Carrier Selection: How AI Picks Which Provider Gets an Order

Key Takeaways

1. AI picks a delivery provider by pricing one order against every way it could move: each eligible third-party delivery provider, parcel, a batch with a nearby order, and your own drivers at true cost

2. Each quote gets corrected by that provider’s record for the zone, hour, package type, and weather.  

3. The lowest true cost that keeps the promise wins. Speed past the promise is money spent on nothing, and nearly 90% of shoppers say a late delivery weakens their relationship with the retailer (AlixPartners, 2026).

4. AI should earn authority one segment at a time, from Manual to Autonomous, and lose it the moment a hard rule breaks.

5. Every assignment should leave a sealed record of the pick, the price, every runner-up, and why each one lost, with the runner-ups ready as the fallback. On Burq, that record is the Decision Ledger.

It’s 3:38 p.m., the order is packed, and the customer wants it between 4 and 6. You’ve got three ways to get it there, but the right choice depends on more than the rate sitting on the screen.

One provider is at $14, though they’ve been inconsistent lately. Another costs more but has been hitting its windows. Your own van is nearby too, but adding this stop could put the rest of the route behind. What looks like a simple carrier selection decision is really a balancing act between cost, reliability, capacity, timing, and everything else moving through the network.

Your best dispatchers already know how to make those calls. They’ve built up the judgment to recognize when a lower rate is worth the risk, when the fleet has room, and when one extra stop will cause trouble later. The problem is that judgment usually lives in someone’s head, which makes it hard to apply consistently once hundreds or thousands of orders are coming through at the same time.

The natural next step is to make that decision process repeatable. Burq Intelligence takes the same factors your team already weighs, decides which option gets each order, and writes down why.

Take this hypothetical Culver City example. Two bags of chilled groceries need to travel 2.3 miles. The numbers are illustrative, but what the system considers before choosing who gets the order is the point.

Carrier Selection Starts With Every Way the Order Could Move

Before the AI can pick the best option, it has to know what the options are.

For the Culver City order, Burq surfaces four possible ways to move it, including a gig car, a courier sedan, parcel, and the store’s own Van 2. Then it starts narrowing the field. Parcel arrives tomorrow, so it misses the window. A store rule for chilled groceries, based on the team’s own feedback, removes the gig car too.

This first pass is all about feasibility. If an option can’t meet the promise, it shouldn’t survive long enough to compete on price.

Your own fleet belongs in that comparison too. Leave it out, and you can end up paying a third party while an available van sits nearby. Burq’s multi-carrier delivery network brings Uber, DoorDash, Roadie, UPS, FedEx, and others, negotiated rates, and your own drivers into the same decision, without owning drivers or vehicles it needs to favor. We’re also completely neutral and play no favorites.

Each Option Gets Scored by Zone, Hour, and Package Type

Once the impossible options are gone, the remaining quotes still aren’t equal. A $14 offer at noon on a dry Tuesday can be a very different bet at 5 p.m. in the rain.

Burq’s Decision Graph adds that context by looking at how each provider performs in this zone, at this hour, with this type of package. In the Culver City walkthrough, for example, the lane is running 98% on time, rain is due at 3 p.m., and traffic is adding 12 minutes.

Now a provider’s history starts changing how its quote should be read. If Provider C scores 81 during the weekday 4-to-7 p.m. window and typically runs six minutes late, the system can account for that before comparing it with the other options.

A citywide average can’t tell you any of this. Carrier selection gets useful when reliability becomes specific to the order in front of you.

The Lowest True Cost That Keeps the Promise Wins

The lowest true cost that keeps the promise wins. A cheap quote can lose on risk, and a fast one usually loses on price.

In Burq’s walkthrough, Van 2 takes the Culver City order at $6.40 and lands inside the window with 36 minutes of slack. The courier sedan could make the same window for $3.50 more. Another walkthrough order, three boxes promised for tomorrow, goes parcel at $9.80 while a 90-minute option sits at $24.60. Nobody ordered that speed, and somebody would have paid for it.

Burq decides promise-first because the promise is what customers hold you to, and grocery promises are tight. Shoppers expect free grocery and food delivery within 0.9 days, against 3.2 days for large general merchandise, and nearly 90% say a late delivery weakens their relationship with the retailer even after an apology. 

It's why Burq's founders argue for predictability over raw speed: the promise is the product.

Authority to Book the Winner Is Earned Segment by Segment

Finding the best option is only half the job. The next question is whether Burq can book it on its own.

Burq handles that segment by segment. Same-day orders under 10 miles, for example, can move through four modes: Manual, Shadow, Review, and Autonomous. Your team starts in control, then gives the system more authority as its decisions prove reliable.

The Culver City order's segment already runs in Autonomous, but it didn’t get there on day one. In Burq’s walkthrough, a segment has to hit human-approved thresholds before moving up: 60% agreement over 150 decisions to move from Shadow to Review, and 90% acceptance over four weeks to move to Autonomous, each with a person signing off.

That’s the point of earned autonomy: the system gets more freedom only after your own operating history says that it’s earned it.  

Every Assignment Leaves a Reason and a Priced Fallback

Autonomy works better when your team can see exactly why the system made the call.

Every assignment goes into Burq’s Decision Ledger with the winning price, the alternatives, why each one lost, who had booking authority, and how the delivery turned out. For the Culver City order, the record is simple: Van 2 won because it could meet the promise at $6.40, then arrived at 5:24.

The runner-ups matter too. Because Burq prices them before booking, the next-best option is already there if the first choice declines or conditions change. You’re not starting the decision over from scratch.

That audit trail also makes AI easier to trust. Burq calls it the third record: alongside what moved and what it cost, you keep a record of why the decision was made.

Peak Season Delivery Will Test Every Carrier Selection Engine

Peak season is where all of this gets tested for real, and we made a whole peak demand playbook for it, because nothing ever stays stagnant. The carrier rule that worked in September will be wrong by Cyber Monday when rates move, capacity tightens, and your own fleet is suddenly busier than expected.

Carrier selection has to move with the day. Burq Intelligence looks across your provider network and your own fleet, scores the options against the customer promise, and chooses the best fit based on what’s happening now. Dispatch OS can book that choice at the authority each segment has earned, while the Decision Ledger keeps the reason, the alternatives, and the fallback on record.

FreightWaves named Burq to its 2027 FreightTech 100 in September, but co-founder Salman Habib puts the idea more simply: “No order in commerce should move on a guess.”

Book a demo and give Burq one of your own orders to decide.

FAQs 

How does AI decide which delivery provider gets an order?

It prices the order against every eligible provider and your own drivers, drops options that can’t meet the order’s weight, handling, or window, then scores the rest on each provider’s record in that zone, at that hour, with that package type. The lowest true cost that keeps the promise wins, executed at the authority that order's segment has earned.

What’s the difference between carrier selection and rate shopping?

Rate shopping compares quoted prices and takes the lowest. Carrier selection also asks whether each provider will keep the promise on this lane, and it weighs your rules, contracts, and fleet’s true cost. A less expensive courier that runs late at 5 p.m. in the rain can cost more once a redelivery, a refund, and a support ticket are counted.

Can AI compare third-party delivery providers with my own drivers?

Yes. Burq prices your in-house fleet as one more provider on every order, at the true cost of the run: who’s free, who’s already on a route, and whose shift the stop would overrun. Your drivers win when they keep the promise for less, and third-party delivery fills the rest, so build versus buy gets settled on every dispatch.

Can I keep my negotiated carrier rates and block providers I don’t want?

Yes. Sync existing agreements, such as a UPS account or a regional courier contract, and those rates compete with network rates on every order. Pin preferred providers, exclude others, and set rules by zone, order type, or service level. Weekly minimums and fee ceilings apply before price, and each decision names the rule it followed.

How should carrier selection change during peak season?

Reprice every order instead of trusting rules written in September. Surcharges, courier capacity, and on-time performance shift through the day in November and December, and Adobe expects U.S. online holiday sales to hit $275.1 billion, up 6.7%. Keep new order types under human review, and pull autonomy from any segment that breaks a hard rule.