Hi friends,
Engagement was the one number everybody agreed on. Platforms paid on it, brands bought on it, creators optimized for it. This week it got demoted from three directions at once.
X announced it is replacing its engagement-based payout with one that rewards original content. Kantar looked at 15,000 sponsored creator posts and found platform engagement lines up with actual brand impact about a third of the time. And Digiday found brands starting their creator searches inside AI chatbots, where engagement is not an input at all.
Three replacements, none of which agree with each other.
Let's get into it.
🐦 X is scrapping the payout model that rewarded engagement farming
X is changing how it pays creators. From Sept 8, the Original Content Rewards Program replaces the revenue share it has run since 2023. The old system paid on volume: clear an impression threshold, hold a premium subscription, get paid, whether you made the video or reposted someone else's. That funded a lot of aggregation and ragebait. The new one is meant to pay for original work, though X has not said how it will judge that.
Buyers were unimpressed. Viral Nation's Mat Micheli said his creators have not mentioned X in years. The brand safety problem is unchanged: Forbes reports nearly 80% of X's trust and safety engineers were cut by 2024, and there is active litigation over sexually explicit imagery made with Grok. One dissent is worth keeping. Greenlight Group's Doug Landers calls X the best place anywhere to organize a fandom, which is a real asset even if the payouts are not. If you sell into sports, finance or gaming, that is the reason to watch Sept 8, not the money.
📎 Digiday
📊 Engagement predicts brand impact in one creator post out of three
Kantar looked at more than 15,000 sponsored creator posts on TikTok, Instagram and YouTube and asked a simple question: do the posts that get the most likes and comments also move the brand? About a third of the time. Among the posts with the strongest engagement, fewer than one in five did much for the brand, and only about one in eight shifted short-term sales.
The useful part is what separated the posts that worked. Saying the brand name out loud was the single biggest factor, and only 49% of posts did it. Product demos worked too, and appeared in just 16%. Creator content beat regular digital ads on how involving and relevant people found it, and lost on whether anyone remembered the brand. None of that needs new budget. It needs three lines in a brief.
🔎 Brands have started searching for creators inside AI chatbots
Olive oil brand Kosterina says it often begins its search for creator partners in Claude, and agency executives say AI visibility is starting to appear in briefs. This is a different discovery surface from a platform search, because a chatbot does not see follower counts or engagement rates. It sees what has been written about someone, where, and by whom. SponsorCX puts nearly one in five consumers already substituting AI for traditional research.
Which quietly rewrites who gets found. Creator Colin Rocker ran a citation audit and discovered he surfaces through his LinkedIn profile and podcast appearances rather than his social output. Uncomfortable for anyone whose whole footprint is short video, and an opening for mid-tier creators with earned media behind them. For brands the caution is simple. Sourcing partners through a chatbot means inheriting whatever that model ingested, which skews toward whoever has been written about rather than whoever performs. Use it to build a longlist. Do not let it build your shortlist.
📎 Digiday
🎤 Seller Spotlight
The creator with 267,000 followers and $23 million in sales
Brandon Hans runs @be.lush on TikTok Shop. He is the platform's second-highest independent creator by all-time GMV, with more than $23 million in sales and this year's Home Creator of the Year award. His following is about 267,000, which is not a number he spends much time on.
He started by selling roughly $2 million of books in about four months from one publisher's catalog, before anyone at that publisher knew who he was. He messaged them the day after the first sale to introduce himself. When competitors copied the format he rebuilt across home appliances, kitchen tools, automotive, fashion and sports gear, on the logic that selling in any category was the only proof the business would survive one.
@be.lush Ninja CrushBOSS Kitchen System #foodprocessor #blender #ninjakitchen #crushboss #tiktokshopcreatorpicks
Here is the part worth your attention. Ask how he picks products and the product comes up last. His three criteria are inventory depth, ad spend, and the seller relationship. Inventory because a viral video can clear stock in a day. Ad spend because GMV Max judges a video inside its first thousand views and organic-only wins have largely stopped. Relationship because at his volume, knowing a seller's promo calendar changes how he reads his own analytics. He is scoring you before you score him.
He also works on commission rather than brand deals by choice, capping deals at a handful of videos a month. Commission pays for results, placement pays for placement, and he learned early that thirty videos for a flat fee produces thirty mediocre videos. A mid-size following moved $23 million, and it moved on whether someone stopped scrolling, not whether they liked the post.
🧠 Brand Move of the Week
BetterHelp turned the comments section into an acquisition channel
Last November, BetterHelp's social team left a comment on a viral TikTok of college students performing a cappella. Four words riffing on Pitch Perfect. It got more than 300,000 likes, and the team realized something was there.
What they did next is the actual story. Rather than chasing another viral moment, they built a repeatable process. Social and community manager Kerrie Ngo now comments on 200 to 350 posts a month, of which 10 to 20 clear 10,000 likes. In July, 19 passed that mark. Chief growth officer Sara Brooks says daily new followers went from about four to about 450, against 537,400 TikTok followers and more than 635,000 on Instagram.
The guardrails are what make it scalable. The team screens conversations against two questions: would a therapist weigh in on this, and would you say it to a friend. And they never sell. Brooks is explicit that the strategy is deliberately not direct response, because pitching would break the trust the whole approach is built on. The unglamorous cost line is Ngo spending roughly 20 hours a week on TikTok, which is most of a job.
Three things transfer. The output is one person and a set of rules, not a budget. The restraint is the mechanism, not a nicety, since a comment that sells reads as an ad and dies. And it has started producing creator partnership conversations as a byproduct, so the cheapest top-of-funnel play in this issue is also a sourcing channel. Brooks thinks it is available to most brands. She is right, and the constraint is not money. It is whether you can leave a sales line on the table 300 times a month.
🛠️ From Our Desk
The highest-GMV creator in our last affiliate campaign only had a 3% engagement rate
FIT (Future Impact Tracking) Score
Three percent is not bad. That is exactly the problem.
Nobody cuts a creator at 3%. She lands mid-roster, gets a standard brief and a standard product allocation, and the campaign gets built around the people whose numbers look exciting. That is not a dramatic mistake. It is the ordinary invisible kind. Not a creator you lost, a creator you underused.
We expected a lot from her, and we said so before the campaign ran, because FIT (Future Impact Tracking) scored her near the top of the roster. She then produced more tracked GMV than anyone else in it.
What the FIT score shows us isn’t just how many likes or comments she gets on her posts - it dives into intent. People ask where to buy things, ask questions about products or mention that they’ve purchased using her code for example.
The distinction is understanding what her audiences are saying and how they interact (including saves and shares) which gives you a better indication of performance.
So we built FIT around two questions instead of one. Do they get attention, and do they drive sales? The attention side still uses engagement rate, weighted heavily: shares, saves, likes and comments should be weighted differently based on what we are trying to predict. Also, comparing against accounts of the same size. ER was never the problem. Comparing a 5K account to a 500K one was.
The sales side reads the comments under sponsored posts and counts the buying language, one comment at a time, because that is where intent shows up before revenue does.
What FIT is not: a scoreboard. Tracked revenue is the scoreboard and we have not seen a serious argument otherwise. FIT is a shortlisting tool, built so allocation runs on evidence instead of on a number that only tells you whether someone clears a bar everyone clears.
We have been refining the score for the past 9 months and it’s proving to be more and more useful.
If you’re interested, reply with a few handles. We will run them and send back the scores, the comments underneath them, and an honest "not enough data" where that is the truth. Alternatively, create your account here and we’ll demo the platform.
📌 Other stories worth your attention
Institutional money is buying creator channels outright. Forbes profiled Electrify Video Partners, a London firm founded in 2021 that takes majority equity stakes in established YouTube channels and then staffs them up. It owns at least nine, including control of Derek Muller's Veritasium and its 21 million subscribers, bought in 2023. Chairman Tim Shey's pitch to advertisers is longevity: a channel with a proven track record can sign multi-year commitments in a way an individual creator cannot. Worth reading alongside OpenAI's purchase of TBPN and HubSpot's of Futurepedia. If audiences can be bought, sponsorship stops being the only way in. 📎 Forbes
US commerce media spend is forecast to hit $142 billion by 2030. EMARKETER has it growing from $83.71 billion this year, but the growth rate decelerates from 22.2% to 10.1% over the period. Amazon still holds over 75% of retail media, which is pushing everyone else toward travel, financial services and cross-retailer plays. The number to sit with: advertisers working with nine or more networks went from 8% in 2023 to 33% in 2025. Complexity is the tax on this channel. 📎 Net Influencer
TikTok is building peer-to-peer payments into direct messages. Bloomberg found the code in the US iPhone app. No launch timeline, and it builds on TikTok Pay infrastructure already live in Vietnam, Malaysia and Thailand. Early, but the direction is a closed loop: consumers have spent $2.9 billion inside the app this year, and US users now spend more on TikTok than on YouTube, Facebook or Instagram. A wallet is what turns a shopping surface into a bank. 📎 Net Influencer
Ok, see you next week. In the meantime, Sell! Sell! Sell! 🫡