CRM Marketing

For most of its history, marketing meant reaching strangers. Newspaper ads, flyers, the top slot in a search result — all of it points outward at people who have never heard of you. There was almost no channel back to someone who had already bought. A shop could keep names and addresses in a ledger, but mailing a few thousand of them meant printing and postage, and nothing in the process told you whether the letter brought anyone back. A sale was a single event, and the only way to grow revenue was to find more strangers.

Two things changed when products moved online. The signup form started collecting an email address, so a list of reachable people accumulated on its own without anyone building it. And sending one message to that list dropped to almost nothing: a stamp costs real money, while ten thousand emails cost a few dollars. A third change arrived alongside them. Every login, every click, every abandoned setup screen started leaving a record. The list of who to contact, the means of contacting them, and the evidence for when to do it all became available at the same time.

That flipped the arithmetic. At Pagedeck, acquiring one team through paid ads runs past fifteen dollars, while sending that same team one more email costs a fraction of a cent. What came out of last quarter's review was the gap between those two numbers going unused. Ravi reported 1,200 new teams for the month, Grace asked what had happened to the 1,100 from the month before, and the answer turned out to be that half of them had never come back after their first week. Money was going into getting people to the door, and nobody was speaking to them once they were inside.

CRM marketing is the name for the work of speaking to that half. Inside marketing, search, content and paid advertising bring in people who don't know the product; the relationship branch calls back people whose name and contact details are already on file. Different audience, different script. The first has to start by saying who you are. The second can start with whatever the person left unfinished, because the record already says what that was. The CRM in the name is short for customer relationship management, which usually refers to the system that stores every contact a company has had with a customer. This work runs on what that system holds, which is why it borrowed the same three letters.

Seventeen terms sit under this heading, and they sort into five groups that follow the path one message travels. Design decides what goes out and when. Purpose decides who you are trying to reach and what you want back from them. Channel decides how it travels. Metrics read the result. Consent decides whether you were allowed to send it at all. Running the groups out of order costs real work: check consent at the end and a finished campaign goes in the bin.

Design covers the shape and timing of what gets sent. There are five terms here, and they split by what the send time is anchored to.

1. Lifecycle marketing — picks the message based on the stage a customer is currently in. Someone who signed up and did nothing, someone using it alone, someone who has invited a team, and someone already paying each get different mail. Send one message to everyone and paying customers receive a pitch to start paying.
2. Customer journey — a scenario of conditions and branches drawn in advance, with each person flowing along whichever path their behavior puts them on. If the first email goes unopened, a version with a different subject line follows three days later; if it was opened but nothing was clicked, a different message goes instead.
3. Triggered message — fires the moment a specific action happens. The first time a document is shared, a failed payment, three days left on a trial. Because the timing is attached to something the person just did, it reads as a notice rather than an ad.
4. Drip campaign — a set of messages spaced at fixed intervals. A four-part introduction sent on day zero, day three, day seven and day fourteen is the standard shape. The timing hangs on the calendar rather than on behavior, which is what separates it from a triggered message.
5. Personalization — varies the contents per person inside a single campaign. The range runs from dropping in a first name to naming the document the person opened last.

Purpose splits by how far the customer has already drifted. Winback targets people who have actually left — cancelled, stopped paying — so it usually leads with a discount or something built since they went. A dormant user is someone who has not left but has gone quiet. Where the line falls depends on the product: fourteen days of silence marks a daily-use tool as dormant, while a tool people open a few times a month might wait sixty. The two are kept apart because the card you play differs. A dormant user often just needs to hear about one feature they never tried, while someone who already left will not return unless you can show that the reason they left is gone.

Channel is the pipe the message travels down. There are four, divided mainly by whether the person is inside the product at the time.

1. Email marketing — the oldest and cheapest pipe. It holds long content and leaves a record, but once it slips down an inbox it gets opened days later or never.
2. Push notification — an alert the app puts on the device's lock screen. It lands instantly, and the cost of overusing it is that people switch notifications off entirely, which no later campaign can undo.
3. In-app message — a panel shown to someone already inside the product. Its reach is small because only active users see it, and its response rate is high for the same reason: it appears while they are in the middle of the task it refers to.
4. Alimtalk — a business messaging format carried over a national chat app, used where that app is the default way people communicate. Open rates run far above email, but promotional content is largely off limits and templates have to be approved before anything goes out.

Metrics read what happened, and the four line up as gates a message passes through. Deliverability is the share of sent mail that reached an inbox rather than a spam folder; when it falls, every number behind it is meaningless. Open rate is the share of recipients who opened, which tests the subject line and the sender name. Click-to-open rate is the share of openers who clicked, which tests whether there was any reason to go further. Unsubscribe rate is the share who left the list afterward, which tests whether the volume or the content crossed a line. Read one number alone and you fix the wrong thing: a 40 percent open rate against a 2 percent click rate says the body is the problem, not the subject, while a 5 percent open rate against a 3 percent unsubscribe rate says the mail is reaching people who never wanted it.

Consent governs whether you may send at all, and the two terms differ by when permission is collected. Opt-in means sending only to people who confirmed in advance that they want the messages. Opt-out means sending by default and letting people remove themselves. Most markets require prior consent for promotional messages, so opt-in is the working default, and this group differs from the other four in what a mistake costs. Everywhere else a bad decision produces a weak number. Here it produces a fine.

The five groups form one loop. Consent supplies the list, purpose selects who on that list to contact, design sets the timing and the words, channel carries it, and metrics send a result back that reshapes the design. Leave one out and the rest spin. Skip the metrics and the same mail goes out for a year; keep the channel without the design and the work collapses into weekly sends to everybody.

"Let's send the winback mail only to teams quiet for three months or more. A full send just grows the unsubscribe list."

Widening the audience costs you customers who were perfectly happy until the mail arrived.

"Open rate is 42 percent and click-to-open is 2. The subject works, there's just nothing inside worth clicking."

The fix belongs in the body and the page behind the link, not in the subject line.

"This can't go out as an Alimtalk. It's promotional, so it only reaches people who gave consent first."

The content fell outside what the channel permits, so both the channel and the audience need rethinking.

Ravi owns this work at Pagedeck and brings last week's numbers to a Tuesday morning review. Last month he rewrote the triggered message that goes out three days before a trial ends. It used to show three pricing tiers side by side; now it leads with how many documents that team created and how many people they invited, with the pricing underneath. Click-to-open went from 3 percent to 9.

He and Mia collide at the same meeting fairly often. When Ravi wants an in-app message for a new feature, Mia points out that three are already running that week. An in-app panel belongs to marketing and to the product screen at once, so the two of them agreed on a weekly cap rather than settling it case by case. Grace checks one thing every quarter: whether revenue from this side came cheaper than revenue from ads. Last quarter ads took a hundred thousand dollars against eight thousand here, and while paid acquisition produced more conversions outright, the cost per conversion on this side was a twelfth of it.

The most common mistake is treating this as another name for sending email. Most of the messages really are email, and plenty of companies call the role email marketing outright, so the habit sets quickly. But email is one item inside one of the five groups. Narrow the field to that item and the purpose group, which decides who to contact, and the design group, which decides when, both vanish — leaving a weekly newsletter to the whole list as the entire practice. Six months of that, and the list is the same size while the open rate has slid from 30 percent to 12, at which point no subject line brings it back.

The second is expecting the tool to do the job. Automation platforms ship with branching, scheduling and templates already built, so it looks like signing the contract is the hard part. The first question the tool asks, though, is who should receive what and when, and the answer is not inside the tool. Which behavior defines a segment, how many days to wait, what the message should say — those come from people who know the product. Teams that buy first and answer later usually end up using the bulk-send button and nothing else until the annual contract runs out.