Fleet Route Optimization Software That Shows Its Working
An optimiser will always return an answer. The question is what it was told to minimise and what it was allowed to ignore. KO Fleetz makes the objective an explicit choice, builds service times from what your stops actually take, and reports an infeasible plan as infeasible rather than dressing one up.
The solver said eight vehicles. The yard needed eleven.
The pilot goes beautifully. A model chews through the week's drops and comes back with fewer runs and less distance than the planner managed, and everyone in the room can see the number. Six weeks later the drivers have gone back to the old sheets and the licence is up for renewal with nothing to show.
The reason is almost never the mathematics. It is that the model minimised distance while the real constraint was drivers' hours. It gave every stop ten minutes of service time because that was the default, and the retail park drops take forty. It did not know the depot can only load three vehicles at once, so it built a plan where nine trucks leave at six. Each of those is a fact the yard has known for years.
So the optimiser gets the blame for a plan the operation could not run, and the fleet concludes that this kind of software does not work here. The plan was not wrong. It was a correct answer to a description of the operation that left out the parts that bind.
State the objective. Show the constraints. Compare the runs.
Distance, drive time, vehicle count, total cost and on-time performance pull in different directions, and no plan wins on all of them. The shortest set of routes is frequently not the cheapest once shift length and detention are priced. KO Fleetz makes you name the objective before it runs, and shows the same set of drops solved against different objectives so the trade-off is a decision somebody made rather than a default they inherited.
KO Fleetz measures service times rather than assuming them. Arrival and departure records from previous visits give each site its own realistic dwell, which is usually the single change that moves an optimiser's output from theoretical to runnable. Depot loading capacity, backhaul opportunities and the return leg are constraints inside the model rather than things the yard absorbs afterwards.
When the constraints cannot all hold, the run does not quietly relax one to produce a tidy result. It returns infeasible and names what bound it: these nine drops cannot be served inside their windows with six vehicles from this depot. That is a more useful sentence than any plan built by silently widening a window nobody agreed to widen. Nor does it re-plan a running day on its own. A disruption produces a proposed revision, and a dispatcher accepts it.
Capabilities
What KO Fleetz route optimization gives your team
Explicit objective selection
Pick distance, drive time, vehicle count, cost or service level before the run. The objective is recorded with the result rather than assumed.
Historical service times
KO Fleetz builds each site's dwell from its own arrival and departure record, so the retail park that always takes forty minutes is modelled as forty.
Infeasibility reporting
A plan that cannot hold every constraint returns as infeasible with the binding one named, instead of a clean result built by quietly dropping a rule.
Scenario comparison
Two runs of the same drops sit side by side with their objectives and their costs, which is how a fleet-size decision gets made on evidence.
Fleet sizing questions
Ask how few vehicles could serve this week's orders and get an answer with the constraint that stops it going lower, rather than a bare number.
Depot throughput constraints
Loading bay count and dispatch windows are modelled, so a plan cannot assume nine vehicles are loaded simultaneously at a yard with three bays.
Backhaul and return legs
Return journeys are part of the model, and collections that fit the way home are offered against them rather than planned as separate work.
Disruption re-runs on approval
A breakdown or a cancelled drop produces a proposed revision to the remaining stops. It reaches drivers only once a dispatcher accepts it.
How it works
How KO Fleetz does it
Step 1: Start from the planned constraints
The optimiser reads the same windows, vehicle profiles, capacities and access rules that route planning already holds. It invents nothing of its own.
Step 2: Choose what you are minimising
Name the objective for this run. Distance and vehicle count give different plans, and the choice belongs to the operation rather than the solver.
Step 3: Run and compare
Scenarios are produced side by side with their vehicle counts, distances and estimated costs. Infeasible runs report the constraint that bound them.
Step 4: Accept and publish
The chosen scenario becomes a plan version and goes to the dispatch board. Nothing reaches a driver until somebody has accepted it.
Step 5: Feed the actuals back
Real dwell times, delays and detentions from the executed day update the service-time model, so the next run is built on a truer picture.
Outcomes
What changes
- The objective is stated, never assumed by the solver
- You choose
- Infeasible runs report the constraint that bound them
- Named
- Dwell times taken from history rather than a default
- Measured
- Scenarios compared before anything reaches a driver
- Side by side
Frequently asked questions
Planning produces one sequence that works. Optimisation asks whether a materially better one exists and searches for it, which is a different kind of question and only worth asking when there is slack to find. If your planner sequences twelve stops a day on lanes he has run for a decade, the search will mostly agree with him. If you are sizing a fleet against next quarter's orders, or arranging four hundred drops across a region, no human enumerates those alternatives.
We will not quote you a figure, and be careful with anyone who does before seeing your data. The gain depends entirely on how much slack sits in the current plan, and slack varies enormously. A fleet whose runs were drawn on a map years ago and never revisited usually has a lot. A fleet already sequencing tightly against hard windows may have almost none. Run it against a historic week of your own orders. That comparison is real and it costs you nothing but the data.
Take the drivers seriously, because they are usually describing a constraint the model does not have. The common ones: a site that takes far longer than its recorded dwell, an approach that only works in one direction, a customer who will not accept before their delivery team arrives regardless of the stated window. Each of those is fixable as data. An optimiser that gets argued with and corrected improves. One that gets overruled and ignored is a licence you are paying for.
It can propose. A vehicle failure, a refused delivery or an urgent collection triggers a fresh run against the stops that are still outstanding, and the result is offered to the dispatcher as a revision with the change highlighted. KO Fleetz does not push a new sequence to a driver mid-route on its own. A driver who has had their afternoon silently rearranged twice stops following the sequence at all, and then you have neither the optimisation nor the plan.
It depends on the combinatorics, not on the vehicle count. Six vans doing two hundred parcel drops across a city is a genuinely hard problem where a solver beats a person. Six trucks doing three trunk legs each on fixed lanes is not, and the honest advice there is to spend the money on planning and proof of delivery instead. The rough test: if your planner can hold the whole day's shape in his head and defend it, you do not need a search.
Typical daily planning volumes for a regional distribution fleet solve inside a workable overnight or early-morning cycle, and a mid-day re-run against remaining stops is faster because the problem is smaller. Realistic expectations matter: these are hard problems, and beyond a certain size the result is a very good answer rather than a provably optimal one. Tell us your stop counts and vehicle numbers and we will run your data rather than guess.
Run the optimiser against a week you already know
Give us a historic week of orders and the fleet you used, and KO Fleetz will lay out the alternative plans and where they break.