Logistics Automation: You Cannot Save More Time Than the Step You Automated

Logistics automation removes manual effort from a step, and logistics orchestration coordinates steps so one decision accounts for several. Both work, and the most common complaint after a successful go-live is that the end-to-end numbers did not move. The explanation is usually sought in adoption, data quality or configuration. It is simpler than that. The amount of cycle time you can remove is bounded by the share of cycle time the automated step occupied, and no amount of speed changes that bound. Automate a step worth 30% of the cycle and you cannot reduce the cycle by more than 30%, even if the step becomes instantaneous.

That ceiling is knowable before the project starts, from a measurement most operations have never taken. Locus, the world’s first Decision-Intelligent, Agentic TMS, produces dispatch-ready plans in roughly two minutes against more than 250 real-world operating constraints, and the operations that convert that speed into cycle time are the ones that checked first how much of their cycle planning actually was.

Why the ceiling exists and where it comes from

The principle is borrowed and exact. Gene Amdahl set it out at the 1967 AFIPS conference in the context of parallel computing, and Amdahl’s law states that overall speedup is limited by the portion of the work you did not improve, expressed as one divided by the quantity one minus P, plus P over N. Substitute a logistics cycle for a program and the arithmetic is unchanged: the part you left alone sets the floor.

Applied to a cycle, the time remaining is one minus the automated share, plus that share divided by the speedup. The reduction available looks like this.

Read across the 30% row and the message is that speed saturates quickly. Going from five times faster to ten times faster buys three percentage points. Now read down the infinite-speed column: moving the target from a 30% step to a 50% step buys twenty. What you automate dominates how fast it goes, and most selection processes evaluate vendors on the second.

The second half of the problem is that the constraint does not disappear when you remove it, it relocates. This is Goldratt’s point in the theory of constraints, set out in The Goal in 1984, whose five focusing steps end with an instruction to repeat the cycle precisely because elevating one constraint promotes another. Take a cycle composed of planning, dock and load, transit and exception handling, and automate planning tenfold.

The cycle fell 27%, which is a real result. Planning went from the largest stage to the smallest, and transit and dock are now 75% of what remains. Any further investment in planning speed is chasing 4% of the cycle.

That produces three numbers, all accurate, which is why post-go-live reviews go badly. The automated step is ten times faster. End-to-end cycle time is 27% shorter. Throughput may be unchanged, because throughput is governed by whatever the constraint is now, and the constraint is no longer planning. A vendor reports the first, the program reports the second, and the operations director reports the third.

The financial stakes sit in the part you did not automate. ATRI’s 2026 report puts the industry-average cost of operating a truck at $2.336 per mile in 2025, a record for the series and 3.4% above the prior year, and transit is where most of that is spent. Where a leg runs on a third party the constraint may not even be yours to elevate: the American Trucking Associations reports almost 580,000 active US motor carriers as of June 2025, of which 91.5% operate 10 or fewer trucks.

How to pick an automation target that moves the cycle

1. Measure the cycle before choosing anything

Time the end-to-end journey and decompose it by stage, in the same units, for a representative period. Most operations can name their stages and very few can state what share of elapsed time each consumes. That single measurement determines the ceiling on every automation option available, and taking it costs a week.

Measure elapsed time rather than touch time, because the two diverge most in exactly the stages you are considering automating. A planning stage may involve forty minutes of planner effort spread across four hours of waiting for inventory confirmation, and it is the four hours that sits in the cycle. Take the sample across a full operating cycle including at least one disrupted period, since the stage shares on a clean day are not the shares that govern your throughput.

2. Compute the ceiling for each candidate before evaluating vendors

For each option, the best possible outcome is the share of cycle time that stage occupies. Put that number next to the option before anyone sees a demonstration. An option capped at 15% should not be compared against one capped at 50% on the basis of feature depth.

Worked through, the comparison usually reorders the shortlist. Suppose planning is 30% of the cycle, dock and load 25%, transit 30% and exception handling 15%. A best-in-class planning tool running ten times faster caps at 27%. A mediocre dock scheduling tool running only twice as fast caps at 12.5%. A carrier mix change cutting transit by a fifth delivers 6% with no software at all. The planning tool wins here, but had planning been 15% of the cycle rather than 30%, its ceiling would have dropped to 13.5% and the ranking would have inverted. The ceiling is a property of your operation, not of the product, which is why it has to be computed per buyer.

3. Automate the largest stage, not the most tractable one

These are rarely the same, which is the trap. The most tractable stage is the one with clean data and a willing owner, which is usually why it was already the most efficient. Tractability correlates with small share, so following the path of least resistance systematically selects low-ceiling targets.

4. Predict where the constraint will move, before it moves

Recompute the stage composition assuming your automation succeeds. The stage that becomes the largest share is your next constraint, and you now know it months ahead of experiencing it. This is the cheapest planning exercise available and it is almost never done, because roadmaps are organized by capability rather than by constraint.

Do it for two rounds rather than one. After planning is automated, transit leads at 41%; if you then address transit through carrier mix, dock becomes the constraint. Knowing the second and third constraints in advance tells you which teams to involve early and, more usefully, when to stop. A roadmap that reaches a stage you cannot influence, such as transit on a third-party leg, should say so rather than continuing to schedule work.

5. Measure three numbers and report all three

Step performance, end-to-end cycle time and throughput. Reporting only the first makes the project look transformative. Reporting only the third makes it look like a failure. Publishing all three, with the ceiling calculation alongside, is what keeps a successful project from being judged against an arithmetic that was never available to it.

6. Re-measure composition after every go-live

The stage shares change every time you automate something, so a roadmap built on last year’s composition is optimizing a cycle that no longer exists. Re-measuring after each phase is what turns a capability roadmap into a constraint roadmap, which is the difference between spending and improving.

What the three numbers each tell you

The last row is the one to add to your reporting pack. It is not an outcome measure, and it is the only one that tells you what to do next. It also reframes an awkward conversation: a stage that has dropped from 30% of the cycle to 4% is evidence the investment worked, even in a quarter when throughput did not move, because the composition shifted exactly as predicted and the next target is now visible rather than theoretical.

Five questions before committing to an automation target

What share of cycle time does this stage occupy? If nobody can answer, the ceiling is unknown and so is the business case.

What is the theoretical maximum cycle reduction? It equals that share. State it in the paper, because it is the number the project will eventually be judged against.

Which stage becomes the constraint afterwards? Recompute the composition assuming success. Whoever owns that stage should be in the room now rather than surprised later.

Are we choosing this stage because it is large or because it is easy? Both are legitimate answers. Only one of them is consistent with a cycle time target.

Does the next constraint sit inside our company? If the answer is transit on a third-party carrier, elevating it is a commercial negotiation rather than a systems project, and the timeline changes accordingly.

What this looks like in enterprise deployments

A leading North American retailer running multimodal logistics automation across several hundred stores replaced six legacy systems, reduced manual dispatch by more than 80%, achieved 99%-plus on-time store delivery with 95%-plus route compliance, and resolves exceptions in under two hours, breaking even in year one. The exception number is the one that matters for this argument. Exception handling is a stage most programs leave alone while automating planning, and addressing both is what prevents the constraint simply relocating from one to the other.

A Fortune 50 parcel operation running centralized dispatch across a 120-country network and 51 sites lifted weekly execution adherence from 75% to 92% and surfaced more than $14 million of unused capacity, including $565,000 at a single site. Adherence is a whole-cycle measure rather than a stage measure, which is why it is the right number to quote. A stage can improve tenfold without adherence moving at all if the constraint sits elsewhere in the journey.

Four mistakes enterprises make choosing automation targets

Evaluating speed rather than share. Vendor comparisons run on how fast a step becomes. The ceiling is set by how much of the cycle that step was, and no vendor is asked that question because it is a question about the buyer.

Automating the tractable stage. Clean data and a willing owner usually indicate a stage that was already efficient, and therefore small. The path of least resistance is a reliable route to a low ceiling.

Continuing to invest where the capability lives. After a successful planning automation, the roadmap fills with planning features, because that is where the team, the vendor relationship and the expertise now sit. The constraint has moved and the spending has not.

Reporting one number. Step performance alone oversells, throughput alone undersells, and the gap between them is what destroys credibility at the benefits review. All three, plus the composition, or none. The version that survives scrutiny states the ceiling up front, so a 27% cycle reduction reads as 90% of what was available rather than as a disappointing fraction of 100%.

How Locus fits a constraint-led roadmap

Locus, the world’s first Decision-Intelligent, Agentic TMS, produces dispatch-ready plans in roughly two minutes against more than 250 real-world operating constraints, and re-optimizes continuously through the route planning system. That speed converts into cycle time in proportion to how much of your cycle planning occupies, which is a question worth answering before procurement rather than after.

Where the platform reaches beyond a single stage is the point of it. Allocation runs across owned fleet, contracted transporters and more than 1,000 carriers, which touches transit rather than only planning. Exception handling runs through Autonomy Levels and Human Review, which addresses the stage most often left untouched. Explainability and Traceability record the trigger, context, reasoning, action and outcome of each decision, which is what makes stage composition measurable after a change rather than estimated. Together those cover more of the cycle than a planning tool does, and covering more of the cycle is the only way to raise the ceiling.

Two boundaries belong here. Locus does not shorten transit, which is a function of distance, traffic and the carrier performing the leg, and where transit becomes your binding constraint the answer is network design or a commercial conversation rather than a planning engine. And Locus does not automate dock and yard processes running in other systems, so if loading is your largest stage, the highest-ceiling investment is not a TMS at all. Saying that plainly is more useful than implying the cycle is ours to compress end to end.

Locus supports more than 360 enterprise customers across 30-plus countries, with over 1.5 billion deliveries optimized, more than $320 million in documented client logistics savings and 99.99% uptime. It has been recognized by Gartner for seven consecutive years, featured in the 2026 Hype Cycle for Supply Chain Execution and Logistics Technologies, named a Leader in TMS by QKS Group (SPARK Matrix), and ranked #1 in Route Planning on G2’s 2026 Best Software Awards.

In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

So why did the cycle time barely move after a successful automation? Because the reduction available was capped by the share of the cycle that step occupied, and that ceiling was set before anyone chose a vendor. Automating a stage worth 30% of cycle time yields at most 30%, and at a realistic tenfold speedup, 27%. Speed saturates fast, with 5x to 10x worth three points, while moving the target from a 30% stage to a 50% stage is worth twenty. The constraint then relocates rather than disappearing, so planning falls from the largest stage to the smallest while transit and dock become 75% of what remains, which is why step performance can be tenfold better while throughput is flat. Locus covers more of the cycle than a planning tool does, with two-minute re-planning against more than 250 constraints, allocation spanning owned, contracted and 1,000-plus carrier capacity, and exception handling governed by Autonomy Levels, and it does not shorten transit or automate a dock process running in another system. Request a Locus assessment to measure your stage composition before you pick a target.

Frequently Asked Questions

Why did our cycle time barely improve after automating dispatch planning? Because cycle reduction is capped by the share of the cycle that planning occupied. If planning was 30% of elapsed time, 30% is the theoretical maximum and about 27% is realistic at a tenfold speedup. The other 70% was never in scope.

Does making the automated step faster help? Very little, past a point. On a stage worth 30% of the cycle, moving from five times faster to ten times faster improves total cycle reduction from 24% to 27%. Choosing a stage worth 50% instead would have been worth twenty points.

What is the constraint migration problem? Removing one bottleneck promotes another. Automating planning tenfold takes it from 30% of the cycle to 4%, leaving transit at 41% and dock at 34% of what remains. Throughput is then governed by transit, so it does not move until transit is addressed.

Why do step performance and throughput disagree? Because they measure different things and both are correct. The step really is ten times faster, the cycle really is 27% shorter, and throughput really is flat while the new constraint is untouched. Reporting only one of the three is what makes benefits reviews contentious.

How do we choose the right automation target? Measure stage composition first, compute the ceiling for each candidate from its share of cycle time, and choose the largest stage rather than the most tractable one. Tractability usually indicates a stage that was already efficient, and therefore small.

What should we do after a successful automation? Re-measure the composition. The shares have changed, so the next constraint is different, and a roadmap built on the old composition is optimizing a cycle that no longer exists. This is the fifth of Goldratt’s focusing steps applied to a logistics cycle.

What if the new constraint is transit on a third-party carrier? Then elevating it is a commercial matter rather than a systems project. Network design, carrier mix or a contract change are the levers, and the timeline is a negotiation rather than a deployment.