that matters is where correction is applied and which figure is authoritative at each step. A system storing a single volume field, without recording whether it is gross or net and at what temperature it was measured, cannot reconcile reliably regardless of how good the rest of the design is.
Compliance Overhead and Where Software Reduces It
Hazardous materials transport carries a documentation and qualification burden consuming administrative time whether or not it is systematized.
Drivers require commercial licences with hazmat and tanker endorsements, periodic training, medical certification and background checks, each with its own expiry. Cargo tanks require periodic inspection and testing with documentation retained. Shipping papers must accompany each load with prescribed information, and placarding must match the load. Hours of service rules apply, with the additional consideration that some fuel hauling operations qualify for particular exceptions depending on operating radius and other conditions.
None of this is optional and none is a differentiator. What software changes is the cost of compliance and the risk of a lapse. A qualification register tracking every driver and vehicle credential with expiry dates, preventing dispatch of a lapsed driver or unit, removes exposure otherwise managed by a spreadsheet and someone’s attention. Shipping papers generated from actual load composition remove transcription error. Electronic capture of delivery tickets with time and position produces the record supporting a detention charge and the evidence trail if a delivery is disputed.
Roadside inspection outcomes and safety scores affect insurance cost and, for some shippers, eligibility to carry product. The connection between operational software and those outcomes runs through whether violations are prevented at dispatch rather than discovered at the roadside.
Measuring the Change Without Fooling Yourself
Executives approving investment here should insist on a measurement design before implementation, because retrospective attribution in fuel distribution is unusually difficult. Volumes move with fuel price, weather and economic activity independently of any system change, so a quarter showing better cost per gallon may reflect a cold snap rather than operational improvement.
The metrics isolating operational effect are activity-based. Average delivered volume per drop, reflecting forecast quality and load planning. Drops per shift, reflecting routing and wait time. Load fill percentage at departure. Shift hours by category: loaded moving, empty moving, terminal wait, site wait, other. Run-out incidents per thousand deliveries. Retain volume as a percentage of lifted volume. Reconciled variance per load.
Each needs a baseline established before any change, measured the same way from the same source. Operations that cannot establish one because the data does not exist have learned something useful: the first phase of the project is measurement, and its value is independent of what comes after.
Financial effect can then be derived from activity change using the operation’s own cost structure rather than asserted. An improvement in drops per shift translates into either more volume on the same assets or the same volume on fewer, and which is realized is a management decision rather than a system output.
Sequencing the Investment
Data capture comes first: electronic BOL capture, delivery confirmation with volume, time and position, and a consolidated record of what each shift did. This changes no decisions, disrupts nothing, and produces the baseline everything else is measured against. It also surfaces the data quality problems that would otherwise derail a later optimization project.
Inventory visibility follows: tank monitoring where economic, consumption modelling elsewhere, and a forecast of when each site reaches its reorder point. Dispatchers can use this as an input to their existing process before any automated planning exists, and forecast accuracy can be validated against actual deliveries during that period.
Planning support comes third, once the forecast is trusted: load construction against compartment configuration, route building with terminal selection, and dispatch decisions considering delivery windows, driver hours and equipment availability together.
Reconciliation and analytics sit alongside rather than after, because they depend on the same captured data and do not require the planning layer to exist.
The temptation at executive level is to procure the whole capability at once. That sequence puts the most operationally sensitive component, dispatch planning, into production before the data underneath it has been validated, and dispatchers who lose confidence in a planning system in its first week will work around it permanently.
For a distributor evaluating oil and gas transportation management software against this sequence, the useful question is which phase a vendor would start with and why. A vendor proposing to begin with dispatch optimization, before establishing whether tank data is reliable and shift time is measurable, is proposing to optimize against inputs nobody has checked. A vendor proposing to begin with capture and measurement is proposing a slower start and a considerably higher chance that the optimization, when it arrives, optimizes the right thing.
