An abstract illustration in which a booking sheet, a map-like area, a dispatch sheet, and notification signals are placed separately, while lines and shapes in the middle connect them into one route.
Platform Rulesthrough Where the Work Breaks

A gap in route migration opens up tools for running the work around routes

Moving to new routing features creates an opening for small operations tools that connect booking sheets, maps, dispatch plans, and customer notifications.

Published 2026. 11. 12.

Legacy routing is blocked in new projects

Google classified the Directions API as a legacy product on March 1, 2025. An API is a connection point that lets one program ask another for a route or arrival time. From October 2026, new Google Cloud projects will not be able to enable this feature, so anyone building a map service now must use the newer Routes API.

Here, a project means the workspace where map billing and permissions are managed together. Projects that already have legacy routing enabled will continue to be supported, so an existing service will not stop immediately. Google also has a policy of giving at least 12 months’ notice when an actual shutdown is set.

The legacy browser tools that calculated routes and drew them on a map were also classified as deprecated as of February 25, 2026. That means teams need to review not only the server-side route calculation, but also the part of the customer-facing screen that draws route lines. This is not finished by keeping the existing interface and changing a name.

The new feature differs in how a route is requested, what distance and time fields are called in its results, and how it reports errors. You must explicitly choose the result fields you need, and the formatted origin and destination addresses included in legacy results do not appear in the same way. First identify which parts of address display, arrival-time calculation, and route-line drawing depend on the legacy feature.

The way costs are counted can also differ from the number of calls visible on screen. Comparing travel times for eight technicians and 20 stops might be one request, but it can involve 160 origin-destination combinations. Google’s pricing documents count a single route and a comparison across multiple combinations as different types of usage, so compare both the results and the number of calculations before and after migration.

Where the work breaks between the booking sheet and the customer text

Consider the operator of an appliance repair business with 12 employees. A booking coordinator enters addresses and requested time slots received by phone into a booking system. In the afternoon, they export the next day’s bookings into an Excel file. They paste addresses into a map one by one to check travel times, then write the visit order back into Excel.

They send the completed order to group chats for each technician. A technician pastes the first address into navigation, then moves each next address in the same way after finishing a job. When traffic causes a delay, the technician calls. The booking coordinator opens the booking sheet, finds the next customer’s phone number, and sends a delay message.

In this process, the same address repeatedly moves from the booking system to Excel, from Excel to maps and messaging, and from messaging to navigation. If a customer name and detailed address are mismatched, or the visit order changes, it is hard to find the stage where the error occurred. A routing-feature change is a reason to revisit this connection structure.

A tool that closes this gap could let an operator upload a booking file once and receive candidate routes for each technician. The screen could show the distance, estimated arrival time, and visit-order differences between the old and new methods side by side. Once the operator confirms the order, the address and sequence go to the technician, while customers receive arrival updates only when needed.

Do not compare every technician with every stop by default. First narrow candidates by service area, the customer’s promised time, and the types of products the technician can repair. Then calculate travel times within that set. This avoids unnecessary usage such as comparing all eight technicians with all 20 stops every time.

Some judgement still belongs to people. A difficult-to-park apartment complex, a repair that needs parts first, a fault that only a specific technician can handle, or a customer who does not answer the phone cannot be decided from a map alone. The tool should show a recommended order and its reasons, while leaving final confirmation and exception handling to the operator.

People running an existing service need an inventory of usage locations and comparison tests more than an urgent full replacement. Record which project powers which screen, how many route calculations happen each day, and how much arrival times differ in the new results. During migration, a screen can also collect affected bookings and hand them over to manual dispatch and customer communication when errors occur.

Elsewhere, teams changed the connections first

In the United States, ServiceNow connects Google Maps to route display and route optimization in its field-service dispatch product. Businesses use the ServiceNow product, while Google route-calculation usage is handled through each business’s own maps account. Its official documentation says that customers using legacy routing need a compatible product update to use the new map connection settings.

What matters in this case is that routing was not attached as an isolated feature. The dispatcher’s screen, technician routes, and travel information shown to customers moved together with the product version. ServiceNow reorganized these connections so that its current product uses the Routes API. No public time or cost savings figures were provided.

Japanese delivery service menu (menu) originally queried the legacy Directions API separately for each courier and destination. It later introduced a Routes API feature that compares real road travel distances across multiple couriers and multiple pickup candidates at once. This was not simply replacing a single-route request with a new method; it rebuilt the process for selecting a courier.

According to a Google customer story, average pickup travel distance was cut in half and combined convenience-store orders doubled in two months. It also says that route queries previously split across multiple requests were consolidated into one or two requests. These performance figures were published by the company and supplier, so results in another market need separate verification.

Canadian healthcare delivery company Bayshore HealthCare moved from entering addresses into Google Maps one at a time to using Onfleet. Onfleet combines route planning, driver dispatch, real-time changes, and proof of delivery in one place through a monthly subscription model that includes completed delivery and pickup jobs. It is an example of turning the work people moved around before and after route calculation into a product.

According to an Onfleet customer story, Bayshore increased daily volume from 300 to 2,000 jobs, while route-planning time for one dispatcher fell from 30–60 minutes to 5–10 minutes. This is also supplier-published customer performance and needs separate verification. Still, it makes clear what an interface connecting address entry, dispatch, and delivery confirmation can reduce.

Small things worth building here

1. A map-feature usage finder

This is an audit service for businesses that run booking or field-visit services through outsourced development and do not know which screens use Google routing. It takes map usage specifications, billing records, and screen addresses, then creates a list of where routing, address search, and map display are connected.

Now that new projects cannot enable the legacy feature, hidden dependencies can surface when a business changes its maintenance provider or builds a new screen. The first screen should show the feature used by each service screen, the connected project, and whether replacement is needed, using traffic-light colors.

2. A legacy-versus-new route comparison testbed

This is a service for food suppliers that operate their own deliveries with 10–30 vehicles and want to upload past delivery records to compare the two approaches. It uses the same origins and visit order to find where distance, arrival time, route lines, and calculation volume differ.

Because this is not a name-only replacement, businesses need to review both customer-facing screens and costs using data from real operating days. The first screen should include an overlaid map of the two routes, a list of deliveries with large differences, and estimated calculation volume.

3. A dispatch tool connecting booking sheets and technician instructions

This is a service for repair, cleaning, and installation businesses that build the next day’s schedule in Excel. After a booking file is uploaded, it first applies service areas and appointment times, then calculates travel time to create each technician’s visit order and customer-message copy.

It can use the new routing feature as its starting point, without forcing an old screen to be copied. The first screen should show only unassigned bookings, each technician’s schedule, and the three visits most likely to be late.

4. Customer communication when routing fails

This is a service that helps rental-collection and on-site maintenance businesses with many same-day bookings keep operating during routing errors or migration work. It collects bookings where calculation failed and provides the last confirmed route, technician phone number, and customer notification text in one place.

When map rules change, teams need to define manual work for failure cases as well as normal routes in advance. The first screen should show the number of affected bookings, the manual-dispatch waiting list, and a button to send messages.

One thing to check in 30 minutes today

Call one person responsible for dispatch or visit scheduling. Ask them to walk through the screens for five recent bookings, from intake to handoff to a technician. If an address, arrival time, or visit order was manually moved more than twice in at least three of the five bookings, that connection point is worth testing as a small tool regardless of the routing migration.

Why this matters where you are

Check where addresses, arrival times, and visit orders are copied between booking, mapping, dispatch, and customer-message tools in your own operation or market. The map provider, pricing model, and migration timeline may differ, but the handoffs around routing can still be measured. Start by tracing a small set of recent jobs and identify the repeated manual transfer that could be removed first.

Sources

8 sources

Every fact in this article came from the pages below. Check them yourself.

A gap in route migration opens up tools for running the work around routes | Prometheon