Route-planning software takes each day's stops, their delivery or appointment windows, how long each stop takes, each vehicle's capacity and each driver's shift, then decides which vehicle does which stops and in what order, keeping driving and late arrivals to a minimum. You import the stops each morning, review the plan, and send each route to the driver's phone.
Two things surprise most small fleets. First, most of the saving comes from deciding which vehicle takes which stops, not from reordering stops within a route. Second, the plan is only as good as the stop data: a wrong map pin, an unrealistic ten-minute stop that really takes twenty-five, or a delivery window nobody actually needs will cost more miles than any software saves.
How a route optimiser works it out
The puzzle has a name, the vehicle routing problem, and it gets hard quickly: with 40 stops and three vans, the number of possible plans is too large to check one by one. Route planners use search methods that find a very good plan in seconds rather than a perfect one in hours.
They juggle rules of two kinds:
- Must-haves: every stop visited once, inside its time window; no van loaded beyond capacity; no driver past the end of their shift; vehicle-specific stops (a tail lift, a refrigerated van) only on the right vehicle.
- Preferences to balance: fewer kilometres, fewer hours, routes of similar length so one driver isn't always finishing late, and sometimes the same driver for the same customers.
The "AI" part has grown in recent years around that core: predicting travel times by time of day, learning how long stops really take from drivers' completion times, cleaning up badly written addresses, and re-planning in the day when a stop is added or a van breaks down. For a small fleet, the learned stop times are often the most valuable of these, because they're the number people guess worst.
Delivery fleets and mobile teams need different settings
The same software serves both, but the bottleneck is different:
- Delivery fleets have many short stops (two to ten minutes), so driving dominates the day and capacity (weight, volume, cages) often limits how many stops a van can take. Parcel couriers add collections and failed-delivery retries; route planning for a small courier firm covers those.
- Mobile teams (cleaners, maintenance crews, carers, mobile groomers) have fewer, longer stops of an hour or more, so time on site dominates and the order within a route matters less than who does which job. Crews carrying equipment face their own version, described in route planning for landscaping crews.
If your team's day is mostly skilled jobs with qualifications and customer relationships, you need dispatch as well as routing; AI job scheduling and dispatch for field service teams covers that side. The rest of this tutorial is about the route itself.
A three-van laundry round, before and after
An illustrative commercial laundry delivers clean linen and collects used linen from 41 customers a day: boutique hotels, guest houses and holiday-let managers. It runs three vans with drivers on 6am to 2pm shifts. For years, the routes were fixed by area: van 1 north, van 2 town centre, van 3 south and the coast.
The problems were familiar. Hotels wanted linen before 11am for their housekeeping teams, but the fixed order meant some got it at 12:30. The coast van ran long every Saturday in summer, when holiday-let changeovers doubled its stops. And nobody had revisited the routes since the laundry took on eight new customers in the town centre.
The laundry trialled a route planner for two weeks. Each morning, the office exported the day's orders from its order system as a CSV, uploaded it, and compared the planner's routes with the fixed ones. Illustrative results for a typical weekday:
| Measure | Fixed routes | Planned routes |
|---|---|---|
| Total driving | 312 km | 251 km |
| Hotels served after 11am | 5 | 1 |
| Last van back | 2:20pm | 1:35pm |
| Van with spare time | None | Van 2, back by 12:30 |
The planner didn't simply reorder the old routes. It moved six town-centre customers onto the north van, which passed them anyway, and gave van 2 the early hotel windows. The quick sum: 61 km a day saved, over 22 working days, is about 1,340 km a month. At an illustrative running cost of $0.35 per km for fuel and wear, that's about $470 a month, against a planner costing $125 to $150 a month (see prices below). The bigger gain was van 2 finishing by 12:30, which let the laundry take on new customers without a fourth van.
The stop data that makes or breaks the plan
Every planner accepts a spreadsheet import. These columns, filled in properly, matter more than which tool you buy:
| Column | Example | Why it matters |
|---|---|---|
| Address, or map coordinates | Full address; coordinates for rural sites | A wrong pin sends a driver kilometres off |
| Window start and end | 07:00-11:00 | Only add windows customers really need |
| Service time (minutes) | 8 | Too low and every route runs late |
| Load | 3 cages | Stops vans being overfilled |
| Requirements | Tail lift | Matches stops to suitable vehicles |
| Notes for driver | "Use rear gate, code on job" | Saves the phone call from the car park |
Addresses and pins. Rural properties, holiday cottages and sites with several entrances are where map pins go wrong. At the laundry, one holiday-let cluster was pinned to the centre of the nearest village, and the planner kept routing the coast van 6 km in the wrong direction before doubling back. The driver knew and ignored the order; the office didn't know until the trial showed that stop always arriving "late". The fix was to save exact coordinates for that customer once. Do this for every stop a driver says is misplaced, because the planner will repeat the error every day.
Service times. The laundry had set every stop at 5 minutes. Proof-of-delivery timestamps from the drivers' app showed hotels averaged 14 minutes (the linen goes up in a lift, and the old linen has to be counted), while guest houses averaged 6. With realistic times, the planner stopped giving van 1 more stops than it could finish.
Time windows. Windows constrain the plan more than anything else, so don't add them by default. In figures, from the laundry: one guest house asked for "between 9 and 9:30". Keeping that window added 9 km to van 3's route because the van had to leave the coast road and return. When asked, the owner was happy with "before 11". Across the customer list, relaxing windows that nobody really needed saved another 14 km a day.
When the plan breaks at 10am
Morning plans rarely survive the whole day. A van breaks down, a hotel phones with an urgent extra order, or a road closes. Most planners let the office add or remove stops and re-plan the remaining routes, keeping completed stops fixed. At the laundry, a flat tyre on van 3 at 9:40 left nine coastal stops unserved. Re-planning split them between vans 1 and 2 and showed that two guest houses would now get deliveries after their windows. The office phoned those two customers; the planner couldn't do that part. Before accepting any re-plan, look for exactly that: which stops now miss their windows, and whether any driver now runs past the end of their shift.
Where a chat assistant helps, and where it doesn't
It's tempting to paste a list of stops into ChatGPT or Claude and ask for the best order. For a handful of stops it will produce a plausible order, but it's estimating distances from general knowledge of places, not calculating them from road data. For anything beyond a few stops, use a real planner.
Where a chat assistant does help is preparing the stop list. The laundry receives many orders by email, with windows and notes buried in the text. A prompt like this turns them into import rows:
Turn these customer order emails into rows for a route planner CSV.
Columns: customer | address | window_start | window_end | service_minutes |
load_cages | requirements | driver_notes
Use our defaults unless the email says otherwise: window 07:00-11:00 for hotels,
07:00-14:00 for others; service 14 min hotels, 6 min guest houses and lets.
If an email changes a regular order, add "CHANGED" to driver_notes.
If anything is unclear, write UNCLEAR in that cell rather than guessing.
Emails: [paste]An illustrative output row: Harbour guest house | [address] | 07:00 | 14:00 | 6 | 2 | none | CHANGED: extra cage this week, leave in the porch. What needed fixing in the first week: the assistant put one email's "can you come after 8?" into the window as 08:00-11:00, applying the hotel default to a guest house. The owner added "if a customer gives only a start time, use our default end time for that customer type" to the prompt.
For very small runs, Google Maps is free and fine: it allows up to 10 locations in one route, counting the start and end, so around eight stops. You choose the order yourself, and it doesn't handle windows or split stops between vehicles.
What route planners cost
Two examples checked on the vendors' pricing pages in September 2026, both charging by volume rather than per vehicle:
- Routific: free for up to 100 orders a month; $150 a month flat for up to 1,000 orders, with unlimited drivers and dispatchers; then from $0.15 per extra order, falling at higher volumes. SMS customer notifications are an add-on.
- Spoke Dispatch (formerly Circuit for Teams): Starter $125 a month for 1,000 stops, then $0.04 per extra stop; Premium $200 for 2,000 stops. Unlimited drivers on each plan, with time windows and proof of delivery included.
The laundry's volume, 41 stops on 22 days, is about 900 a month, which sits inside both entry plans. A business doing 60 stops a day would pass 1,000 a month and pay per extra stop, so do the sum on your busiest month, not your average one. If you already use field-service or property-management software, check its built-in routing first: Jobber, for instance, includes route optimisation on several of its plans, and it may be enough for a mobile team.
A holiday-let changeover day: time, not distance
A mobile team's routing problem looks different. Suppose a holiday-let manager has 14 changeovers on a summer Saturday, three cleaning crews, and a hard window of 10am to 4pm at every property. Each clean takes 90 to 150 minutes depending on the size of the cottage, and drives between properties are 10 to 25 minutes.
Here, the planner's main job is assigning cottages to crews so every clean fits the window. The manager entered each cottage's clean time, the 10am-4pm window and each crew's start point. The planner produced five cleans for one crew and four and five for the others, and flagged that one crew's last clean would finish at 4:10pm. The fix was a person's decision: that cottage's arriving guests had said they'd arrive at 6pm, so the manager moved its window to 10am-5pm, and the plan worked. The software found the problem at 7am instead of a crew discovering it at 3:30pm.
Two extra settings helped. Linen for each cottage was loaded in the order of the crew's route, so nothing had to be dug out of the back of the car. And the hot tub checks, which only one person was trained to do, were tagged to that person, so the planner built her route around them.
Volunteer drivers: a church food bank's delivery round
Routing isn't only for businesses with vans. Take a hypothetical food bank run from a church hall, delivering about 36 parcels on Friday afternoons using six volunteers in their own cars, each available for about two hours. The coordinator used to hand out lists by neighbourhood, which left some volunteers with 45 minutes of driving and others with 10.
With a free-tier planner (36 stops a week is well under 100 a month on Routific's free plan, for example), the coordinator entered each volunteer's start address, a two-hour shift, and each car's capacity in parcels: six for most, ten for the volunteer with an estate car. Routes came out balanced, and the longest round fell from about two and a half hours to under two.
The detail that mattered most wasn't mileage. Recipients' addresses are sensitive personal data, so each volunteer received only their own route on their phone, not the full list, and routes were deleted from the planner after each week's run. If you handle vulnerable people's details, check your planner's data retention settings and follow data-protection law such as the GDPR; ask your data-protection adviser if unsure.
Getting drivers to follow the plan
A route plan that drivers ignore saves nothing. Drivers ignore plans for good reasons more often than bad ones: they know where parking is impossible at 8am, which customer's loading bay is blocked on Tuesdays, which road floods. Treat every deviation as information.
- Ask drivers why when they change the order, and turn the answer into a window, a pin correction or a note.
- Load in reverse stop order, so the first stop's goods are nearest the door.
- Keep routes stable where you can. Drivers and customers both like familiarity; many planners let you favour the same driver for a customer.
- Share the results. At the laundry, telling drivers the coast van now finished 40 minutes earlier on Saturdays did more for adoption than any instruction.
- Let the planner text customers. Most planners can send an estimated arrival time, which cuts the calls drivers take on the road; automatic delivery updates shows the wording and the effect on "where is it?" calls.
Proving the savings with real mileage
Compare the same weekdays before and after, because Mondays and Saturdays differ more than any software change. Track five numbers for a month:
- Kilometres per stop, from vehicle mileage, not the planner's estimate.
- Stops per driver hour.
- Stops arriving outside their window.
- Overtime or late returns.
- Customer complaints about delivery times.
Use the vehicles' real mileage for the first. Planners report their own estimated distances, which assume drivers follow the plan exactly. If real kilometres per stop don't fall within a month, look for the usual suspects: wrong pins, unrealistic service times, too many narrow windows, or drivers who have good reasons for not following the order that nobody has written down yet.
Route planning questions from small fleets
Can I just use Google Maps for multi-stop routes?
For a short run, yes. Google Maps allows up to 10 locations in one route, counting the start and the end, so about eight stops in between. It doesn't reorder them for you, handle time windows or split stops between vehicles. Once you have more than one vehicle or more than about eight stops, a route planner saves real time.
Do drivers have to use the planner's app?
Most planners send each driver their route in a phone app with navigation and proof of delivery, and that's how you get accurate stop times back. Some let drivers open each stop in their preferred navigation app. If drivers won't use the app, you lose the data that improves future plans, so involve them early.
How much can route optimisation save?
It depends on how your routes are planned today. Fixed routes drawn years ago and never revisited often leave a lot of spare mileage; routes a sharp dispatcher plans daily leave less. Run the planner alongside your current routes for two weeks and compare kilometres and late stops on the same weekdays to find your own figure.
What about vans with a tail lift or drivers with different licences?
Most planners let you tag vehicles and stops with requirements, so a stop needing a tail lift only goes to vans that have one. Check this during the trial, because it's the kind of rule that is easy to forget until a driver arrives unable to unload.
Further reads
- Jobber, Housecall Pro, and ServiceTitan AI Features Compared — Built-in routing and dispatch in the main trades platforms.
- Photo Checklists and Quality Control for Cleaning Teams Using AI — Photo checklists for cleaning crews once routes are sorted.
- AI Shift Scheduling vs a Rota Spreadsheet for Small Teams — Plan who works which day before planning their routes.
- How Mobile Dog Groomers Use AI for Routes, Reminders and Rebooking — Routing plus reminders for a one-van mobile business.
- AI Appointment Scheduling for Small Businesses: Tools and Setup — Booking time slots that routing can actually meet.
- Fleet Admin for Small Businesses: Automate Services, Checks and Fuel — Automate the servicing and checks for the vans themselves.
- How Dry Cleaners and Laundries Use AI for Orders and Collections — Where AI fits between the counter and collection: phone agents that book pickups, ready texts, an uncollected-garment ladder and calmer claim replies.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Routific pricing page; Spoke Dispatch pricing page; Google Maps help community and route-planner documentation on the 10-location limit.