Analysing competitors for creating an AI influencer

It’s been quiet for a while. Time for an honest update about my AI Influencer project.

And I’ll be upfront: things aren’t going as smoothly as I’d hoped.

The Numbers Don’t Lie (And They’re Not Pretty)

Over the past few weeks, I’ve published several new tracks on Spotify under BoxxBoxx’s name. They’re doing… practically nothing. Sure, you could say that’s expected. But I’ve been running multiple artist accounts where I test all kinds of things, especially around Spotify SEO. And I regularly see songs take off out of nowhere.

I was hoping that would happen with BoxxBoxx. It hasn’t.

I’ve also published several vlogs where BoxxBoxx talks to the camera about Formula 1. Short videos. I create them with Veo from Google – a video model where I use an image of BoxxBoxx as input. I make those images with Nano Banana, which I’ve mentioned before.

The reach? Disappointing across the board.

On Instagram, it stays extremely low. On TikTok, everything hovers around 700 views. On YouTube Shorts, I’m getting the best results with 1,000 to 2,000 views per vertical video.

What stands out? I’m not going viral. And I’m barely gaining any followers. Growth is painfully slow.

Studying Success: The AI Rockband Blueprint

So I started analyzing. Really diving deep.

Last weekend, someone sent me articles about an AI rockband in America. A rockband with over a million monthly listeners on Spotify alone. They even got a record deal.

I dug into what they did to grow. And I found two main strategies: PR and vertical videos.

Here’s how it worked:

They created vertical videos on TikTok, Instagram, and YouTube Shorts. These videos introduced people to their music and made them search for it on Spotify. Spotify noticed. And Spotify rewards that behavior.

The more people you send to Spotify, the more Spotify pushes your music. It’s a feedback loop.

But here’s the thing about PR: you need data first. You need something to show.

So they generated reach with vertical videos first. Then they approached the media saying, “Look, an AI artist with serious streaming numbers.” That created an explosion of even more streams. Spotify rewarded that. An upward spiral began.

They repeated this pattern. Each explosion in streams became fuel for the next PR push. The result? A beautiful upward trajectory and a record deal.

The Four Types of Viral AI Content

I also analyzed what AI videos I actually encounter online when I’m scrolling. I came up with four categories:

40% falls into humor. Often pretty crude or vulgar humor.

40% is fake news. Content deliberately designed to mislead people. Like when Max Verstappen became a father, my Facebook feed showed AI-generated images of Lewis Hamilton visiting him in the hospital. Completely fabricated.

10% is extraordinarily original, beautiful, touching, or nostalgic.

10% is what I call “what the fuck” content. Completely bizarre things. Like last weekend when I kept seeing videos on TikTok of a chiropractor helping an elderly woman on a massage table. Suddenly, the chiropractor grabs the woman and throws her through a wall. It makes zero sense. But that high “what the fuck” factor gets people sharing.

Looking at my content? It doesn’t really fit any of these proven categories.

Sure, I sometimes try humor or share opinions. But it’s not funny enough. Not original or touching enough. Not “what the fuck” enough.

And I don’t want to do fake news.

But this is something to think about. I could try to create a new successful category. Or I could be smarter and work within categories that have already proven to work.

The challenge? I don’t want to be too offensive or too crude. I don’t want to insult teams or drivers. I like humor, but I don’t want to damage people. And long-term, I want to remain attractive for brands to work with.

The more you lean into offensive humor, the harder that becomes.

The Sjonnie Slijptol Success Story

Here’s where it gets interesting. With another account I haven’t talked much about, I’m actually using the same approach as that American rockband.

The artist is called Sjonnie Slijptol. I work on this with a friend. It started at the beginning of this year as a carnival joke – making music for people from my region. After two days, it was already played on RadioNL. Pirates stations picked it up. Spotify saw the positive signals and started pushing it.

The upward spiral began.

Between YouTube and Spotify, I’m expecting to hit half a million streams in about a month. That’s substantial.

And I can see exactly where the peaks come from. They align with moments when I was active with vertical videos on TikTok. Or when radio stations played the track. Or when I appeared in two newspapers last week.

Each time I appear in a newspaper, I see two effects: First, people go directly to Spotify. Then, after that direct effect fades, an even higher peak appears because Spotify noticed people searching for it and pushes it even harder.

It’s fascinating to see this mechanism in action.

Why BoxxBoxx Isn’t Ready For PR Yet

For BoxxBoxx, PR is too early. I don’t have reach. The story isn’t interesting enough yet for a newspaper or TV station.

I actually got approached by a Dutch TV station about Sjonnie Slijptol. They wanted to position it against the traditional music industry, focusing on the ethical side. But I know how TV works. You talk for an hour, they cut it down to three sentences. Since they were setting up two opposing groups, I was afraid they’d strip out all nuance and do more harm than good.

So I declined.

But the lesson is clear: With BoxxBoxx, I need vertical videos first. Find the right angle. Go viral a few times. Repeat what works. Then pursue those PR moments.

The Random Nature of Viral Success

I learned something fascinating last Sunday. I posted a vlog – created with Veo again. Eight seconds long. A crude, edgy joke.

Until that point, everything I posted on Sjonnie Slijptol’s Instagram had a maximum reach of 2,000.

This one video? 140,000 views in two days. On Instagram alone.

The funny thing? That exact same video did the absolute worst on TikTok of everything I’ve posted in months.

What you see is that success is incredibly random. Something that works on one platform with the same description and hashtags doesn’t necessarily work on another. Sure, there are slight differences in audience demographics. My Instagram audience is a bit older than TikTok, which is also older than YouTube Shorts.

But still, you’d expect some correlation. If something reaches 140,000 people in two days and people find it funny, you’d think it would do better on other platforms too.

I also gained 200 followers from that video. Not massive numbers, but if you can repeat that viral success a few times, you reach a point where new content goes viral more easily because you have more followers who like that content.

The Patience Game

Here’s another realization: I’ve been posting content on Instagram for quite a while that did nothing. All in the same style, with the same humor.

Apparently, if you stay consistent long enough and create enough content, one can unexpectedly take off.

With BoxxBoxx, I’m not there yet. I haven’t created and posted enough.

The learning? Patience.

Looking back, I think I was a bit naive. When I see content from real influencers pass by, I often find it pretty average. A lot of influencers just post about what they’re eating or other things that don’t interest me personally.

That gave me the impression that if you post that kind of surface-level content, you’ll achieve the same results.

But that’s simply not true.

This is hard work. Trying lots of things. Building on what works. Persisting.

Maybe I’ve been spoiled by past examples where things took off immediately. That’s not happening here.

I need to keep searching, keep tweaking what I’m doing. But I’m convinced it will come.

Building The Network

Meanwhile, I’m continuing to network.

I’ve scheduled a podcast with Rudy van Buuren in November. One with Henk Kuipers next week. I’ve also made an appointment with Richard Otto, owner of Spreekbuis.nl – a platform read by many people from the media world who work at Mediapark in Hilversum.

And something important: In the podcast with Patrick Kicken, he mentioned that I need a proper profile for BoxxBoxx. Who is she? What does she stand for? Her background. Where does she live? What does she like? Does she have pets? Is she in a relationship? How old is she?

Having that defined allows me to account for it in my content and build a deeper connection with followers.

On one hand, this feels unimportant right now since I don’t have those followers yet.

On the other hand, it’s better to have it from day one. So I created it. I’ll talk more about that in another episode.

I also have a meeting in two days with a former colleague who started making music with AI, to discuss what’s needed to get this off the ground.

The Bottom Line

This journey is teaching me something crucial: viral success with an AI influencer isn’t about shortcuts or tricks. It’s about understanding proven content categories, being patient enough to create volume, and strategic enough to leverage early wins into PR opportunities.

BoxxBoxx isn’t there yet. But now I have a clearer roadmap based on what actually works.

The next phase? Finding that angle that fits a proven category while staying true to what I want BoxxBoxx to represent. Creating more content. Testing relentlessly.

And when the breakthrough comes – because I believe it will – being ready to turn it into that upward spiral.

Stay tuned.