Internet provider: what one added subscriber really costs

“The network kept growing, the report kept showing a profit, and what was left in the bank each month was so little that I was afraid to buy a second switch. I thought I was bad at collecting payments. It turned out I was good at connecting people.”

Vadym Oliinyk runs a residential network in Chernivtsi. Eleven years in telecom, seven of them as the owner. Fourteen people: installers, support, two in the NOC. There are 4 200 subscribers, and a few more every year.

The report for the year showed 4 032 000 ₴ of profit. Of that, 264 000 ₴ was left in cash. The rest went on things that did not look like a cost in the report at all.

What the report says

The year looked calm and clear.

Twenty-eight per cent for a residential network is good. Vadym looked at that line and could not understand why there was never money for development. The bookkeeper showed him the report, the bank showed him the balance, and the two never agreed.

Every time a batch of switches had to be ordered or a new district opened up, the money came together with difficulty. Vadym took a short loan, closed it within a quarter and went back to waiting. He took that for a feature of telecom: the trade is capital-heavy, everyone lives like that.

One subscriber for one month

It all starts with one flat.

The tariff is 280 ₴. Transit and peering take 28 ₴. The payment systems take five per cent, which is 14 ₴. Variable costs come to 42 ₴ in all.

That leaves 238 ₴.

This is the figure Vadym carried in his head for seven years. It is correct and it hides nothing. It simply answers a different question from the one he was asking.

The question “how much do I earn on a subscriber” has two different meanings, and they diverge by an order of magnitude. The first: what is left of the payment after everything that travels with the traffic. The second: what is left after everything that keeps the network alive. Between them sit the whole support payroll and all the electricity at the nodes.

“Two hundred and thirty-eight hryvnias from a flat. I knew that figure by heart and could have said it at three in the morning. It turned out it is not profit, it is halfway to it.”

Two contributions, not one

Somebody pays the fixed costs too.

Spread 7 963 200 ₴ over 4 200 subscribers and twelve months and you get 158 ₴ per subscriber per month. The contribution then looks different.

Why the gap is so wide

Eighty hryvnias a month is the sum the network actually lives on. Switches are bought out of it, a new district is paid for out of it, the next subscriber's connection is taken out of it.

Support, the NOC, the vans, the rent, the licences, the electricity at the nodes — none of it disappears when a subscriber leaves. It stands where it is and waits for the next one. Per flat that is two thirds of the contribution.

The difference between 238 ₴ and 80 ₴ is not somebody's mistake. It is that the first figure describes traffic and the second describes a business.

What a connection costs

A connection is not a cost of the month. It is capital that goes into the wall and stays there as long as the subscriber pays.

• cable from the node to the flat, with fittings — 740 ₴
• the router handed to the subscriber — 1 450 ₴
• the work of the installation crew — 560 ₴
• the node equipment's share for one port — 390 ₴

That is 3 140 ₴ for one flat.

The difference between a cost and capital is not theoretical here. A cost spoils the month and is forgotten. Capital spoils the year and is never forgotten, because it lies in the wall exactly as long as the subscriber pays, and it vanishes together with him.

What comes back out of it

The cable never comes back: it was pulled for one particular flat. The port on the node is freed and waits for the next person. The router comes back if the subscriber brings it in, and that gets a separate conversation below.

The same figure, two answers

Divide 3 140 ₴ by the gross contribution of 238 ₴ and you get 13.19 months. A little over a year. It sounds excellent, and that is exactly how Vadym counted for all seven years.

Divide the same 3 140 ₴ by the net contribution of 80 ₴ and you get 39.25 months. More than three years.

Payback depends not on the subscriber but on what you agree not to count.

The gap between those two terms is 2.98 times. Almost three, though three is never crossed. And the whole gap rests on one item that is missing from the cost of a single connection, because it looks like something shared by everyone.

What was tried before

The first attempt was about money coming in, not money going out.

Vadym raised the tariff by 20 ₴. On paper that is 4 200 subscribers times 20 ₴ times twelve months — 1 008 000 ₴. A million from one decision that takes five minutes and does not need a kopeck of investment.

“I put it up by twenty hryvnias and waited for a million. A quarter later I looked for it and could not find it. Revenue really did grow, and there was less money than before.”

Where the million went

Churn grew. It had been 900 subscribers a year; it became 1 150. An extra 250 flats.

Every one of them has to be replaced, because the network has to stay full. Replacing 250 connections is 250 times 3 140 ₴, that is 785 000 ₴.

Out of the million, 223 000 ₴ was left.

The ones who left were not the ones Vadym expected. The quietest ones left:

• those on a cheap old tariff who never thought about the network at all
• those who already had one unclosed ticket and took the new price as the last straw
• those whose neighbour happened to mention a competitor's promotion that very month
• those who paid once a year and saw the whole rise in one sum

Why this is the hardest thing to catch

Vadym held the new price for two years. In that time the churn turned into a habit: on the installers' schedule a disconnection and an installation stood side by side, and nobody asked any more why there were so many of them.

It is 22.1% of what was promised — less than a quarter. But it is not a minus. The tariff rise did not create a loss, and the report for that year looked even better than the one before.

Decisions that come out negative are visible at once. Decisions that deliver a fifth of what they promised live for years, because formally they work.

Three in four

Now the main figure of this whole story.

Over the year the network connected 1 200 flats. Over the same year 900 left. The growth was 300 subscribers.

What that means in money

Connections took 1 200 times 3 140 ₴, that is 3 768 000 ₴. The base grew by 300 subscribers. Divide one by the other — 12 560 ₴ per added subscriber.

A connected subscriber costs 3 140 ₴. An added one costs 12 560 ₴. Exactly four times more, because there were exactly four times more connections than growth.

Three connections out of every four add nobody — they replace somebody who left.

That figure does not depend on how much you connect. It depends on the ratio. Had the network connected two thousand flats instead of a thousand two hundred, the growth would have been bigger. But the price of one added subscriber would have fallen only partly. Every new connection would still cost the full 3 140 ₴, and three quarters of them would still be closing a hole.

“The installers worked as never before. A hundred connections a month, the schedule packed to the edge. And over the year the network grew by three hundred flats, and I could not work out where the other nine hundred had gone.”

Where that money is in the report

It is not there at all.

The 3 768 000 ₴ spent on connections is 93.5% of the profit for the year. The money left the till, but it cannot be seen in the cost lines. The cable and the router became fixed assets. The installation smeared itself across the crew's wages. Nobody counts the port on the node separately at all.

The report showed 4 032 000 ₴. The till showed 264 000 ₴. That is 1.87% of the revenue for the year — exactly the sum that makes buying a switch frightening.

Formally everything is right. Capital investment should not stand in the cost lines, or profit would tell a lie in the other direction. The problem is not the bookkeeping, it is that the owner makes his decisions by the profit report and pays for them out of the till. Two documents describe one year, and 3 768 000 ₴ sits between them.

The routers that never came back

Here the story goes completely quiet.

There is one more item that nobody sees.

Of the 900 subscribers who left, sixty per cent return the router. That is 540 devices. The other 360 stay in the flats along with a promise to bring them in some day.

Multiply 360 by 1 450 ₴ and that is 522 000 ₴ a year, 12.9% of the profit.

“I once started a table of how many subscribers we had given a router to and how many routers came back. I looked at it once. I did not want to look again.”

Why this is invisible

Because a lost router has no document. The subscriber did not steal it — he simply did not return it. Nobody writes it off, because the device formally exists somewhere. Three years later such routers quietly leave the books as worn out, and their cost mixes in with normal depreciation.

Why growth does not help

The owner's logic is simple: if there is little profit, connect more.

It is not even wrong. Every new subscriber really does bring a contribution, and the network really does get bigger. What is wrong is the assumption that new subscribers are added to the old ones rather than replacing them.

But every connection adds 3 140 ₴ of capital today and 80 ₴ of contribution each month. For those 3 140 ₴ to come back, the subscriber has to stay almost forty months. And while he pays, his neighbour has already gone, and his place needs another 3 140 ₴.

Growth on high churn is not growth, it is expensive parking.

The nastiest part is that every working indicator looks excellent while this goes on. The installers are loaded. Revenue is growing. The subscriber count is growing too, slowly as it is. Not a single figure flashes red, because the one that would flash is counted nowhere: how much capital goes into growth and how much goes into holding the size.

“I thought I was building a network. In fact I spent three years repairing the same network at the price of a new one.”

The move that works

If growth is expensive because of churn, then churn is also what makes it cheaper.

Vadym did not start connecting more. He took 480 000 ₴ for the year and spent it on what he used to think optional:

• a call to the subscriber after the second fault, not after the notice to terminate
• a speed upgrade on the same tariff for everyone who has been with the network over two years
• replacing old routers before they start falling over
• a separate visit schedule for the buildings with the worst ticket history

What exactly was changed

No new tariff. No promotion. No advertising. Only the order of actions in situations that used to be handled with whatever time was left over, because a new flat always seemed more important than an old one.

Not one of these decisions needed anybody's agreement from outside. All of them had been in the owner's hands the whole time, and not one of them had ever looked important enough to make the plan for the year.

Churn fell from 900 to 600 a year.

What it brought

The arithmetic here is short.

Saved: 300 times 3 140 ₴, that is 942 000 ₴. It cost 480 000 ₴. The net effect is 462 000 ₴, or 11.5% of the profit.

Against the 223 000 ₴ from the tariff rise that is 2.07 times more. More than twice as good, and not one subscriber paid a single hryvnia extra.

There is a second part, hard to count and easy to see. When churn falls, the installation crew stops driving to the same buildings twice: first to take the kit out, then to put it back. The schedule frees itself up, with nobody hired.

“The strangest thing is that the subscribers would not even call it a promotion. Things simply stopped breaking, and we were the ones calling them, not the other way round.”

What retention does not do

It does not remove churn. Six hundred subscribers leave anyway — people move, buildings are cleared, somebody simply chooses differently.

It does not add subscribers by itself. The growth stayed the same: 300 for the year.

Then what is the point

The point is that the same growth now costs 2 826 000 ₴ instead of 3 768 000 ₴. The difference is not in the number of subscribers, it is in the price of each added one.

And in the fact that the installation crew is working on growth for the first time in seven years, instead of on replacement.

For comparison. To take the same 462 000 ₴ out of the tariff, it would have to be raised so far that churn would grow harder still. Next year an added subscriber would cost even more. Retention works the other way: every year of low churn makes the next one cheaper.

What to watch every month

Three figures are enough, and none of them is revenue.

Revenue in a residential network barely moves. It grows exactly as much as the base grew, which is why watching it is pointless: it will confirm any decision you like.

• connections for the month and disconnections for the month — as separate lines, never as a difference
• the capital spent on connections during the month — in live money, not in depreciation
• the routers that did not come back — in units and in hryvnias

If disconnections run at more than a third of connections, you are paying to stand still. Divide churn by the subscriber count and you get a percentage that says nothing. Divide capital by growth and you get the price you are really paying for every new subscriber.

These three lines do not need a new accounting system. Connections and disconnections are already in the billing. The capital on installation is already in the delivery notes and the time sheet. The routers are already in the store. All that is missing is to bring them together in one place and divide one by the other once a month.

The money does not vanish — you simply do not see it

Telecom looks like a business about traffic and equipment. In fact it is about how long the subscriber stays in the wall you put 3 140 ₴ into.

Vadym did not have a problem collecting payments. He had a network that spent almost all of its profit every year on staying the same size.

The profit showed 4 032 000 ₴. The growth turned out to be 300 subscribers at 12 560 ₴ each. Both figures are true, and between them lies the whole difference between the report and the till.

How to see this in your own network

All the arithmetic above rests on one row the profit report does not have: how much capital went into connections this month and how many subscribers were actually added. While connections and disconnections are collapsed into a difference, neither the cost of an added subscriber nor the fact that three installs in four close a hole is visible. When the same data is collected automatically, the price of growth is visible every month — which is exactly why owners of small and medium businesses use Finmap: to see the financial picture rather than piece it together from memory.

Read also

• MRR, ARR and churn in real money: what the growth dashboard hides
• Three pickup points: one eats what the other two earn

Topic foundation: Management Accounting Explained: What It Is and Why Owners Need It