If you've been following along, you probably remember my quiet year. Back in What Is a KOL Anyway, I wrote about starting this Hong Kong travel account. For an entire year, it was just silent "moving postcards"—pretty footage set to music that almost nobody ever watched. Then came that late May night on the tram, which I shared in Look, There's Nobody in Causeway Bay!, when an overheard phone conversation pushed me to finally pick up my phone and record my first voiceover.

By that summer, as I wrote in A Tour Guide on a Screen, and a Unicorn, an AI-generated draft for Pok Fu Lam Village that read like a boring history textbook jolted me awake. I threw it out, rewrote it in my own voice without any preachy fluff, and used the full Pok Fu Lam walking guide to test a real structure: an immediate hook to stop thumbs from scrolling, a fast teaser montage of the day's highlights, and a step-by-step walkthrough of the stops. To top it off, I drew a custom walking map so people could actually follow the route. That reel broke 20,000 views within 24 hours.

Looking back, I wasn't just messing around on Instagram trying to become an influencer. I was running a really fun experiment.

Every day, I sat down with Gemini for our little "performance review," dissecting why one video blew up while another flopped, using the numbers to tweak my scripts and editing pace. Eventually, I started running controlled tests. What would happen if I took the exact same place and the exact same footage, but gave it a voiceover? Did the views explode simply because I started talking?

Place Names on a Screen Mean Nothing to Strangers

Let's rewind to March. Back then, the tram night was still months away, and I hadn't had a single viral reel. One day, I posted a cycling video documenting the ride from Tai Wai to Tai Mei Tuk. No voiceover at all. I just overlaid some text onto the screen listing the itinerary: rent a bike in Tai Wai, cruise along the Shing Mun River, pass Science Park, Tai Po, and Ting Kok, ride 20 km, and finish on the dam at Plover Cove Reservoir.

In my head, laying out the trip like that felt like "providing value." But to someone casually swiping through their feed, those random location names meant absolutely nothing. Most people had no idea where those spots were, and nobody is going to stop what they're doing to search for a bike route just because they saw a few pretty clips. It's too much work. So the reel quietly topped out at around 300 views. My bike ride never even made it into their heads.

Fast forward to July. By then, I had already produced over forty voiced reels. I was putting together a neighborhood series on Sha Tin, and while digging around for footage, I remembered the cycling ride. Tai Wai is technically part of Sha Tin, so it was a great fit. Best of all, I still had my original editing project file on my phone. All I needed to do was re-edit the clips and lay down a voiceover—an easy new video.

Turning a Journey Into a Game You Can See

Opening up the old files on my phone, though, I knew right away that simply recording a voiceover wasn't going to cut it. The original video followed my day in chronological order, which meant it opened at a bike rental shop. But watching someone pick out a bike is boring and visually flat. If that was my opening shot, nobody was going to stop and watch it.

So I shook things up. I made the first shot the most breathtaking view along the entire route to set the mood, followed by a montage of highlights. Then, in a voice that was completely confident yet warm and approachable, I dropped the line: "This is easily the best bike ride in Hong Kong!"

Next, I reassured them: "It's super easy, highly scenic, and takes you through different landscapes with zero steep hills." That way, they knew I wasn't asking them to climb mountains or exhaust themselves just to see nice views.

The challenge was how to show that visually. I had already teased some cycling clips in the intro, so continuing with more bike footage would just feel repetitive. Instead, I built an animated 3D map. It traced the entire route with clean red dots for the stops—no cluttering place names, just a clean image. Then I added a cute little animated cyclist pedaling along the line, with a cheerful sound effect popping off at every red dot, just like beating stages in a video game. Right on top of the animation, I stamped a clean title: "24 KM, Tai Wai to Tai Mei Tuk. Easy, scenic, zero hills."

That animated map solved several problems at once. In that second, people could understand the whole route. The clean route and the red dots planted a clear message: "There are lots of great stops," getting them excited to see what each one looked like. The game-like animation gave them that itch to level up and check them all off. Once that anticipation was locked in, I took them through the stops one by one to show them the real scenery. It was the same framework from the Pok Fu Lam guide, just reimagined in a much more playful format.

Around that time, I also started experimenting with CTAs (calls to action): leave a keyword in the comments, and I'd send a free route map directly to their DMs. The map I sent wasn't just a static picture—it was an interactive map that opened straight in Google Maps, so they could follow the route with their phone's GPS. My viewers didn't have to spend hours researching or worry about getting lost.

To send out the map, I taught myself how to use an auto-reply tool. What I didn't expect was just how crazy popular the video would get. Everyone who saw it genuinely wanted to go. The avalanche of comments asking for the map quickly used up the free plan's monthly limit. With the automation locked up, I had to sit there pasting saved messages into DMs by hand, one by one.

Within 24 hours of posting, that remade cycling reel pulled in over 30,000 views. Three months later, it was sitting at 117,000 views and had brought in 810 new followers.

Honestly, at the time, I didn't even realize I had just run an accidental A/B test. I simply thought I was being thrifty with old footage.

A Real Experiment at Shing Mun Reservoir

By August, though, I wanted to run a real, deliberate test.

Back in January, I had posted a reel about Shing Mun Reservoir. There's a grove of paperbark trees growing along the water's edge, and after heavy rains, the rising water drowns the lower trunks, creating a surreal "flooded forest" reflection. It's gorgeous. That original reel was pure moving postcard: pretty shots accompanied by elegant music. It stalled at around 300 views.

After making neighborhood walking guides for a while, I decided to start a "Summer Hikes" series, sharing shaded, easy trails that are pleasant even in the brutal Hong Kong heat. Shing Mun Reservoir was a perfect fit. So that August, I decided to remake it. But this time, I had a specific theory to prove.

With the bike video, I had swapped out the opening shot to grab attention, and I changed the music because the original track was too fast and loud, drowning out my voice. Strictly speaking, it wasn't a clean test.

This time, for Shing Mun Reservoir, I kept the controls tight: I reused the original footage, the exact same background track, and even the exact same opening frame.

So what did I actually change?

On top of the original shots, I added new clips for the one thing the postcard didn't have: the logistics. Which green minibus to take, where to get off, and a clear walking map once you arrive.

When people see stunning scenery online, their immediate reaction usually isn't, "Wow, why didn't I know about this?" Their real questions are: "How do I get there? Do I have to hike forever to see this? Will I get lost? Will I get heatstroke? Can I actually finish the walk?"

These worries are what I call "itinerary anxiety," and they are usually the exact reason people give up and stay on the couch. That's why I started offering GPS-linked maps—to strip away that hesitation so people could just get outside.

The loop around Shing Mun Reservoir is actually pretty straightforward, and it's hard to get lost. But the full loop is 9 km, which is still a decent trek even on flat ground. So in the reel, I split it into two options: what you'll see if you go counterclockwise, and what you'll see if you go clockwise. Viewers could pick the option that suited their energy and mood without feeling pressured to tackle all 9 km. By handing the choice back to them, the trip felt entirely under their control—without feeling pressured or stressed.

On top of all that, the video needed a proper name. Not just "Shing Mun Reservoir"—that's just a geographical label. I needed a name that set the scene, built a recognizable brand, and stuck in people's minds.

I thought back to the first time I saw those trees in the water, and their glassy reflections. That dreamlike setting, the sunlight dancing across the water... it felt as if the forest were magical.

That was it. I called the reel "Hong Kong's Magic Forest."

Within 24 hours, the Magic Forest reel picked up nearly 30,000 views. More than two months later, people are still watching it, and it has climbed to almost 150,000 views.

Remember: the exact same footage as a silent postcard had only managed 300 views.

Looking at those two wildly different numbers, everything finally clicked. I understood what "providing value" actually meant.

I used to think that digging up hidden gems and offering weekend ideas was value. But not knowing a place exists isn't what stops people from going out. What stops them is itinerary anxiety: not knowing how to get there, being afraid of getting lost, or dreading the hassle of mapping out a route. Now, I act as their personal tour guide and automatic navigator, mapping out the walk in advance and whispering the sights into their ears. They get to join an effortless virtual tour, simply following along.

That's the kind of video that is genuinely useful to the person on the other end of the screen. That was the value I was giving them.

The Curve That Stood Up

Between June and August, my account gained about 8,000 followers. At some point that summer, looking at the analytics, I realized the growth of my account had shifted.

In the first month, I had clawed my way to 1,000 new followers. But by the end, I was pulling in 1,000 in a single week. The growth wasn't inching along anymore; the line was standing up, bending into an exponential curve.

Looks like everyone really wanted to join Sophia's online tour group.