Issue #45: The Second Number 💎

Issue 45
August 27th, 2026

Last week I told you to screenshot your 28-day view totals before Monday, because YouTube was about to change what a view is and you'd want the clean before-and-after. I did mine. And then I sat there Monday afternoon watching a number go up for no reason at all, which is a genuinely strange feeling.

Here's what I didn't expect. That change turned out to be the smallest version of a thing that happened four separate times this week.

Because if you look at it plainly, YouTube just formalized something every creator already knew and mostly tried not to think about: you have two numbers. There's the number that goes up, the one on the front of the video, the one in your media kit, the one people congratulate you on. And there's the second number, the one that actually converts into money. YouTube used to let those live in the same box. As of Monday they're in different boxes with different names, and the gap between them is now a permanent, visible, measurable thing.

Then Messi's hydration brand shut down with half a billion followers behind it. Then two of the most respected filmmaking creators on YouTube took money from an AI video company and got absolutely hammered by their own audiences for it. Different stories, same shape. In every one of them somebody found out that the big number and the number that pays are not the same number, and that you can't spend the first one.

In this issue:

Let's dig in.

A $5.4 billion AI video company paid two beloved filmmakers to say nice things about it, and their audiences sent them the invoice.

On August 17, Higgsfield announced a $400 million Series B led by DST Global, at a company-reported $5.4 billion valuation. Normal enough. AI video is where the money is going right now and everybody knows it.

Four days later, The Verge reported that Matti Haapoja and Sam Kolder, two filmmaking creators with enormous credibility in the "learn to actually shoot things" corner of YouTube, had both posted videos showing off what Higgsfield's Seedance 2.5 could do and framing it as the future of production. Neither video was labeled as an ad. Neither creator answered questions about whether it was paid. Higgsfield, when asked directly, confirmed the arrangements were formal partnerships and that the creators were compensated through a negotiated mix of cash and platform credits.

The reaction was fast and it was not gentle. Marques Brownlee publicly pushed back on Haapoja's comparison of generative AI to the Canon 5D Mark II, pointing out the obvious thing, which is that the 5D didn't get built by ingesting the uncredited work of the people it was sold to. Other creators started posting screenshots of the partnership offers they'd received from PR firms working on Higgsfield's behalf, which is its own kind of message: here is the going rate, here is what they thought I'd take.

I want to be careful about what the story actually is, because "creators shouldn't take AI money" is a take and I don't think it's the interesting one. Plenty of creators have taken AI money without anything happening to them.

The interesting thing is the structure. Look at how these two got paid: cash plus platform credits. That's a compensation package where part of your fee is denominated in the exact product your audience is upset about. If the credits are worth anything to you, you use the tool more, which deepens the association. If they're worth nothing to you, then you sold the endorsement for less than the headline number and took the reputational hit at full price either way. There is no version of that trade where the credits are neutral.

And it landed in a week where we finally have decent numbers on what that hit is worth. Five days before the Higgsfield videos went up, CreatorIQ published its State of Creators report, surveying 5,095 creators. The headline finding is brutal if you sell sponsorships for a living: only 15% of creators say they fully trust sponsored content from other creators when they're deciding what to buy. These are professionals evaluating their own peers. 11% said flatly that they're skeptical because creators are paid to say positive things.

The other number in that report is the one I keep coming back to. 53% of creators with 500,000 or more followers report feeling tension between what their audience wants and what the brand wants. More than half. At the tier where the deals get big enough to matter, the majority of people are already living inside this conflict and mostly just absorbing it quietly.

What Haapoja and Kolder ran into isn't new, then. It's the ordinary condition of sponsored content, except AI turned the volume all the way up because it's the one category where the product itself is an argument about whether the creator's craft has value. You can take a mattress deal and nobody thinks you've betrayed something. Take a generative video deal when your entire channel is built on teaching people to frame a shot, and the sponsorship isn't adjacent to your work, it's a position on your work.

The bottom line: every sponsorship has had two numbers all along. There's the fee, and there's what it costs you in audience, and until now the second one has been unmeasurable so most of us have been treating it as zero. It isn't zero, and CreatorIQ just gave us the first decent proxy for it. If only 15% of your peers fully trust sponsored content, your audience's number isn't higher than that. Which means the real question on any deal isn't "is this fee good," it's "is this fee good relative to what I'm spending." Most creators have never done that second calculation because there was no way to. There sort of is now. There's a pro tip at the bottom of this issue that walks through how I'd actually run it.

Also worth saying plainly, because it's the cheapest lesson in the whole story: the thing that turned a debatable endorsement into a trust event was the missing disclosure. In the same CreatorIQ data, explicit sponsor disclosure is one of the named reasons creators trust a peer's sponsored content. Labeling the video wouldn't have made the AI question go away. It would have kept the conversation about AI instead of about honesty, and those are very different conversations to be having with people who subscribed to you.

Three data points that tell a bigger story.

Lionel Messi has 516 million Instagram followers, and Digiday reported this week that his Mas+ hydration brand is shutting down less than two years after launch. In the same piece, Alex Cooper's Unwell beverage line is being discontinued this fall. Neither of these happened in a dying category, which is the part that should get your attention: functional beverages grew 29% year over year in 2026. The demand was there. The audience was there. The repeat purchase wasn't. Sunny Bonnell of the branding agency Motto put it in a line I haven't been able to shake: "There is a big difference between having attention and having permission to enter a category." A beverage distribution veteran quoted in the same story was blunter about why follower counts don't translate. "Retail is a velocity business. Those are not the same currency." Two exits don't prove creator-led brands fail, and I'd resist that reading (Mas+ also had a saturated category and trademark litigation working against it). What they do prove is that the conversion rate from attention to purchase is a real number, it's specific to your audience and your category, and nobody can tell you what it is in advance. Launch reach is the first number. Reorders are the second one.

Everybody covered the view count change, including me. Almost nobody put it next to the thing YouTube announced in the same breath on August 10: starting February 1, 2027, new creators need 8,000 qualified watch hours in 365 days (up from 4,000) or 20 million qualified Shorts views in 90 days to get into the Partner Program. Double the current bar. There's also a new 10 million Shorts view threshold in 90 days just to share in Shorts ad and subscription revenue. Existing YPP members are grandfathered and unaffected. So run the two announcements together and the shape is unmistakable. On August 24 the number the public sees got easier, because a view now counts from the first frame across every format. On February 1 the number that gets you paid gets twice as hard. Those aren't contradictory decisions, they're the same decision: YouTube is deliberately decoupling the reach number from the revenue number and telling you which one it prices on. Digiday's winners and losers breakdown is the best read on the trade-offs. If you're pre-monetization, the actual news is that you have until January 31, 2027 and the bar you're aiming at is not the bar on the help page today.

The Hollywood Reporter documented a genuine bidding war between Whatnot and TikTok Shop over live auction sellers, and buried in it is the most encouraging number I saw all week. One seller said TikTok pushed a World Cup jersey auction to roughly 40,000 viewers on an account with about 13,000 followers. TikTok reps reportedly met with another seller repeatedly while he weighed Whatnot's 30-cent transaction fee against the alternative. Two caveats worth stating: these are seller-supplied figures, and gross merchandise sales are emphatically not creator take-home. But the mechanic is real and it cuts against everything else in this issue in a useful way. In live commerce the audience that shows up and buys is decoupled from the audience you accumulated, and it can be decoupled upward. Your follower count is not the ceiling on your revenue in a format where distribution is algorithmic and the transaction happens inside the stream. That's the good version of having two numbers.

What the platforms shipped this week and why you should care.

Patreon announced roughly 30 new or revamped features on August 20, spanning discovery, short-form Clips and Quips, topic-based Niches, live Q&As, fan profiles, and a rebuilt earnings and membership analytics layer. Availability varies a lot by feature, and "30 tools" is doing some work there: it bundles shipped, rolling out, in test, and planned.

Why this matters for creators: the direction is what's worth reading, not the count. Patreon spent a decade as the place people went after they'd already been discovered somewhere else, and this release is a serious attempt to become a discovery surface in its own right. Practically, the thing to do this week is open your Patreon analytics and look at the new membership and earnings breakdowns before you touch anything else, because if Niches and discovery actually work you'll want a baseline from before they turned on. Also go set your Niche yourself rather than letting it get inferred. Categorization you didn't choose is how you end up in front of the wrong audience.

Instagram rolled out First Draft on iPhone on August 25. Select your clips and it trims them, strips the dead air and pauses, and hands back a starting cut you can then edit. Instagram says the first pass can take under 10 seconds.

Why this matters for creators: this is aimed squarely at the formats where editing is pure tax, so talking-head, multi-clip, day-in-the-life. If that's you, the honest move is to test it on a video you were going to make anyway and time yourself, both the edit and the result's performance, rather than switching your whole workflow on the promise. The thing I'd watch for: auto-trimming pauses is an aesthetic decision, not a neutral one, and it will make your Reels sound like everybody else's Reels. If your pacing is part of why people watch you, cut manually and let it handle the B-roll.

As of August 20, Twitch's Sponsorship Dashboard is open to Affiliates globally, not just Partners. Affiliates can now be included in brand campaigns they qualify for and run the whole thing end to end in one place. There's a gate: you have to be Creator Sponsorship Certified. Twitch's reasoning is the interesting part, which is that in recent campaigns more than half of the participating paid streamers were already Affiliates.

Why this matters for creators: go get certified this week, before you need it. That's the entire action item. Campaign inclusion is criteria-based, so the certification is the difference between being eligible when something relevant comes through and finding out about it after it's allocated. If you've been streaming as an Affiliate and assuming brand money starts at Partner, that assumption expired on the 20th.

Before you say yes to a sponsorship, price the second number. It takes about ten minutes and you only have to do the math once.

Here's the thing about the Higgsfield situation. Neither of those creators made an obviously stupid decision. They got offered real money by a real company in the hottest category in the industry. What they almost certainly didn't do is put a dollar figure on the other side of the ledger, because most of us never have, because how would you.

So here's how I'd do it. Open your analytics and get your average engaged views per video over the last 90 days (engaged views, specifically, not the new public number, since that's the one that behaves like an audience). Then take your typical sponsorship fee and divide it by that, and you have your revenue per engaged view. Call it your RPEV. For a lot of mid-size creators this lands somewhere between half a cent and three cents.

Now the part nobody does. Estimate what a deal costs you in audience. Not "will people unsubscribe," which is unanswerable, but "if this deal cost me 1% of my engaged audience permanently, what's that worth?" Multiply 1% of your engaged views by your RPEV, then multiply by however many sponsored videos you expect to run in the next two years. That's the number. For a creator doing 200,000 engaged views a video at two cents, 1% is 2,000 views, times 40 future sponsored videos, and you're looking at roughly $1,600 of future revenue riding on a single yes. Suddenly a $3,000 deal that makes 1% of your audience think differently about you is close to a wash, and a deal that risks 3% is straightforwardly a bad trade no matter what the check says.

Two refinements that matter. First, when part of your compensation is credits, product, or equity in the sponsor, your real cash fee is only the cash. Price the deal on the cash and treat everything else as zero, because that's what it's worth on the day your audience decides how they feel about it. Second, the risk percentage is not the same for every category. A tool your audience already uses is near zero. A tool that competes with the skill you teach is the high end, and you should assume the top of your range, not the middle.

None of this makes the decision for you and it isn't supposed to. What it does is stop you from comparing a real, specific number (the fee) against a vague feeling (the vibes), which is a comparison the fee wins every single time. Write your RPEV on a sticky note. Next time an offer comes in, do the ten minutes before you answer the email, not after.

That's all for this week. Go find your second number, in YouTube Analytics, in your reorder rate, in your reply-to-follower ratio, wherever yours lives. It's the one that was telling you the truth the whole time. And if someone forwarded this to you, sign up to get your own issue every Thursday.