The biggest cost for a small taxi service is not software or telephony. It is the people who answer the phone. You need operators around the clock, including nights, weekends and holidays, while order flow is uneven: queues in the morning and evening, silence during the day and at night. So the service either overpays for idle time or loses calls at peak hours. Both cost money.

This article explains how a distributed call center works in taxi software, what it costs compared with in-house operators, who benefits from handing calls off, and who can take other services’ calls and earn from it. We also cover how an LLM voice robot takes routine calls off operators before they ever reach a person.

The problem: operators are expensive, and calls still get lost

Let’s run the numbers for a service with 100 orders a day in a city of 100–200 thousand people.

24/7 call handling in-house. You need to cover 24 hours, 7 days a week. That is 168 hours a week. With a normal 40-hour week, that is at least 4.2 full-time positions, or 5 people once you account for vacations and sick leave. At a salary of 30,000 ₽ in a small city plus payroll taxes, that is 180,000–200,000 ₽ a month. Across 3,000 orders a month, it comes to 60–65 ₽ per accepted order. For a service whose driver commission is 30–40 ₽ per order, such a call center loses money on its own.

Cutting corners at night. Many services leave one operator at night or simply forward calls to the dispatcher’s mobile phone. Quality drops, some calls go nowhere, and night orders are the most expensive ones.

Peak hours. From 7 to 9 in the morning and from 17 to 19 in the evening, a third of the day’s calls come in. If only one operator is working, the queue grows, and after 30–40 seconds of waiting customers hang up and call a competitor. To cover the peak, you have to keep a second operator all day for two hours of load.

Hence the dilemma: pay for idle time or lose calls. Classic forwarding to a third-party call center only partly solves it. An external operator does not see your software, does not know the city’s addresses and fares, and cannot assign a car. They just write down the order and pass it to you, which adds minutes and errors.

What a distributed call center is

The distributed call center in UpTaxi lets one dispatch service take and process calls for another dispatch service inside the same software. Not «forward to a phone», but actually take the call in its own operator console: a call from your service’s customer reaches an operator at another company, who sees an order card for your city with your map, your fares and your cars. The operator creates the order, and it goes straight into automatic dispatch to your drivers.

Nothing changes for the customer: they called the familiar number, someone answered, quoted a price, and the car arrived. For you, the only change is that you do not pay an operator’s salary. You pay for results.

Inside the UpTaxi system there are 8 independent call centers competing with each other to handle calls. You choose where to route your calls and can change your choice if the quality does not suit you.

How it works technically

• Multi-console. An operator takes calls from different cities in a single console. The software uses the number to determine which city the customer is calling from and opens the order card of the right service, with its map, addresses, fares and car statuses. The operator does not «switch between programs». They just work.
• Priorities and queues. A call is first offered to your own operators (if you have any), then to the connected call center. The rules are configurable: for example, only calls not answered within 15 seconds go to the queue.
• Remote operators. An operator needs only a computer with a headset and an internet connection. They can sit in any city. This lets call centers hire where labor is cheaper.
• Asterisk and SIP telephony. Calls go over IP telephony, so routing between dispatch services needs no hardware. Everything happens inside the software. More on telephony on the voice and SMS auto-notification page.

The mechanics are described in detail on the multi-console and call center page.

Three scenarios: fully, partly, at night

A distributed call center does not have to mean «giving up operators». It is more often used selectively.

Scenario 1. No operators at all

Suits new services and services with up to 100–150 orders a day. The LLM voice robot holds the first line: it answers the call, asks where from and where to, clarifies details and creates the order itself. Calls the robot could not handle go to the connected call center, while the owner focuses on drivers, advertising and quality. Combined with order bots in Telegram and Max and the customer mobile app, the share of calls you pay a live operator for drops even further. All these channels are included in the base package; see order intake.

Scenario 2. Overflow only

You have one or two operators who handle most of the day. At peak hours, calls that wait in the queue longer than a set time go to the call center. You pay for the overflow and do not lose customers at peak hours, without hiring someone for two hours of load.

Scenario 3. Nights and weekends

Your own operators work during the day. From 22:00 to 7:00 and on Sundays, a 24/7 call center takes the calls. This is the most popular option among services that used to «hand the night to a mobile phone».

All three scenarios are set up in the software and switched without involving the vendor.

What it costs

The UpTaxi rate for call handling by the distributed call center: 1 ₽ per accepted order plus 4 ₽ per completed order. So for an order the operator accepted and the driver successfully completed, you pay 5 ₽. For an order that was accepted but cancelled, 1 ₽.

Compare this with keeping your own operators for a service with 100 orders a day (3,000 a month).

Option	Calculation	Per month	Per order
5 in-house operators, 24/7	5 × 36,000 ₽ (incl. payroll taxes)	180,000 ₽	60 ₽
2 operators by day + night on a mobile phone	2 × 36,000 ₽	72,000 ₽	24 ₽ (plus night losses)
Distributed call center, all calls	3,000 × 5 ₽ (at 100% completion)	15,000 ₽	5 ₽
2 operators by day + call center for nights and overflow (30% of calls)	72,000 + 900 × 5	76,500 ₽	25.5 ₽

The salaries are illustrative; plug in your own. But the order of magnitude is clear: taking a call for 5 ₽ is 5–12 times cheaper than through your own operator. And when volume drops in the low season, costs fall proportionally. Payroll does not.

There is one caveat worth stating plainly. Giving up your own operators completely means you do not control who talks to your customers and how. Call centers in the system compete on quality, and you can switch, but an external operator will not reproduce the tone of «our Marina, who knows everyone by voice». In a small city where the service runs on personal relationships, that may be a reason to keep at least one «of your own» for daytime and outsource only nights and peaks.

Run the numbers on your own data: how many orders a day, how many of them at night and at peak hours, what an operator costs in your city. Access to the software with the distributed call center is free for 30 days. Get access.

The flip side: earn with your own operators

If you run a large service with enough operators, the distributed call center turns from a cost into a source of income. You connect several small services and take their calls. Your operators get a more even workload, and you get paid for every accepted and completed order.

When this pays off most:

• Your operators cost less than 4 ₽ per order. In regions with lower wages and in CIS countries, the cost of one accepted order in-house can be lower than what services from big cities pay. The difference is your income.
• You have gaps in workload. During the day your operators sit without calls. Let them take other services’ calls.
• Different time zones. If you team up with a service in another time zone, your peak hours do not overlap: when it is night for you, it is evening for them, and vice versa. Each side covers the other’s peaks, and both miss fewer calls.

This builds a market inside the system: some services buy call handling, others sell it. Neither side hands its customer base to the other. The order is created in the card of the company the customer called, and it stays there.

How this differs from plain forwarding and an external call center

	Forwarding to a phone	External call center under contract	Distributed call center in the software
Operator sees your map, fares, cars	No	No	Yes
Order goes straight into automatic dispatch	No, must be re-entered manually	No, or via integration	Yes
Caller identification and order history	No	No	Yes
Payment	Call minutes	Monthly flat fee or per minute of talk time	Per accepted and completed order
«Nights only», «overflow only» setup	Partly	Per contract	In the software, on your own
Quality control	No	Reports	Call recordings and reports in your account

The key difference is that the order is created inside your software from the first word, not retold. This removes double entry, address errors and minutes of delay.

What else reduces operator load

The distributed call center answers the question «who picks up». At the same time, it is worth cutting the number of calls that need a human at all.

1. LLM voice robot. This is not an IVR with «press 1» buttons. The robot answers the call, asks where from and where to, clarifies the entrance, time and preferences, understands natural speech, pauses and follow-up questions, and creates the order without an operator, around the clock. It offers a regular customer to repeat their previous address. Requests the robot cannot handle (an unusual route, a dispute) are transferred to a person, either your own or from the distributed call center. Only a handful of taxi software products have an LLM robot; most competitors have none at all, or offer a classic voice menu.
2. Auto-notification. A call saying «a blue Opel, number 323, is on its way» and an SMS on arrival remove repeat «where is my car» calls. In services’ experience, this is one of the most common reasons for incoming calls.
3. App and bots in Telegram and Max. Younger customers prefer not to call. The bot takes an order in three taps (geolocation or an address with suggestions, car class, and options such as a child seat) and shows statuses «driver found / car arrived / on the way» with the driver’s car make, plate number and phone. The bot is branded for the service, and sending promotions and promo codes through it is free, unlike SMS. The shared «UpTaxi Taxi» app is included in the plan; a branded app costs 20,000 ₽ per OS.
4. Announcing queue position. If a queue does form, the customer hears «you are third in line» and hangs up less often.
5. Automatic order intake from landline numbers. The software detects the number, recognizes the customer and fills in their details.

A service that uses all of the above usually finds that a live operator is needed only for a minority of requests: routine «home to work» orders go to the robot, bots and app. For such a service, buying handling of the remaining «live» calls from the distributed call center costs a few thousand rubles a month.

Robot and call center: how they split the flow

It helps to think of call handling as three lines:

Line	Who answers	What requests	Cost to the service
First	LLM voice robot	Routine orders with clear addresses, repeat orders from regular customers	Included in the plan
Second	Your own operators (if any)	Complex routes, corporate customers, complaints	Salary
Third	Distributed call center	Overflow, nights, weekends, everything if you have no operators	1 ₽ + 4 ₽ per order

The better the first line works, the lower the load and costs on the second and third. The robot does not get tired, does not get sick and is just as polite at three in the morning.

How to connect: step by step

1. Assess your call flow. In the manager’s account, look at calls by hour and the share of missed calls. This determines the scenario: everything, nights or overflow.
2. Choose a call center. There are several call centers in the system; UpTaxi support will tell you who handles calls 24/7 and who has experience with cities of your size.
3. Set the rules. Wait time before a call goes to the queue, the hours when calls go outside, priorities.
4. Check the scripts. Give the call center your service’s specifics: district names, how to explain the price, what to say about arrival time.
5. Listen to recordings. During the first week, spot-check calls; recordings are available in your account. If you are not satisfied, switch call centers.

If you want the opposite, to take other services’ calls, contact support: you will be connected to the distribution system as a call center, and calls from other services will start arriving in your console.

Summary

A distributed call center solves the old dilemma of a small taxi service: pay for idle operators or lose calls. Order intake for 1 ₽ plus 4 ₽ per completed order instead of 25–60 ₽ through an in-house operator, while the order is created in your software and goes straight to your drivers. For a large service, the same mechanism becomes extra income.

It is one of the modules included in the UpTaxi base package, along with the LLM voice robot, order bots in Telegram and Max, the customer app, road-based automatic dispatch and the driver exchange. Customers order however they like: by phone, app, website, or bot in Telegram or Max. A server for your company is set up within a day, and 24/7 support is free.

Get 30 days free · Message on Telegram · Message on Max · Call +7 900 28-28-328

The full feature list is on the about the program page.