Six AI videos, ¥153,030 in the first month — one video with 4.22 million views drove two-thirds of TikTok monetization revenue

Six posts, ¥153,030 in revenue, in the very first month of TikTok monetization. That’s the figure AI video creator Ponzu published on note.

The numbers alone look efficient, but opening up the breakdown changes the impression. About two-thirds of the revenue came from just one of the six videos. This is a case of “hit once, and it’s worth ¥150,000”, not one of “make six videos, get ¥150,000.” How to read that difference is the center of this article.

View counts and revenue for the six videos

The recorded period runs about one month, from June 3 to June 30.

Meanwhile, the account’s total view count shown by the creator exceeds 10.48 million, leaving a gap of about 2.1 million views versus the sum of the six. The article gives no explanation for this gap. It’s possible it includes posts from before June, but there’s no way to confirm that.

Revenue across the six videos totaled ¥153,030, of which about two-thirds came from Video A, according to the creator, roughly ¥100,000.

The effective unit price differs by nearly 2x between videos

Converting this breakdown into a per-view rate reveals an interesting gap.

Video A generated roughly ¥100,000 from 4.22 million views, or about ¥0.024 per view, an RPM-equivalent of about ¥24. The remaining five videos together generated about ¥50,000 from roughly 4.16 million views, or about ¥0.012 per view, an RPM-equivalent of about ¥12. Same account, same period, same creator, yet the effective unit price differs by nearly 2x.

This is exactly what the creator points to in emphasizing: “calculating a per-view rate is entirely meaningless.” Only views shown through the For You feed count toward revenue. Views from followers are excluded. Mixed within the raw view count, then, are views that count toward revenue and views that don’t. Given the same 1 million views, a video with a higher proportion of For You-driven views earns more. Since videos that go viral tend to have a higher proportion of For You-driven views, Video A’s higher unit price reads as a natural result.

One video was the decisive factor

The turning point in this case is clear-cut: Video A recorded the highest view count ever for the account. The creator describes this as “an unexpected event.”

The before-and-after figures are notable too. Followers stood at 3,000 as of May 15, and had reached 32,000 at the time of writing. June’s increase is recorded as +21,000 followers, and the account’s total view count grew by +433.7%. Note that 3,000 plus 21,000 doesn’t add up to 32,000. The article offers no explanation for this discrepancy either. Regardless, it’s clear that a change in scale of an order of magnitude occurred within a single month.

According to the creator, Video A was produced based on a note course titled “Building Viral Content.” So this is positioned not as an accident but as something made deliberately, which then hit.

What worked

Extremely low production costs. The creator’s banner is “Story × AI,” produced using video-generation AI tools like Runway, Kling, Pika, and Vidu. No filming, no on-camera appearance, no location shoots. Six videos in a month may look like a small number, but achieving this scale of views with live-action footage in six videos would be difficult. To compress this case into a single line: only those who can miss cheaply get to draw a hit. The advantage of AI video lies less in image quality or speed than in being able to keep trying at a price that tolerates five misses.

Being self-aware about the “waiting for a hit” structure. The fact that two-thirds of revenue is concentrated in one video isn’t something the creator hides. It’s foregrounded. This is also an acknowledgment that the business doesn’t earn evenly across all six videos, but depends on a single outlier. Understanding this structure means that if revenue drops next month, it won’t feel abnormal.

Holding a hypothesis about format. The creator has hypothesized that TikTok currently favors one-minute videos, and reports that one-minute videos have significant growth potential, citing a separate case of “1.7 million views, ¥20,000 with a one-minute video.” This is a pattern of testing what the algorithm favors in short cycles and shifting formats accordingly.

Not relying on a single revenue source. In that same June, the creator began selling explainer articles on note, earning about ¥70,000, and also launched a LINE official account. Against TikTok’s ad revenue of ¥153,030, content sales brought in ¥70,000, a ratio of over 30% coming from income unaffected by platform rate fluctuations. Building this composition from the very first month is early.

Risks not to overlook

The biggest issue is that there’s no guarantee this ¥153,030 will recur next month. If a hit on the scale of Video A’s 4.22 million views doesn’t happen again, revenue could drop to the level of the remaining five videos combined, around ¥50,000. The editorial team reads this ¥150,000 as the cash value of a winning ticket rather than as monthly income. Revenue that depends two-thirds on a single outlier is, by definition, extremely unstable.

The creator raises this issue too, citing over-focus on revenue as a problem. “Staying nimble and turning various ideas into form is important,” and “sowing seeds for the future matters more than immediate view counts”, reflecting a wariness that revenue-consciousness narrows creative freedom. This reads as an awareness of a structural risk: chasing a repeat of a hit by imitating past winning patterns risks falling behind as the algorithm changes.

In addition, the downward trend in TikTok’s rates itself has been pointed out in other creators’ published data. Video A’s effective RPM of ¥24 is a favorable level for now, but it isn’t permanent.

Conditions and limits of replication

What’s easy to replicate is the AI video production itself. Tools like Runway, Kling, Pika, and Vidu are accessible to anyone, and initial investment fits within monthly subscription costs. No filming equipment or on-camera talent required, with only six attempts, entry cost is low.

What’s hard to replicate is landing the hit itself. The ¥153,030 result in this case only materializes because of the single 4.22-million-view video. Using the same tools and the same volume, if that one video doesn’t happen, the number lands in the ¥50,000 range. The creator has a sales background and holds a perspective of “not words that sell, but footage that moves”, producing according to a template acquired from a course, yet even so, the creator describes it as “an unexpected event,” meaning it isn’t written as something achieved by intent.

And this is one month’s worth of record. What happens in month two or three isn’t covered in this article. Whether the first month’s numbers hold up requires an entirely separate verification.

What’s worth taking from this case is structure, not the amount. Social media ad revenue is determined by outliers, not averages. Revenue is generated by For You-driven views, not raw view counts. And there’s meaning in establishing a separate line of content sales from the very first month. These three points apply to every business built on the same revenue model, AI video being only one of them.

Related reading

• Photo AI — solo development and AI products
• Hitode’s blog revenue