Customer Journey

For a long time there was one way to send email: load the whole list and send everyone the same thing on the same day. So someone who signed up yesterday and someone who had been paying for a year got identical messages, and people who had already upgraded kept receiving pitches to upgrade.

Marketers who saw the problem split the list by hand. Every Monday they exported the signup list, sorted it into day-7 accounts, day-30 accounts and paying accounts, and sent a different email to each group. Five segments meant repeating the work five times, and one wrong filter put the free-trial message in front of paying customers. Because a person was doing it, the step that removes everyone who already got the message last week was the step that kept getting skipped.

A customer journey moves that sorting and sending into something you draw once and stop touching. You lay out a path made of conditions and branches, and each person climbs onto it and flows through on their own schedule. If the first email goes unopened, a second one goes out three days later with a different subject line. If it was opened but nothing was clicked, a different message goes out instead.

The key is that send time is attached to a person's state rather than to a date. A blast means everyone gets the same text on the same day. A journey means the same email arrives on different days for different people.

Building one path usually takes one to two weeks and breaks into five steps.

1. Write in one sentence what you're trying to change, and pick the number that tracks it. Something like: cut the share of teams that create no documents in their first two weeks from 40 percent to 25 percent. Write down the threshold you'd use later to decide whether to turn the path off.
2. Define the entry condition. Whatever decides who climbs onto the path has to be phrased so a machine can check it — "account is seven days old and has created zero documents" works because every clause is a value the system can read.
3. Draw the branches and the waits. Decide how many days to wait after the first message, and where the path splits depending on whether it was opened or clicked. Two or three branches is plenty at the start. This is also where you decide how far to take personalization, which fills different content in for different people.
4. Write the exit condition and the exclusion rules. Anyone who does the thing you wanted leaves immediately and receives none of the remaining messages — that's what stops a customer who just upgraded from getting an upgrade pitch three days later. Set a cap on how many messages one person can receive in a day so separate paths don't stack up.
5. Turn it on for a slice and fix it against the numbers. Run it for 10 percent of the audience for two weeks, check the open rate and the unsubscribe rate, then widen it. Switch it on for everyone from day one and a single bad line has nowhere to be walked back.

The value peaks when the same situation recurs on a different day for each person, and what to say about it is already settled. Accounts that did nothing in their first week, accounts three days from the end of a trial, accounts whose payment just failed. In all three the timing varies per person while the message stays the same every time.

It's a poor fit for news that only needs to go out once. A feature announcement can go to everyone on the same day, so a blast is faster. Four messages at fixed intervals belong in a drip campaign, and a single message fired the moment a behavior happens belongs in a triggered message.

The hardest limit is that journeys are easy to start and hard to stop. A path nobody is watching keeps running, so two years on you have twelve live journeys sending mail that the person who wrote them left the company. Screens get renamed in the meantime, and the button named in the email no longer exists in the product. Put a quarterly review of the journey list on the calendar and switch off the paths that aren't earning anything.

Adding branches makes judgment harder, which is the other limit. More branches means fewer people per branch, so a twenty-branch journey pushes about thirty people down each one and you can't tell which wording worked. Set the broad shape with lifecycle marketing, which decides what to say at each stage of a customer's life, and keep each journey down to two or three branches inside that.

"Let's pull that out of the blast and into a journey. Day seven with zero documents is the condition."

It isn't something everyone should get on the same day; it's something that hits each person at a different moment.

"Three journeys are overlapping right now — one customer got four emails in a day."

Nobody set exclusion rules per path, so the sends piled up.

"Only 40 people went down that branch, so the numbers can't settle it."

The branches were sliced too finely to tell which side is better.

At Pagedeck, Ravi owns the onboarding path. The entry condition is an account three days old with zero documents, and the first email asks them to create one. People who don't open it get a second message two days later with a different subject. People who open it but still create nothing get a different message suggesting they invite a teammate. Creating a single document drops them out of the path on the spot.

Mia looks at the numbers with him. Comparing the two weeks before and after launch, the share of accounts creating no documents fell from 41 percent to 29 percent, and the unsubscribe rate stayed at 0.4 percent.

The most common misreading is that a more complicated journey is a more sophisticated one. More branches feels like a closer fit for each person, which is a reasonable instinct. But every extra branch thins out the population inside it until nothing can be told apart. Run a twenty-branch journey for six months and you end up with flat results and a diagram nobody dares touch.

Reading "automated" as "runs itself" is the same mistake in a different place. The word suggests no human involvement at all. What a journey automates, though, is the sending — nothing in it checks whether the content is still true. Rename a screen or change the pricing and the message becomes wrong that day, and the wrong version goes out to hundreds of people daily.