How dispatch strategies work

A dispatch strategy is one saved ruleset you build once and reuse — on an order, as your org's default, or from a workflow. This article is about what that ruleset actually does when a dispatch happens: how it narrows and orders the providers you allow, what "reliability" means when you lean on it, and what happens automatically when a provider falls through. For the strategy object itself — where you find it, how it's attached — start with dispatch strategies. For how a dispatch turns an order into a job in the first place, including everything that decides which providers and prices are even in the running, see how dispatch works — this article picks up after that: once the eligible, priced quotes exist, the strategy is what chooses among them.

The model

Every dispatch strategy works from the same two ingredients. The first is yours: the allowed-provider list you set on the strategy — the providers (or your own fleet) you've told it it's allowed to consider at all. The second isn't yours to set here: the eligible quotes the rest of Nash's configuration produced for this order — which contracts qualified, and at what price. A strategy never sees a provider you didn't add to its list, and it never sees a quote that didn't come back eligible in the first place. What a strategy actually does, every time it runs, is take the overlap of those two sets and decide which one wins.

That's the whole job: order or weight the eligible, allowed quotes, and pick one. Everything else in this article — the five ways it can do that, the reassignment rules for when the pick falls through — is detail on top of that one operation.

The five selection modes

The mode you choose on Provider Selection decides how the strategy orders the providers you've allowed. Two modes are deterministic — the same inputs always produce the same order. Three are probabilistic — they influence the pick toward an outcome without guaranteeing the single top-ranked provider wins every time.

Mode	How it picks
Lowest Cost	Deterministic. Ranks your allowed providers by price and takes the cheapest eligible quote first.
Manual Ranking	Deterministic. Walks your allowed-provider list in the exact order you set and takes the first one with a valid quote.
Most Reliable	Probabilistic. Favors providers with a higher reliability score — a weighted pick biased toward reliability, not a guarantee the single most-reliable provider is chosen every time.
Weighted Distribution	Probabilistic. Splits volume across your allowed providers by the per-provider percentages you set (they must sum to 100), honoring those shares over time rather than on any one delivery.
Dynamic	Probabilistic. Balances reliability against price together, rather than optimizing for either alone. Covered on its own below.

That distinction is worth sitting with, because it's easy to read "Most Reliable" or "Weighted Distribution" as a fixed rule and be surprised when a single delivery doesn't follow it. It isn't a fixed rule. Most Reliable favors the more reliable providers in your list — it doesn't lock in the top-scored one every single time. Weighted Distribution distributes toward the percentages you set — a 70/30 split shows up as roughly 70/30 across many deliveries, not as a guarantee on any individual one. Only Lowest Cost and Manual Ranking behave like a strict, repeatable ordering; the other three are designed to spread outcomes across your allowed list rather than hand the same provider the win every time.

Whichever mode you use, the strategy is only ever ordering an operator-controlled allow-list — the providers you added, filtered down by whatever else made a given quote eligible for this particular order. Adding a provider to a strategy doesn't guarantee it ever gets picked, and leaving one out means the strategy never considers it, no matter how good a price it might have offered.

Reliability score

Most Reliable ranks providers by a reliability score that Nash computes from each provider's past performance — completion rate, cancellation rate, and other signals — and keeps up to date on its own. You don't calculate it, tune it, or set it anywhere; the only thing you do is opt in to using it, by choosing Most Reliable as your strategy's selection mode. There's no published formula or weighting behind the score, and no field in the portal to adjust it directly — it's a Nash-side signal you can lean on, not a setting you configure.

Keep that distinct from the Reliability Index, a separate dial that appears only under Dynamic mode — covered further down. The reliability score ranks providers for Most Reliable; the Reliability Index tunes how much weight reliability gets inside Dynamic's own balancing act. They sound alike and both involve "reliability," but they're different controls behind different modes.

Auto-reassign

A dispatch strategy doesn't stop working once a provider is picked. Three independent triggers can hand the delivery to the next eligible provider in the strategy — or to a different strategy entirely, if you've configured one as the fallback:

• On cancellation. If the assigned provider cancels, the strategy reassigns automatically to the next eligible provider on your list.
• On pickup time. If the delivery hasn't advanced past a status you choose by the time you set relative to pickup, the strategy reassigns it.
• On creation time. If a delivery goes unassigned for a number of minutes after it's created, the strategy reassigns it to the next eligible provider.

Each of these is its own switch — you can turn any combination on, and each one fires independently of the others when its own condition is met. Two more behaviors shape what reassignment actually does once it's triggered:

• Smart Reassignment doesn't just retry blindly. Before handing a delivery to the next provider, it weighs whether another attempt is likely to succeed at all — and if it isn't, the delivery is canceled instead of being reassigned repeatedly toward a very unlikely success.
• Split Batch applies when a whole batch, not a single delivery, is canceled by a provider. Rather than reassigning the batch as one unit, it splits it into individual deliveries and reassigns each of them.

Warning

Don't turn on auto-reassign on creation time and on pickup time together on the same strategy. The two triggers are watching for different conditions on the same delivery, and running them at once can work against each other rather than reinforce one another. Pick the one that matches when you actually want a reassignment to fire.

It's worth being precise about what auto-reassign actually moves. Nash dispatches; the provider or your own fleet performs the delivery. Auto-reassign moves the dispatch — it hands the job to the next eligible provider — it doesn't reach into an in-progress delivery and take it away from a driver who's already carrying it. A reassignment that lands close to the moment a provider starts the delivery can, in a tight race, still incur a cancellation fee — not something a strategy can guarantee away entirely.

Dynamic mode and the Reliability Index

Dynamic is the fifth selection mode, and it works differently from the other four: instead of optimizing for one thing — cost, your manual order, or reliability alone — it balances reliability against price together, aiming for the better overall outcome rather than the cheapest or the most reliable option in isolation. Like Most Reliable and Weighted Distribution, it's probabilistic — it favors a good balance, not a single guaranteed pick.

Dynamic exposes one adjustable control: the Reliability Index, a dial that shifts how much weight reliability gets relative to price in that balancing act. Turn it up and Dynamic leans further toward reliable providers even at a higher price; turn it down and it leans further toward the cheaper option. This is a different lever from the reliability score covered above — the score is a Nash-computed ranking signal, while the Reliability Index is a dial you turn to tell Dynamic how much that signal should matter to it.

Note

Dynamic mode is available depending on your organization's setup — reach out to Nash if you don't see it. It's closely related to Nash's broader automated-monitoring capabilities; see what Nash Agent is for the wider picture of what Nash's AI agents do, of which dynamic balancing is one piece. This article covers only what Dynamic mode does inside a dispatch strategy, not agent setup.

What affects this

None of the following live on the strategy's own selection-mode field, but each one shapes what the strategy actually has to choose from, or how it chooses.

Input	Where it's set	Effect on a dispatch strategy
Allowed providers and selection mode	The strategy itself	Which providers are even in the running, and how the strategy orders or weights them
Providers, contracts, prices, and zone coverage	Configure ▸ Network — owned by how dispatch works	Sets the pool of eligible, priced quotes the strategy chooses among — a provider missing from that pool never reaches the strategy at all
Reliability score	Nash-computed	Ranks providers under Most Reliable and feeds the Dynamic balancing act
Your org's default strategy	Settings preference	Which strategy applies to an order that doesn't name one directly
Workflow "Set dispatch strategy"	Automate ▸ Workflows — see what workflows are and build a workflow in the portal	Assigns a strategy to matching orders automatically, based on the conditions you define

Two strategies that look identically configured can still behave differently in practice if the providers, contracts, or zone coverage behind them differ, or if one is leaning on a reliability score that's shifted since you last checked it. When a dispatch doesn't go the way you expected, it's worth checking whether the provider you wanted was ever in the eligible set to begin with — the strategy can only choose among quotes that actually arrived. Nash dispatches; the provider or your own fleet performs the delivery.