Customers
How many pay each month, and how many join and leave.
- Flows
- Gross adds, churn, net adds
- Split
- Postpaid and prepaid
Telecom case interview, subscribers, churn and the network bill
A telecom case is usually about a carrier whose revenue is a monthly subscription and whose biggest costs are a network it has already built. Work it per customer: revenue per user each month, how long the customer stays before churning, and what it cost to win them, against the capital it takes to serve them. Most prompts, from a price war to a fiber build, come down to how long a customer stays and what they are worth over that time. Soreno, an AI voice interview practice app for consulting, investment banking and other finance candidates, has a free pricing case you can run with no sign-up, good practice for the price-war questions telecom cases turn on, plus full mock cases with scored feedback.
Updated
- Unit of account
- One customer, one month
- ARPU, churn and margin are all monthly
- Average customer life
- About 1 ÷ monthly churn
- At 1% a month, about 100 months (a simplification)
- T-Mobile postpaid phone churn, 2025
- 0.93% a month
- Up from 0.86% in 2024, per its 10‑K
- Worked case
- Price cut loses $72 million
- An invented carrier, every number shown
How is a telecom case built?
A carrier spends heavily up front on spectrum, towers and cable, then earns a small monthly fee from each customer for years. Serving one more customer on a network that is already built costs little, but winning that customer and keeping them costs a lot. So a telecom case is worked per customer and per month, and the first question is usually whether the problem sits in how many customers there are, what each pays, how long they stay or what the network costs.
Revenue per customer
What each customer or account pays a month.
- Measure
- ARPU, or ARPA per account
- Moves with
- Plans, promotions, add-ons
Cost to win and keep
Acquisition cost against lifetime value.
- Win
- Marketing, commissions, devices
- Keep
- Retention offers, service
The network
Mostly fixed, mostly paid for in advance.
- Measure
- Capex ÷ revenue
- For a build
- Homes passed, take rate
Worked example: Tamarack Wireless (ours)7 steps
Tamarack Wireless, an invented regional carrier with 2 million phone customers, has seen monthly churn rise from 1.0% to 1.5% since a national rival cut its prices. The sales team wants to cut every plan by $5 a month. Should Tamarack do it? Tamarack and every figure in its case are made up by Soreno for practice. To keep it short, yearly figures assume Tamarack replaces every customer who leaves, so its base stays at 2 million.
Clarify
Monthly churn, postpaid phones, one rival
Confirm that churn is monthly (T-Mobile, for one, reports it per month) and covers postpaid phone customers. Each pays $50 a month, Tamarack keeps 60% of that after running the network, customer care and billing, and it costs $400 to win a new customer. The goal is to protect profit, not to win back every customer.
Size the churn
240,000 to 360,000 customers leaving a year
At 1.0% a month, 240,000 customers left a year. At 1.5%, 360,000 do, 120,000 more. The average customer used to stay about 100 months. Now it is about 67.
Lifetime value
$3,000 to $2,000 a customer
Each customer brings $30 of margin a month. Over 100 months that was $3,000, and over 67 months it is $2,000, before any discounting. A customer is still worth far more than the $400 it costs to win one, so retention is worth paying for, if it is aimed well.
Price the problem
$48 million a year to stand still
Replacing the 120,000 extra leavers at $400 each costs $48 million a year, against $720 million of yearly margin from the base. That is the size of the problem any fix has to beat.
Option A
$5 off for everyone: minus $72 million a year
Cutting every plan by $5 costs 2 million × $5 × 12 = $120 million of margin a year, since serving a customer costs the same at either price. Even if churn fell all the way back to 1.0%, it would save only the $48 million of extra acquisition. Net, minus $72 million a year. Most of the discount goes to customers who were never going to leave.
Option B
Credits for the at-risk group pay if churn falls below 1.125%
Offer $10 a month for 12 months only to the 300,000 customers the data flags as likely to leave: $36 million. It pays for itself if it saves 90,000 leavers a year, which is overall monthly churn down 0.375 points, from 1.5% to below 1.125%. Put another way, 30% of the flagged group must stay who would otherwise have gone.
Recommend
Target retention, don't cut prices across the board
Reject the across-the-board cut: it loses $72 million a year even in the best case. Test the targeted credit in one region first and keep it only if churn falls by more than 0.375 points. Each customer kept is also worth more than its $400 replacement cost if Tamarack can't win new customers as fast as it loses them, so the test is a cautious one. Then find out why customers leave: price, coverage, a bundle the rival offers with home internet, or a handset deal. Next step: churn by plan, region and contract status for the last six months.
A fiber build in one town (ours)4 steps
A second invented example: should a provider run fiber past the 12,000 homes of a small town? Every number is ours, computed in fiberFigures(). Simple payback, no discounting.
Build
$12 million to pass 12,000 homes
At $1,000 per home passed, the street-level network costs $12 million whether anyone signs up or not. Each home that subscribes costs another $700 to connect.
Take rate
40% sign up: 4,800 customers
At 40% of homes passed, connecting costs $3.36 million, for $15.36 million in all. At $70 a month and a 50% margin, each customer brings $420 a year, or $2.02 million for the town.
Payback
About 7.6 years
$15.36 million ÷ $2.02 million a year = 7.6 years, once the take rate is reached. Reaching it takes years, so the real payback is longer.
Break-even take
28.6% for a 10-year payback
The build pays back in 10 years if about 3,429 homes sign up, 28.6% of those passed. Ask who else serves the town, including cable and fixed wireless, before you trust any take rate.
Telecom vocabulary you will use8 terms
Where a carrier's 10-K defines a term, its definition is summarized. The rest are our plain-words versions.
Postpaid, prepaid
Billed after, or paid before
Postpaid customers are billed monthly, usually on accounts with several lines. Prepaid customers pay in advance and leave far more often: T-Mobile's 2025 prepaid churn was 2.72% a month against 0.93% for postpaid phones.
ARPU
Service revenue ÷ customers ÷ months
T-Mobile's 10-K defines ARPU as the average monthly service revenue earned per customer. It is built on service revenue, so equipment sales are left out.
ARPA
The same, per account
T-Mobile reports postpaid ARPA, the average monthly postpaid service revenue per account. An account can hold several lines, so ARPA and ARPU can move differently.
Churn
Customers lost ÷ average customers, per month
T-Mobile defines churn as customers whose service was deactivated, as a share of average customers, per month. Verizon counts connections instead: retail disconnects divided by average retail connections over the period. The FCC's 2024 report adds that higher churn means more spending on acquisition just to keep the same number of customers.
Acquisition cost
What one new customer costs to win
Marketing, sales commissions and device promotions. Compare it with lifetime value before approving any promotion.
Lifetime value
Monthly margin × average months stayed
With a steady monthly churn rate, the average stay is about one divided by that rate. It is a rough guide, so discount it in a long-horizon case.
Capital intensity
Capex ÷ revenue
Verizon reported $17.01 billion of capital expenditures in 2025 on $138.19 billion of revenue, about 12.3%, our division. Network cases turn on whether new spending earns more than it costs.
Homes passed, take rate
Reach, then sign-ups
Homes passed are those the network runs past and could connect. Take rate is the share that subscribes. The build cost follows homes passed, while revenue follows the take rate.
Reference points from the FCC and carrier filingsOfficial data
Numbers to sanity-check a case. The FCC figures come from its 2024 Communications Marketplace Report, released December 31, 2024, and describe the market at the end of 2023. The FCC's reports page lists editions for 2018, 2020, 2022 and 2024, one every two years, so check for a 2026 edition before you quote these after the end of 2026. The carrier figures are from 10-Ks for 2025, filed in February 2026.
Market shape
3 nationwide carriers
The FCC's report counts three facilities-based nationwide providers as of December 31, 2023, AT&T, T-Mobile and Verizon, with about 371 million postpaid and prepaid connections between them. At that date it also counted a new facilities-based entrant, regional carriers and resellers that buy wholesale access (MVNOs).
Connections
About 492 million mobile connections
Numbering data cited in the same report show about 492 million mobile wireless connections at the end of 2023. The count covers handsets and other connected devices with a phone number, so size a phone market on people, not connections.
Towers
Mostly owned by tower companies
The FCC's report says most towers are now owned or operated by independent tower companies, after the major carriers sold theirs. A carrier leases space on them, which is why tower sales show up as cases.
Home internet over wireless
T-Mobile about 8.45 million, Verizon 5.7 million
T-Mobile's 10-K counts 7,602,000 5G broadband customers among its postpaid customers and 848,000 among its prepaid ones at the end of 2025, about 8.45 million in all, for a fixed wireless product. Verizon's counts 5.7 million fixed wireless access connections. Any fiber or cable case now has a wireless competitor in it.
Six telecom prompts and where to start6 prompts
Our openings. Each picks the customer or network number to build first.
Churn is up
Who is leaving, and to whom
Split churn by plan, region, tenure and whether customers are in contract, then find where they go. A rival's price, a coverage gap and an expiring device deal call for different fixes.
Price war
Margin given up against customers kept
A cut is paid on every customer and saves only the few who would have left. Compare the two, as in the Tamarack case, before matching a rival's price.
Network build
Cost per home passed against take rate
Build cost, connect cost, take rate and margin per customer give payback. Then check the competitors already serving those homes.
Bundling
Lower churn against a lower price
Mobile plus home internet often keeps customers longer. Size the discount against the churn it removes and the customers it wins.
Tower or asset sale
Cash today against rent for years
Selling towers brings in cash and adds a lease payment. Compare the price with the value of the rent and the control given up.
Merger
Network overlap and spectrum
Where the networks overlap, which sites and costs combine, what spectrum the deal adds and what regulators will ask, then integration cost.
Two openings to the Tamarack case (ours)2 openings
A weak start and a strong one, both written by us.
Weak opening
Agreeing with the sales team
"Churn is up because the rival is cheaper, so matching their price should fix it. I'd look at how much revenue we lose versus how many customers we keep." Half right, but it assumes price is the cause and never asks how many of the 2 million customers were ever at risk.
Strong opening
Per customer, then who is at risk
"A price cut is paid by all 2 million customers and only helps with the 30,000 a month who leave now, 10,000 more than before. I'd like to compare the margin we'd give up with the cost of the extra churn, then see whether we can find the customers at risk and act on them alone. I'd also want to know why they leave, since price may not be the only reason."
What TMT firms say, and what they publish2 firms
From the pages we read on October 31, 2026.
Altman Solon
A TMT-only strategy firm
Altman Solon describes itself as the largest global strategy consulting firm working only in telecommunications, media and technology. Its undated interview guide says a case lasts 30 to 45 minutes, can be interviewer or interviewee led, and suggests practicing cases and researching the business models of firms in TMT.
Altman Solon: Interview guideMcKinsey
No telecom case among its five
McKinsey's interviewing page lists five practice cases: Beautify, Diconsa, Electro-Light, a national education system and Talbot Trucks. None is set at a carrier, so practice telecom cases from guides like this one.
McKinsey: Interviewing
From FCC: 2024 Communications Marketplace Report (December 31, 2024), FCC: Communications Marketplace Reports, T-Mobile US: Form 10-K for 2025, Verizon: Form 10-K for 2025, Altman Solon: Interview process, Altman Solon: Interview guide (undated) and McKinsey: Interviewing (practice cases).
What signals does a telecom interviewer want?
Altman Solon's interview guide says it assesses how logically you break a problem down, how you apply quantitative and qualitative analysis, and how you adjust your thinking to new information. On a telecom case, in our experience, these moves show it.
Per customer, per month
Turning totals into ARPU, churn and margin per customer early, so options can be compared on the same footing.
Lifetime, not one month
Valuing a customer over how long they stay, and comparing that with what it costs to win one.
Aim, don't spray
Seeing that a price change hits every customer while churn comes from a few, and looking for ways to reach the few.
Fixed network, small extra cost
Knowing that serving one more customer on a built network costs little, and that a new build lives or dies on take rate.
The competitor's next move
Asking what a rival, a cable company or a fixed wireless provider will do in response before calling an answer final.
9 telecom questions for practice
All are set at a carrier, a broadband provider or a media subscription. After each is how we would want the answer to begin.
Customers and pricing
Customers are leaving or prices are under pressure.
A regional wireless carrier's churn has risen by half in a year. What would you look at?
Splits churn by plan, region and contract status, finds where the customers go, and compares the cost of keeping them with what they are worth.
A national rival has cut its unlimited plan by $10 a month. Should our client match it?
Compares the margin given up across the whole base with the customers a match would keep, and looks for a targeted response first.
A streaming service's monthly churn has doubled in a year. How would you find out why?
Checks whether new or long-time subscribers are leaving, what changed in price or content, and whether churn is voluntary or failed payments. Practice it, see a model answer
Should a carrier offer a discount to customers who take both mobile and home internet?
Sizes the discount on customers who would have bought both anyway against the churn it removes and the new customers it wins.
Network and deals
Decisions about what to build, sell or buy.
A cable company is deciding whether to build fiber in a town it doesn't serve. Is it worth it?
Works out cost per home passed, the take rate needed to pay back, and how the incumbent and fixed wireless would respond.
A wireless carrier has been offered a high price for its towers. Should it sell?
Compares the cash today with the rent it would pay for years, and asks what control over its network it gives up.
Two regional carriers want to merge. Where would the savings come from?
Looks at overlapping sites and network costs, combined spectrum, overhead and the integration cost, and names what regulators would question.
Sizing
Estimates in phones and connections.
How many smartphones are sold in the United States each year?
Builds from people who own a smartphone and how often they replace it, then checks against the number of phone connections. Practice it, see a model answer
How many homes in a mid-sized U.S. city would sign up for a new fiber service in its first two years?
Starts from households, removes those the network won't pass, then applies a take rate that rises over time and checks it against the other providers.
Telecom errors we hear often
Each of these turns up regularly in practice telecom cases.
Reading monthly churn as yearly
1.5% a month is about 18% of customers a year. Ask which it is before you multiply.
Cutting price for everyone to stop a few
The discount is paid on the whole base. Compare its cost with the customers it actually keeps.
Leaving out what a customer costs to win
Promotions, commissions and device deals can take years of margin to earn back. A customer who leaves early can lose money.
Treating homes passed as customers
A network that runs past a home earns nothing until the home subscribes. Take rate is the number that decides a build.
Mixing ARPU and ARPA, or postpaid and prepaid
Per line and per account give different numbers, and prepaid customers churn far more. Keep the groups apart.
Ignoring capital spending
A plan that adds customers but needs new network spending has to pay for that too. Check capex before you call it profitable.
Do the per-customer math out loud
Soreno's AI interviewer runs the whole case with you. It releases churn, pricing and network numbers when you ask, pushes back on your per-customer math and gives you a scored report.
Questions people ask
What is a telecom case interview?
A case set at a wireless carrier, a cable or fiber provider, a tower company or a media subscription business. Common prompts are rising churn, a price war, a network build, bundling, a tower sale and a merger.
What metrics should I know for a telecom case?
Customers and net adds, ARPU and ARPA, monthly churn, acquisition cost, lifetime value, capital intensity, and for a build, homes passed, take rate and payback.
How do you calculate customer lifetime value in telecom?
Multiply monthly margin per customer by the average number of months a customer stays, which is roughly one divided by monthly churn. In our worked case, $30 of margin a month at 1.5% churn gives about $2,000, before discounting.
What is a normal churn rate for a wireless carrier?
Use the interviewer's number. As a reference, T-Mobile's 10-K for 2025 reports postpaid phone churn of 0.93% a month and prepaid churn of 2.72% a month. Verizon's 10-K puts its Consumer segment's retail postpaid phone churn at 0.92% for 2025, up from 0.83%, close to T-Mobile's. Prepaid customers leave far more often.
How much do carriers spend on their networks?
A lot, every year. Verizon's 10-K reports $17.01 billion of capital expenditures in 2025, about 12.3% of its $138.19 billion of revenue by our division.
Do consulting firms publish telecom practice cases?
Not on the two firm sites we checked on October 31, 2026. McKinsey's five practice cases include none set at a carrier. Altman Solon, which works only in telecom, media and technology, publishes an interview guide that encourages candidates to research the business models of TMT firms.
Do I need to know the telecom industry to do well?
No. The per-customer measures above and the fact that the network is mostly a fixed cost carry most cases. When a figure is missing, ask the interviewer for it, or estimate it and say how.