Beating the fallacy of cheap marketing

The budget review is going well until someone asks why the email program sends so much. The answer arrives quickly and with confidence, the usual defense of cheap marketing: it is basically free. The list already exists, the template is built, and each additional send costs a fraction of a cent. Why would we send less of something that costs nothing?

The same logic shows up in the push notification that fires after every app session, the pop-up survey on every page, and the quick two-minute questionnaire that follows every support call. Each one is cheap by the only measure anyone in the room is counting.

The fallacy is counting only the outbound cost. A channel is priced by what it costs to send, not by what it costs the relationship. The send is nearly free. The unsubscribe is not. The muted notification is not. The customer who has learned that your surveys are noise is very much not, and you will find out what that costs on the day you need an honest answer from them.

What a cheap send actually costs

A blast email costs almost nothing per send and something real per unsubscribe. The unsubscribe is close to permanent, and the person who clicked it was, until that moment, someone who had agreed to hear from you. That permission was hard to get (getting it at all is the subject of why can’t we collect email addresses?), and you spent it on a message that was not worth their time.

A generic push notification is free to fire and expensive in attention. Attention is the scarcest thing a customer gives you, and the notification permission is one tap from gone. Once it is off, the message that matters (the delivery is late, the payment failed, the appointment moved) has no way in.

A “quick” survey sent to everyone after every interaction is the cheapest research there is and the fastest way to teach customers to ignore you. Response rates fall and survey fatigue sets in. The people who still answer are the furious and the delighted, and the middle, where most of your customers live and most of your churn starts, goes quiet. A cheap survey becomes expensive at exactly that point: it has burned the response rate you will need later, when you want to know something that matters, and it has made the data you still collect less representative every month.

Notice the pattern. Every one of these is a withdrawal from an account you did not know you were keeping.

A cost per annoyed customer

Here is a habit for any team that sends things at scale. Before a mass send, estimate the cost per annoyed customer, in two steps.

First, guess who will be irritated. Not the average customer: the specific slices. The person who bought this product last week and is being offered it again. The customer who has answered three surveys this month. The user who opened the app twice today and is about to get a third notification. You will not know the numbers exactly, and you do not need to. A rough share of the list is enough.

Second, put a value on the irritation. An unsubscribe is worth roughly the future margin of an email-reachable customer, minus what you would have earned from them anyway. A muted notification is worth whatever you would pay to reach that person with an urgent message later. A burned survey respondent is worth the research you can no longer do. These estimates will be wrong, but they will be far less wrong than zero, which is the number currently in the spreadsheet.

Now compare. A send that earns a little from many and annoys a few is fine. A send that earns a little from many and annoys many is a loss dressed up as activity. Teams that do this exercise honestly tend to find that a good part of their calendar falls into the second group, and that cutting it changes little about revenue and quite a lot about opt-outs and response rates.

This is also the only way the measurement question gets a fair hearing. If you are asking whether your program is worth the effort, you cannot answer it with the send cost on one side and revenue on the other. The relationship cost belongs in the ledger too. It is the missing line in most attempts to drive profit from a customer program, and once it is there, a surprising number of “profitable” sends stop being profitable.

Three questions before any mass send

1. Who did not ask for this? List the segments on the send who gave no sign of wanting it: recent buyers of the same thing, recent survey respondents, people who have not opened anything in a year. Suppress them, or write down why not.

2. What does the annoyed customer cost? Put a number on the unsubscribe, the muted notification, or the lost respondent, multiply by your best guess at how many there will be, and write it next to the expected revenue. If the two are within shouting distance, do not send.

3. What will this cost us next time? Every send draws down permission and attention. Ask what you will need from this list in three months (a launch, a renewal push, an honest survey about a real problem) and whether today’s send leaves enough in the account to do it.

The survey version of the third question is the one to start with this week. Pull up everyone you surveyed in the last thirty days and ask how many of them you would want to survey again next month about something that matters. If the answer is all of them, again, you have found the cheapest research in the company, and the reason it will not be worth much for long.