Acquisition Channel

Back when a team only counted signups, nobody could say where those people came from. The month ended with one number, 3,000 new accounts, and no one in the room knew whether those 3,000 had found the product through a search result or clicked an ad while scrolling past it. The number going up made everyone happy and the number going down made everyone nervous, but neither direction told anyone what to do more of next month, or what to stop doing.

So budget meetings were won by whoever argued hardest. "We should spend more on ads" ran into "we should publish more articles," and since nobody could say how many of last month's 3,000 each side had brought in, the two sides compared instincts. The outcome was usually one of two things. The team repeated whatever it had done the year before, or whoever joined most recently transplanted the plan from their previous company.

The conversation changed once teams started counting the paths separately. Split 3,000 into 1,400 from search, 900 from ads, 400 from referrals and 300 typing the address directly, and each line raises a question of its own. Which articles pulled in the 1,400? What did each of the 900 cost? Who invited the 400? None of those can even be asked while the number is a single lump.

Once the lines are separate, it becomes obvious that they behave differently. A thousand people from ads and a thousand from search are not the same thousand. One cost 20 million won and the other cost nothing, and the share still around a month later is usually higher on the search side, because someone who arrived while looking up their own problem is a different visitor from someone who tapped an ad between two videos. That is why acquisition channels ended up being treated less as a way to count users and more as a way to decide where the next budget goes.

The terms filed under acquisition channels fall into three groups: how the paths are divided (types), what happens when a working path stops growing (limits), and how to tell whether a path suits this particular product (judgment).

Types split the paths themselves. The dividing lines are whether money was spent and where the visitor came through.

1. Organic acquisition — people who arrive without a single won of ad spend, usually by reading an article that ranked in search results or by hearing about the product from someone. The cost per person is close to zero, but the build-up is slow. An article typically takes three or four months to climb into the top search results, and once it starts bringing in 200 people a month it keeps doing so without further spend. That does not make it maintenance-free. When someone else publishes a better piece on the same topic your ranking slips, so an article usually needs a refresh every six months to hold its position.
2. Paid acquisition — traffic bought with an advertising budget. Put 10 million won in today and people arrive today; switch it off tomorrow and they stop tomorrow. It behaves like a tap. Water runs the moment you open it, stops the moment you close it, and nothing pools up in between. Young teams reach for this group first because results show up within a week, and the reason it's risky is the same one: the day the budget ends, six months of traffic ends with it.
3. Direct traffic — people who typed the name into the address bar or came in from a bookmark. It usually reads as a good sign, since these are people who already know the product by name, but the bucket also exists because some arrivals carry no source at all. A link pasted into a chat app, or a tap from inside another app that strips the source tag, lands here. So a sudden jump in direct traffic prompts one question first: did the brand get better known, or did a source tag somewhere break? A jump that appears overnight rarely comes from awareness, so the usual answer is that a campaign link lost its tags in a recent release.

Limits cover what happens when a path that used to grow stops growing. Channel saturation is the point where more money or more effort in the same channel stops buying more results. Teams normally spot it in the cost column before the traffic column, because the count holds roughly steady for a while as the price per person creeps up. A channel that brought 500 people for 10 million won and then brought 700 for 20 million returned 1.4 times the people for twice the spend. The cost of bringing in one person — the customer acquisition cost (CAC) — went from 20,000 won to 29,000. Saturation isn't a broken channel so much as a channel where you've already met nearly everyone it can reach. If the fifth showing of the same ad is going to the same person, the sixth landing worse than the first is what you would expect. A channel like this behaves like a field: the first few harvests reward the digging, then the same effort starts yielding less, and at that point looking for another field beats digging deeper.

Judgment covers choosing which path to walk into at all. Channel fit is how well a channel suits the product. Traffic counts alone won't settle it. If a collaborative document tool runs short video ads, draws 800 visitors a day, and exactly three of them invite a teammate, that channel produced traffic rather than customers. So fit is checked against what an arriving user is worth — the lifetime value (LTV) — against what they cost to bring in. Spend 30,000 won per person on someone who pays 15,000 won over their whole life as a customer and every extra visitor deepens the hole. Fit also isn't settled once. When the product changes, the channels that suit it change too. A note-taking tool that grows into a team tool finds that the channel that reached individuals never reaches the team lead holding the budget, and last year's best channel becomes this year's most expensive one. The cheapest way to test fit is small and early: run a fixed, modest budget for two or three weeks, then judge it on what those users did in week two rather than on how many signed up in week one.

The three groups work in sequence. Types draw the map of available paths, judgment picks which one to dig into, and limits say when to walk away from it. Teams that only watch one group tend to stall in predictable places. A team that knows types alone lists every channel, touches each a little, and gets nowhere in all of them. A team that ignores limits keeps pouring budget into a channel whose returns bent downward two quarters ago. Most teams also end up running fewer channels than the map offers. Two paths that work beat six that half-work, since every channel carries its own learning cost and none of them pays back until someone has spent months inside it.

"Half of this quarter's signups come from search. Ads are 20 percent."

A read of the channel mix, usually offered as evidence that cutting the ad budget still leaves half the traffic standing.

"Costs on this channel have climbed two months running. I think it's saturated."

More money isn't producing proportionally more people, so the budget should move elsewhere.

"Traffic is fine, but nobody sticks. I don't think this channel suits our product."

A request to judge the channel on what visitors do after arriving rather than on how many arrive.

At Pagedeck, the monthly growth meeting opens on the same screen every time: a table of channels. Ravi splits the month's 3,200 signups into four rows — search, ads, referrals, direct — and each row carries the count, the cost per person, and the share still active a month later. Once the table is up, the conversation almost always starts with the most expensive row.

Last month Grace wanted to double the ad budget. The reason Ravi pushed back sat inside the table: three months ago an ad-driven signup cost 18,000 won, last month it cost 27,000, and the count had barely moved over the same stretch. Doubling the budget would not double the people, so the discussion turned toward putting that money into writing for search instead. Mia pointed at a different row. Referral signups stayed active at nearly twice the rate of ad signups but made up only 12 percent of the total, which made reworking the invite flow look like better value than any ad decision. The meeting settled one budget question, but every input to that decision came out of the rows and columns. Without the table, the same hour would have gone to arguing about whose hunch sounded better.

The most common mistake is reading the biggest channel as the best channel. Volume is the first column anyone's eye lands on, so it reads that way naturally. But volume is only the first box a channel fills. Between a channel that brings 800 people a day and keeps 20 of them a week later, and one that brings 120 and keeps 70, the second produces more customers every month. Teams that pick channels by volume tend to arrive at the same place about six months in: signups still climbing, paying teams flat, ad spend up. By then the signup screen and its copy have been rewritten around the users that channel sends, so turning the ship around costs several more months.

The second mistake is assuming each user belongs to exactly one channel. The table has one row per channel, so people look like they should fit in one row too. In practice someone sees an ad, learns the name, hears about the product from a colleague two weeks later, then searches for it and signs up. That person passed through three channels and lands in the search row alone. The method for deciding which touchpoint gets the credit is called attribution, and the same month's data produces very different tables depending on which method is set. A team unaware of this switches off whichever channel the table scored lowest, and only afterward learns that the channel had been feeding the rows above it. That is why search traffic sometimes drops the month after the ads stop, and why a channel is usually wound down gradually rather than cut in one move.