Building K-Saju 2: the hard part after launch was promotion

If you build the feature, will people just find it?

In the first K-Saju post, I wrote about why I wanted to turn saju into K-culture character cards and friend chemistry tests. At that stage, most of my energy went into shaping the idea, finding the right visual tone, and making the service actually work.

K-Saju now has two main flows.

The first one is My K-Saju.

You enter your birth date, birth time, and a few basic details, then get a K-Saju type, an element balance, a K-pop-group-style role, and a K-drama-like character interpretation.

The second one is Friend Chemistry.

Instead of reading only your own result, you can enter a friend’s birthday and see how the two of you fit together. I wanted this to feel less like a serious “compatibility reading” and more like something you can send to a friend, laugh about, and use as a small conversation starter.

After building those two features, it was tempting to think:

Now people will try it, share it, and maybe send feedback.

But that is not how it works.

Building a feature and helping people discover that feature are two very different problems.

Promotion feels unfamiliar as a developer

I am a developer. When there is a product problem, I can usually turn it into a technical loop: define the behavior, write the code, deploy it, watch the logs, and fix what breaks.

Promotion is different.

What should I show first? Which sentence should I use? Should the tone be playful or direct? Which platform should get which kind of post? Is one announcement enough, or should the same idea be repeated in different formats?

These questions do not produce clear compile errors.

When building the product, I can check whether the button works, whether the result appears, and whether the mobile layout breaks. When promoting the product, it is much harder to know whether the message actually reached people.

If a post gets low views, is the video weak? Is the first frame boring? Is the copy unclear? Was the timing wrong? Did I choose the wrong audience? Or is the product simply not interesting enough yet?

That is the part I am still learning.

For a side project, launching the site is not the finish line. I also need to build the paths through which people can discover it.

This is especially true for K-Saju. It is not a utility people are likely to search for when they have a practical problem. It is closer to something that needs to appear in a feed or a conversation, followed by a quick “Want to try this?”

What I am trying now

Right now, I am testing a few small channels.

The first one is TikTok and Reels.

K-Saju is visual. The result cards, character images, and chemistry screens are a big part of the product. A long text explanation is not always the best way to show that. A short video can say “this is what happens” much faster.

A simple video flow might look like this:

• enter a birth date;
• show the K-Saju result card;
• reveal the K-pop-group-style role;
• show the K-drama character energy;
• move into the friend chemistry test.

One video will not magically create traction. But short video feels like the fastest way to communicate the mood of the service.

The second channel is X and Threads.

For those, I am posting short updates with a video attached. The copy is simple: I built a small side project where you can use your birthday to get a K-pop/K-drama-inspired character type and test your chemistry with a friend.

The third channel is dev.to.

I want to separate the role of my personal blog and dev.to a little.

On this blog, I want to write more openly about why I built K-Saju, how the idea changed, and what I learned emotionally while making it.

On dev.to, I want to write for developers: how the result types are structured, how the character-card data is organized, where AI helps, where I intentionally keep manual control, and what I learned from building a small product around a cultural concept.

Both are useful, but they should not be the same post.

I need a more deliberate strategy

So far, my promotion has mostly been “post something when I remember to post something.”

That does not scale well, even for a tiny side project. Every time I start from a blank screen, I have to write new copy, make or edit a video, choose hashtags, decide where to post, and guess what tone fits.

After doing that a few times, I start spending more energy deciding how to talk about the product than improving the product itself.

So I want to make the process more deliberate.

One idea is to split a single feature into multiple angles.

For example, Friend Chemistry can be explained in several ways:

• something fun to try with a close friend;
• a light relationship chemistry test;
• a K-drama-style character-pairing result;
• imagining the two of you as roles in a K-pop group;
• sharing a result card after the test.

The feature is the same, but the hook is different. One person may respond to the K-pop angle. Another may care more about friend chemistry. Another may simply like the character-card visuals.

Another idea is to change the language depending on the culture.

In Korean, words like “saju” and “compatibility” already have context. In English, starting with those terms can create distance. For an international audience, it may be better to begin with phrases like:

• K-culture personality card;
• K-drama character energy;
• friend chemistry test;
• K-pop group role.

Saju can still be part of the story, but it may not need to be the first word in every piece of copy.

I also need clearer feedback paths.

Instead of vaguely asking people for feedback, I should ask more specific questions:

• Was the result fun to read?
• Was any part hard to understand?
• Would you share the friend chemistry result with someone?
• Did the result card feel visually appealing?
• Did the English phrasing feel natural?

Those questions would make it easier to decide what to improve next.

What I want to semi-automate

The goal is not to automate promotion completely. At least not now.

What I want is semi-automation: keep the human judgment, but reduce the repetitive work.

The rough workflow I have in mind is:

1. Start from a blog post or product update.
2. Generate several short promotional copy variants.
3. Generate TikTok/Reels script and caption drafts.
4. Generate X/Threads post variants in different tones.
5. Generate dev.to article ideas and outlines.
6. Review and adjust everything manually before posting.

I do not want fully automated posting yet. K-Saju is still finding its tone, and cultural wording matters here. A careless sentence could make the product feel too mystical, too deterministic, or simply confusing.

But not starting from a blank page every time would already help a lot.

In development, templates and scripts reduce repeated work. I am starting to think promotion needs the same kind of system.

Should I keep posting videos?

For K-Saju, I think the answer is yes.

Text alone does not communicate the product quickly enough. People need to see the birthday input, the result card, the character image, and the chemistry result. That sequence explains the service faster than a paragraph can.

But I do not think the goal should be one perfect promotional video.

It is probably better to test multiple short formats:

• 7 seconds: show only the result card;
• 15 seconds: show birth date input to result;
• 20 seconds: show the friend chemistry flow;
• comparison hook: “What is our K-drama chemistry?”;
• reaction hook: “I tried my birthday as a K-pop/K-drama character card.”

The important part is to show the actual result screen. The card itself is stronger than a long explanation.

The next experiment

K-Saju is still small. Traffic is low, and I need more feedback.

But that also makes this a good time to experiment. Before adding many more features, I want to understand what people actually respond to.

Is the personal K-Saju result fun? Is the friend chemistry flow shareable? Does the K-pop/K-drama doorway make sense to international users? Do the result cards feel worth saving or sending?

Those answers should shape the next version more than my assumptions do.

Development is still the area where I feel most comfortable. But if I want a side project to grow into a real service, I need to learn how to explain it, test messaging, and ask for feedback.

This second post is mostly me admitting that promotion is hard.

I am not good at it yet. But I am starting to think of it as part of the product, not a separate chore that happens after launch.

Next time, I want to write about the actual promotional copy and video formats I tested, and what kind of response they received.

You can try K-Saju here:

Open K-Saju