Route Familiarity is an Asset You Are Not Pricing: What Every Reassignment Actually Costs

The productivity gap nobody measures

Put two equally capable drivers on the same forty-stop route. One has run it for six months. The other is competent, well-trained, and new to it. The first will finish earlier, with fewer failed attempts and fewer calls to dispatch.

Nothing in that outcome is about skill or effort. It is about accumulated knowledge of a specific geography: which entrance actually works, where you can legally stop, which businesses close between one and two, which gate code is on the keypad rather than the buzzer, and which three addresses need a call ahead or they fail.

Most operations know this intuitively and almost none of them measure it. There is no line in any cost model for route familiarity, no field in the planning system that holds it, and no figure attached to losing it. Which means when a reassignment, a territory redesign, or a resignation destroys some of it, the loss is invisible and the decision that caused it looks free.

This piece argues familiarity is real capital worth pricing, explains why the optimiser structurally cannot see it, and lands on the resolution that matters: familiarity is only fragile because it lives in a person.

What route familiarity actually consists of

Naming the components matters, because most of them turn out to be things a system could hold.

Read the right-hand column. Almost all of it is capturable. Which means the reason familiarity is a depreciating personal asset rather than a durable institutional one is a design choice, not a fact of the work.

The two rows with the strongest existing evidence are failure and dwell. Every operation already records failed attempts and actual arrival and departure times. That data contains most of what the experienced driver learned the hard way, and in most operations it sits in reporting rather than in the planning model.

Why it depreciates, and why the loss is asymmetric

Familiarity behaves like capital with three unusual properties.

It is built only by repetition. No amount of briefing substitutes for having served the address. This is why onboarding programmes, which are genuinely useful, cannot close the gap: they transfer process knowledge, not place knowledge.

It decays with absence. A driver returning to a territory after two months has lost some of it, and the rate depends on how much changed while they were away. That gives familiarity a half-life, which is a more useful way to think about it than a binary.

The operation loses it instantly. A driver’s own decline is gradual. The employer’s is not. When a driver leaves, the entire stock of their place knowledge exits with them on their last day, and the replacement starts from zero on the same geography.

That asymmetry is what makes turnover more expensive than replacement cost suggests. And the turnover context is severe. The National Academies reports average annualised turnover of 92.7% at large truckload carriers against 11.8% for LTL linehaul drivers. At the first figure, a familiarity stock cannot accumulate: the roster is replaced faster than place knowledge can compound. At the second, it accumulates for years.

That comparison is worth sitting with, because the same operations debate whether familiarity matters while running turnover rates that make it impossible to find out.

The optimiser cannot see it

Here is the structural problem, and it applies to any planning system including a good one.

A route optimiser minimises a cost function containing distance, time, vehicle constraints, and service windows. Familiarity is in none of those terms. So when the optimiser reshuffles territory boundaries and finds a plan that is 3% shorter, it reports a 3% improvement and is telling the truth about the model. What the model does not contain is that eleven drivers are now working geographies they have never seen, and their real completion times for the next several weeks will not match the plan that was just optimised.

The result is a specific and common failure: an optimisation that tests well and underperforms in execution, followed by a conclusion that the drivers or the data were the problem. Neither was. The plan was optimal against a cost function missing a real cost.

This also explains a pattern operations teams report and find confusing, where performance dips after a re-optimisation and then recovers over a month or two. That is not resistance to change. That is familiarity being rebuilt.

Also Read: From Static Route Plans to Continuous Re-Optimisation: A European Last-Mile Efficiency Benchmark

The honest tension with continuous re-optimisation

This argument sits awkwardly with a position Locus and this blog have argued repeatedly, which is that static plans decay and operations should re-decide continuously. Both things are true, and the reconciliation is a distinction between two time horizons.

Intra-day re-optimisation should be aggressive. Re-sequencing today’s stops, reassigning a delayed drop, intercepting a failed attempt, reallocating capacity when a vehicle breaks down. None of that touches familiarity, because the driver is working the same geography they already know. Here the case for continuous re-decisioning is unqualified.

Cross-week territory assignment should be conservative. Which driver owns which geography, over weeks and months, is where familiarity is built or destroyed. Churning that for marginal distance gains spends an asset the optimiser is not counting.

Most planning systems make no distinction between these, which is why “we optimise continuously” can mean both something excellent and something quietly expensive. The useful policy is to separate the two: free re-optimisation inside the day, deliberate and measured change to territory ownership across weeks.

It collides with the retention argument, and the collision is productive

There is a well-made case, including in Locus’s own European driver management writing, that static territory allocation carries a retention cost. Monotony is real. So is the unfairness of locking a driver into a difficult territory indefinitely while a colleague holds an easy one.

So the answer is not to freeze territories, and anyone reading this as an argument for permanence has read it wrong.

The resolution is to separate two things that usually move together. Stabilise geography, rotate quality. Keep a driver in a familiar area long enough for place knowledge to compound, and rotate the good and difficult work within that area rather than moving them across the map. And where territory changes are genuinely needed, schedule them at a cadence slow enough to build familiarity and fast enough to be fair, rather than whenever an optimiser proposes one.

That satisfies both arguments. Familiarity accumulates because the geography is stable. Fairness holds because the workload inside it rotates.

What to measure

Five measures, none requiring new instrumentation.

Familiarity index. Consecutive exposures to a territory per driver, over a rolling window. This is your familiarity stock, and most operations have never calculated it.

Completion-time delta. For the same route, the difference in completion time between a driver with high exposure and one with low. This converts familiarity from an intuition into a number, and it is the number that makes the internal case.

Familiarity half-life. How quickly that delta reappears after a driver has been away from a territory. It tells you how long a rotation cycle can be before you are paying full relearning cost every time.

Share of stops served by a low-familiarity driver. Your exposure at portfolio level. If it is rising, either turnover or reassignment policy is consuming the asset.

Knowledge capture rate. Of the six components in the table above, how many are held as system attributes rather than in drivers’ heads. This is the only one of the five that is a solution metric rather than a diagnostic.

Also Read: The Three-Workforce Fleet Reality: How Owned, 3PL, and Gig Drivers Actually Operate at Most Enterprises

How Locus turns place knowledge into system state

Locus, the world’s first Decision-Intelligent, Agentic TMS, addresses this from the capture side rather than the policy side, which is the more durable fix. The relevant capability set sits in rider and driver management alongside dispatch.

Within DiSCO, the Dispatch agent plans against more than 250 real-world constraints per computation, and those constraints are where place knowledge becomes institutional: address-level access attributes, parking and stop-type conditions, per-site dwell expectations, and known failure patterns all enter the plan as modelled inputs rather than as things a driver happens to remember. The Hub agent models facility readiness and dwell from site history rather than a network average, which captures the single most valuable component in the familiarity table. The Capacity agent holds the roster with exposure and qualification state, which is what makes an assignment policy expressible as a rule rather than a habit. And because the cycle is Sense, Decide, Execute, Learn, an exception that a driver would previously have learned to anticipate becomes an input the system anticipates instead.

The claim worth stating precisely: this does not eliminate familiarity as a factor, because a human driving an unfamiliar street will always be slower than one who knows it. What it changes is how much of the gap is knowledge the operation owns rather than knowledge that resigns.

Locus 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.

Two deployments show place knowledge moved into the system.

A global lottery operator running field service across more than 25 US states had the harder version of this problem, because familiarity there is not only geographic but regulatory and technical. Contracts, labour laws, SLA windows and zones differed by state, six job types each required different skills, and rosters shifted through the day, so every assignment was a three-way match of case, skill and location. That knowledge cannot live in individual heads at that combinatorial scale. Locus models each state’s contracts, labour laws, SLA windows, zones and skills as live constraints, with the Capacity agent maintaining the roster and the Dispatch agent matching against it. The operator reported SLA penalty risk down 20%, fuel spend down 18%, and drive distance and time down 15%.

A leading Canadian grocery brand shows what it costs to leave the knowledge in people. Carrier selection was a manual judgement call, with the team checking each order against serviceability sheets line by line and comparing rates and ETAs order by order, so, in the operation’s own description, the allocation logic lived in planners’ heads rather than in a system. That is familiarity as a single point of failure at the planning layer rather than the driving layer, and the same fragility applies. After moving order creation and carrier selection into autonomous orchestration, the brand reported 33% faster deliveries, 15% lower fulfilment costs, and 25% less time spent on manual shipping tasks.

Also Read: The Real Cost of Manual Dispatch in North American 3PLs

Price the asset before you spend it

The practical next step is one calculation. Take a route served by both a high-exposure and a low-exposure driver in the last quarter and compare completion times, failed attempts, and dispatch calls. That difference, multiplied across the reassignments you make in a year, is what familiarity is worth in your operation.

Then decide deliberately rather than by default. Re-optimise inside the day as aggressively as your system allows, because familiarity is untouched. Change territory ownership on a schedule, with the relearning cost acknowledged, rather than whenever a marginal distance saving appears. And start capturing the six components, because the version of this problem worth solving is not choosing between familiarity and flexibility. It is making flexibility cheap by owning the knowledge.

An operation that holds place knowledge in its planning system can reassign freely. An operation that holds it in its drivers is one resignation away from relearning a territory it already paid to learn.

Book a Locus demo to see how address attributes, per-site dwell, and failure history are modelled as constraints rather than left in drivers’ heads.