Home · Solutions · Customer service
Solution · Customer serviceThe phone rang while hands were on a customer. Two minutes later the caller has a text and the number waits in a queue
Missed calls called back
A robot watches the call history of your phone system or business phone. A missed number gets a text within two minutes: we saw your call, we will ring back by a set time, or book yourself if you prefer. The number lands in a callback queue in Teams and stays there until someone ticks the conversation off. Calling stays human; the robot only makes sure no number is ever lost.
Executive summary
In a small service business the phone rings while the paid work is happening: a massage, a haircut, a wheel on the lift. Missed calls pile up in the history, calling back happens in the evening from memory, and some numbers never get a response.
The robot sees missed calls in the phone system history and within two minutes texts the caller with a callback promise and a self-booking link; the number lands in a callback queue in Teams and waits to be ticked off.
Callers stop defecting to competitors for lack of an answer: they get a sign of life at once, a callback in the promised window, or they book themselves. The owner sees for the first time how many calls were being lost and how many were rescued.
call history from the phone system or the carrier; a text within two minutes; a callback queue and report in Microsoft Teams; bookings via Microsoft Bookings
Business problem
The customer does not offer a second chance; the customer waits fifteen minutes and moves on
In a one-person clinic and a three-person workshop the phone keeps the worst possible timetable: it rings while the work the customer pays for is happening. The massage therapist will not interrupt a treatment, the mechanic will not crawl out from under a car, the hairdresser will not put the scissors down mid-cut. The call rings out, drops into the history and starts waiting for the evening.
In the evening the call history looks different than it did at noon: some numbers are unknown, nobody remembers who called about what, and the energy after ten hours of work is what it is. Two or three numbers get a callback, the rest stays for tomorrow, and tomorrow brings new ones. That is how the quiet list grows: customers who wanted to pay and were never answered.
The caller meanwhile runs on a clock of their own. Their shoulder hurts, their pipe leaks, their wedding is on Saturday; they call three businesses in a row and stay with the one that picked up, or at least texted back. It is not about loyalty, it is about order: the first sign of life wins, and the loser never even knows a race was on.
The most bitter part happens after closing. A call at 7:40 pm gets no answer of any kind until morning, though the answer could be simple: we open tomorrow at 9:00, and you can book a slot right now with this link. No human needs to send that message; someone only needs to write it once.
How it works today
Below is what the work looks like before anything is automated.
- PersonThe phone rings during a treatment, a repair or a haircut
- WaitingThe call rings out and drops into the missed-call history
- Risk of errorThe caller dials the next number from the search results; someone there picks up
- WaitingCalling back starts in the evening, from memory and leftover energy
- Risk of errorUnknown numbers and unclear cases stay for tomorrow, and tomorrow brings new ones
- WaitingCalls after closing get no response of any kind until morning
Why the current process costs more than it appears
The bill that never shows up in a budget.
- Count last month’s missed calls in the phone system history; the number alone tends to surprise.
- How many of them got a callback the same day, and how many got no sign of life at all?
- What is one visit, one job, one repair worth, walking away with every lost number?
- How many of the owner’s evening minutes go into calling back from memory, not knowing who called about what?
Cost of inaction
The model assumes 80 missed calls a month, a third of which get no reaction, and some of those numbers are new customers with a concrete need. You will see your own number in the phone system history after one month.
The estimate leaves out reputation: the opinion “you can never reach them” can cost more than every line of this table combined.
A model organisation with realistic proportions – the numbers exist so you can run the same maths on your own data; they are not a client result.
A physiotherapy practice: two therapists, 45-minute treatments, a business line on a VoIP phone system, no reception desk, Microsoft 365. About 400 calls a month, 80 of which go unanswered.
Until now: evening callbacks from the call history, unknown numbers skipped, calls after 6 pm left without any response until morning.
Wednesday, 11:20 am: a patient with a painful shoulder calls for the third time. The therapist has his hands on another patient’s shoulder blade; the phone rings out.
At 11:22 the patient gets a text: “We saw your call. We will ring back by 1 pm, or you can book a slot right away: [link]”. The number lands in the Teams queue with the time of the call.
The patient clicks the link and books Friday 9:00 in Microsoft Bookings; the queue entry ticks itself off with the status “self-booked”. Had he not clicked, the therapist would see a reminder at 12:50 and call back between treatments.
In Friday’s report: 19 missed calls that week, 19 answered with a text within two minutes, 11 called back inside the promised window, 6 bookings from the link. In the modelled practice not one number was left without a reaction. These are model figures, not a client’s records.
Proposed solution
The source is the call history: from a VoIP phone system, the carrier’s panel or the business phone, via export or API. The robot checks it every two minutes and compares it with the list of answered calls; the difference is the missed calls that need a reaction. We record nothing and listen to nothing; the robot sees the number, the time and whether anyone has called back.
The reaction has two layers. The first is immediate: a text to the caller with wording you approve, a promise to call back within a concrete window and, if you want, a self-booking link to Microsoft Bookings. The second is internal: the number joins a callback queue in Teams and waits there with the time of the call until someone ticks the conversation off or the customer books themselves. After hours the text reads differently: it says when you open tomorrow and carries the same link.
A human always makes the call. The robot holds no conversations, impersonates no receptionist and promises nothing beyond the template; it only makes sure every number gets a sign of life within two minutes and that the callback queue has no bottom. Once a week a report shows how many calls were being lost, how many were rescued and at which hours the phone rings most; sometimes that report is a good argument for moving the lunch break.
UiPath Orchestrator: call-history polling every two minutes, the callback queue, retries and a trail of every contact; UiPath Integration Service connectors to Microsoft Teams and Outlook 365
A text to the caller within two minutes of a missed call, a callback queue with ticking and reminders, an after-hours message with a booking link and a weekly report of rescued contacts
A VoIP phone system or the carrier’s telephony via export or API; an SMS gateway; Microsoft Bookings for self-service appointments
How the automated process works
- AutomationEvery two minutes the robot compares the call history with the answered list
- SystemA missed number gets a text: we will call back by a set time, or book yourself
- AutomationThe number joins the callback queue in Teams with the time of the call
- PersonA human calls back between treatments; the conversation stays a conversation
- SystemA booking from the link ticks the entry off; no reaction raises a reminder
- AutomationA weekly report: missed, called back, rescued, peak hours
Human-in-the-loop model
Automation handles
- Detecting the missed call, the two-minute text and the entry in the callback queue
- Reminders about numbers without a reaction and ticking off bookings from the link
- The weekly report: how many were lost, how many rescued, when the phone rings most
People decide
- The phone call itself: a human calls back, not a machine
- The wording of the texts and the callback window promise; the robot writes only what you approve
- Decisions from the report: opening hours, breaks, a second line in season
Before and after
Systems and integrations
The stack is short on purpose: one engine, one execution layer, one place where a person decides.
Inputs
- call history from the phone system or the carrier
- text templates and callback windows
- the calendar of free slots in Bookings
- the team’s ticks on completed calls
Automation layer
- UiPath Orchestrator
- UiPath Robots
- UiPath Integration Service
- UiPath Action Center
Target systems
- texts to callers within two minutes
- a callback queue with reminders in Teams
- a weekly report of missed and rescued calls
Human touchpoints: calling back from the queue between treatments; ticking a conversation off with one tap; a weekly look at the report
Technologies used
call-history polling, the callback queue, retries, a trail of every contact
Athe queue, reminders and reports where the team already looks
Aself-service booking from the link in the text, on the calendar you already run
Acall history via export or API; no call recording
Bmessages to callers within two minutes; on your own provider contract
BIllustrative economic model
Numbers you can check against your own data.
The model assumes a fast reaction rescues part of the missed calls, not all; whoever truly had to, found another firm within fifteen minutes anyway. The volumes belong to the scenario; put your own into the calculator beside.
Run the maths on your data
An illustrative estimate based on your inputs. It models freed capacity, not promised savings.
Business benefits
- Every missed number gets a sign of life within two minutes, after hours too
- The callback queue replaces memory; no number is lost between treatments
- Some callers book themselves from the link before anyone finds a free moment
- The owner gets his evenings back, and customers stop saying “you can never reach them”
- The report shows peak calling hours; the roster can follow the real traffic
The management view
- Sales stop depending on whether hands happened to be free when the phone rang
- The number of lost calls becomes visible and measurable instead of being a hunch
- Peak hours from the report are hard data for decisions about the roster and a second line
Board-level KPIs
share of missed calls answered within 5 minutes · time to callback · bookings from the link after a missed call · contacts rescued after hours · missed calls with no reaction per week
Security and governance
The automation has exactly the permissions it needs. Not one more.
- The robot sees the number, the time and the call status; conversations are neither recorded nor listened to
- Phone numbers are processed solely for callbacks and bookings, in line with GDPR
- Texts go out from templates you approve; the robot holds no conversations and impersonates nobody
- Every text and every tick is logged: what, when, to which number
- Data stays in your Microsoft 365 tenant; the robots run in the EU region of UiPath Automation Cloud
Why now
A customer compares three businesses in fifteen minutes; the first sign of life wins, and silence gets no second chance.
The call history you already have contains a ready-made list of lost customers; it only needs to be read every two minutes.
A deployment of this class fits into weeks, and the first missed-call report tends to be the cheapest sales audit the firm has ever run.
Relevant executive roles
Sees on Friday how many calls were lost and how many rescued; evenings stop being callback time
Calls back from the queue between treatments instead of from memory after closing
Gets a text within two minutes and a callback at the promised time, or books with one tap
Common questions and objections
Most carriers expose the call history in a panel or via an API, and that is enough. If yours does not, a virtual phone system costs little per month and adds a greeting and opening hours along the way.
The text arrives two minutes after their own call, with the firm’s name and a concrete promise: we call back by 1 pm, or book here. It is the answer they were waiting for, not an advert. The wording is yours, and the booking link is optional.
A withheld number gets no text, because there is nowhere to send it; it only joins the queue as “missed without a number”. A wrong dial that gets a text simply will not click or call back; the entry expires from the queue after a set time.
When this is not the right solution
- A business with a reception desk that genuinely answers almost everything: check the history first, the problem may not exist
- Expecting a voicebot to hold the conversation and book the customer by voice: here a human calls, the robot keeps the queue
- Telephony with no access to the call history and no consent to share it: without the source there is nothing to watch
A question for the next management meeting
How many of yesterday’s missed calls have received any sign of life from us by today?
Implementation approach
Scope without ambiguity, before anything is signed.
We deliver
- Call-history watching and missed-call detection every two minutes
- Texts from your templates: during hours and after closing, with or without a Bookings link
- A callback queue in Teams with reminders and one-tap ticking
- A weekly report: missed, called back, rescued, peak hours
- Two weeks of tuning the wording and callback windows after launch
We need from you
- Access to the call history: the phone system panel, the carrier panel or an export
- The text wording and realistic callback windows the team will keep
- A decision whether callers should get a self-booking link
Stages
Discovery
A month of call history: how many missed, at which hours, what happened to them
Rules
Text wording, callback windows, after-hours behaviour, entry expiry
Build
The watcher, the texts, the Teams queue, the Bookings link, the report
Parallel run
Two weeks: the robot runs while you compare the queue with the old memory-based way
Go-live
Tuning wording and windows; a report review after the first month
A fast payoff. The one unknown is access to the call history at your carrier; where a panel or API exists, the rest runs on the Microsoft 365 you already have.
Wednesday, 11:20 am: a patient with a painful shoulder calls for the third time while the therapist’s hands are on another patient. At 11:22 the patient gets a text with a link. At 11:26 he has a Friday slot.
Pull last month’s call history from your panel and count the missed calls. Send us that one number; we will send back the arithmetic of how many of those calls a two-minute text would have rescued.
Count your missed callsThe neighbouring process usually has the same problem
Email, Messenger, phone and a form. Four places where the customer could ask the same question.
See the solution Operations & qualityVehicle check-in and repair status by textThe customer calls a third time asking about his car, and the mechanic crawls out from under the lift again.
See the solution Sales & marketingAvailability enquiries answered the same dayTuesday evening’s enquiry got its answer on Thursday. The guest already sleeps elsewhere.
See the solutionIndustries where we deploy this most oftenAesthetic medicineSmall business & services