The channel that built your business was an arbitrage.
It worked because you got there before the cost of being there caught up with the value of being there. Cheap search terms nobody had bid on. A platform where organic reach was still free. A directory before everyone listed. A referral network nobody else had bothered to build.
Arbitrages close. That is not a flaw in the channel and it is not something you did wrong. It is the schedule.
Which reframes the panic. Your channel is not breaking, it is repricing, and the fixes everybody reaches for first, better creative and tighter targeting, are efficiency gains in a market where your competitors are making the same gains. They buy a quarter. Sometimes two.
The question worth asking is harder and more useful: during the years it worked, what did you convert the cheap customers into? Because that, not the channel, is what you still own.
The three things “it stopped working” can mean
Same symptom, three causes, and they want opposite responses.
The platform changed the rules. An algorithm update, a policy change, an interface that now answers the question your page used to answer. This one is genuinely a marketing problem and it is sometimes recoverable. I went through the organic search version of it in recovering from an algorithm update, including why rewriting everything in week one is the worst available response, so I will not rebuild that here.
The channel got crowded. Your cost per customer rose because more people are competing for the same finite attention. Nothing about your business changed. Nothing about the platform changed either.
Your offer stopped being distinctive. The channel is working exactly as before and is now reporting, accurately, that you are one of six similar options. This is the one nobody wants, because the dashboard is in marketing and the problem is not.
Most of the energy in most companies goes into treating the second and third as though they were the first.
The test that separates them
One question, and it is answerable without seeing anybody’s accounts.
Did your competitors’ costs rise too?
If the whole auction got more expensive, you have crowding. If your cost rose while the market’s did not, something about you got less compelling, which is a positioning problem wearing a performance costume.
You can get at this without their data. In Google Ads, impression share and auction insights show you who is newly present and how often they outrank you. Count the names that were not there eighteen months ago. On any platform, watch your click-through rate against your own history: crowding usually moves cost first and click-through later, while a weakening offer moves click-through first.
And look at the simplest thing, which is whether your conversion rate held. If people are still arriving and still converting, and only the price of arrival went up, that is a market telling you what the slot is now worth. If people arrive and no longer convert, the channel is innocent.
Nobody can tell you what it should cost
A short detour, because you will go looking for a benchmark and I want to save you the trip.
I went through the published 2026 acquisition cost benchmarks and they are not usable. The headline inflation figures range from about 60% over five years to 222% over eight to 260% over nine, depending on whose blog you open. Meta’s cost per customer is given as under $60 in one source and around $230 in another. Google Ads appears as $50 to $130 in one place and over $800 for B2B in another.
These are not measuring the same thing. Some count ad spend only, some count fully loaded cost including salaries, and almost all of them are published by companies selling something adjacent. One of the saturation datasets I found says openly that its figures are an assumption-based adjustment rather than measured data.
The only benchmark that means anything is your own channel’s cost over time, and the only comparison that means anything is whether the auction moved or only you did.
What the window was for
Here is the part I would actually want remembered.
A cheap channel is a temporary subsidy on customer acquisition. The businesses that survive its closure are the ones that spent the subsidy on something that outlasts it, and there are only about four such things.
An audience you own. A list, a subscriber base, a set of people who have given you permission to contact them without paying a platform for the privilege each time.
A name people search for. The ultimate expression of a closed arbitrage being survivable: when the generic term gets expensive, the branded term is still cheap, because you are the only bidder who should be winning it.
A product or delivery advantage. Something that makes you measurably better for a specific customer, which is what lets you outbid people for that customer’s attention.
Relationships that are not transactions. Partners, referrers, the people who send work without an auction attached.
Now the audit. Go back through the good years and ask which of those four you actually built. If the honest answer is that cheap customers produced revenue, and the revenue produced more spend on cheap customers, then the channel closing is not a setback, it is the bill.
That is uncomfortable and it is not a counsel of despair, because the four things above can be started late. They just cannot be started cheaply once the subsidy is gone.
The treadmill, and when it is right
The instinctive response is to go and find the next cheap channel. There usually is one, and for a while it works.
The problem is that this is a permanent condition rather than a strategy. You become a business whose core competence is finding arbitrage, which is a real competence and a precarious one, because each window is shorter than the last and you are always one closure from the same conversation.
In fairness, there are cases where chasing it is correct. If you are genuinely early in a market and need volume to learn, buy the cheap attention and worry later. If a new platform is structurally suited to your product rather than just underpriced, that is not arbitrage, that is fit.
The distinction is whether you would still want to be there after the price normalises. If the honest answer is no, you are renting, and you should be converting the rent into one of the four things while it is cheap.
The channel that still works
Which brings me to the useful version of all this, and it applies to whatever is currently carrying you.
Assume it has an expiry date. Write down roughly when you think it is, and what you are converting it into between now and then.
If you cannot answer the second part, you have a channel rather than a business, and the difference will only become visible on the quarter it stops.
Where I come out
When the numbers soften, spend a week on the test rather than a quarter on the fix. Find out whether the auction moved or you did, because the two have nothing to do with each other.
Then stop asking how to restore the channel and start asking what it was supposed to have bought you. If the answer is nothing durable, the fix is not a marketing fix, and the sooner that is said out loud the cheaper it is.
If you’d like a second opinion on whether your costs rose because the market moved or because your offer did, drop me a line on email, WhatsApp, phone or LinkedIn and we can have a quick chat, and my notes on performance marketing cover the measurement side. I’m taking on a small number of contracts at the moment, so I’ll tell you straight whether it’s something I could help with.