TikTok Creators Sell A-B Testing Data as Brand Research

Lucas: You know, we've talked a lot about creators selling finished content — videos, sounds, thumbnails — but there's another layer that's quietly becoming a big deal: the data from what they tested and didn't end up posting. Luna: Wait, you mean like the 'failed' experiments? The videos that flopped in the drafts? Lucas: Exactly. A creator might film three different hooks for the same product review, post the one that worked best, but the other two still generated real data — click-through rates, completion rates, comment sentiment. And that data is gold for brands. Luna: So instead of just paying for the final video, the brand pays for the insights from the tests. That's smart. Do you have an example? Lucas: Sure. I spoke with a beauty creator named Maya — about 150,000 followers — who regularly tests two thumbnails for each sponsored post. She runs them as unlisted videos for 24 hours, tracks the analytics, then deletes the loser. But before she deletes, she saves the data. Luna: So she's basically building a database of what doesn't work. That's fascinating. What kind of numbers are we talking? Lucas: She found that close-up thumbnails — just the product's texture on skin — had a 28% higher click-through rate than full-face shots. That single insight convinced a skincare brand to change their entire ad creative for a quarter. They paid her five thousand dollars for a monthly subscription to her test results. Luna: Five grand a month for data from videos that never even went public? That's a whole new revenue stream. How does she organize the data? Lucas: She uses a simple spreadsheet with columns for date, product category, thumbnail type, hook type, posting time, and then metrics like view-through rate, save rate, share rate. She also notes whether the test was posted publicly or unlisted. Luna: So it's almost like a controlled experiment. Brands love that because it's cleaner than scraping engagement from live posts. Lucas: Right. There's no algorithmic interference from the For You Page timeline. The data is more reliable. And creators are starting to formalize this. I've seen a few building custom dashboards with tools like Notion or Airtable to sell access. Luna: Is this happening at scale, or is it still niche? Lucas: It's early, but growing fast. I talked to a creator agency that now includes a 'testing library' rider in their top creators' contracts. They offer brands a tiered subscription: basic access to aggregate test data, or premium access with direct Q&A about the tests. Luna: That makes sense. The premium tier is basically consulting. So the creator becomes a research partner, not just a content supplier. Lucas: Exactly. And it's not limited to thumbnails. Some test hooks — like 'I tried this for 30 days' vs. 'Dermatologist vs. drugstore' — and they sell the comparison. Others test posting times, or even the length of the video. Luna: What about the ethics? Are creators worried about giving away trade secrets that could help competitors? Lucas: That's a valid concern. Most creators I've talked to anonymize the test data by removing any brand-specific references. So the brand sees 'beauty product' instead of 'XYZ serum'. They also sign NDAs. But the bigger risk is that brands might try to use the data to bypass the creator and run their own similar tests without attribution. Luna: Have you seen that happen? Lucas: A couple of cases. One creator found that a brand used her thumbnail data to run identical tests on their own TikTok account without paying her for the insight. That led to a renegotiation where the brand agreed to a licensing fee for past use and ongoing access. Luna: So the legal side is still catching up. But it sounds like the upside is huge for creators who treat their testing like a product. Lucas: Yeah. And for brands, it's incredibly efficient. Instead of running their own tests from scratch, they're buying proven hypotheses. The beauty brand I mentioned said it cut their creative testing time in half. Luna: I wonder if we'll see something like a 'testing data marketplace' — a place where creators can list their libraries and brands can browse. Lucas: That's already emerging. There's a startup called TestStack that launched in beta last month — it lets creators upload anonymized experiment results and set pricing per category. Brands subscribe to categories they care about. Luna: And I'm guessing the platform takes a cut. Lucas: Fifteen percent. But for creators who don't want to manage sales, it's a decent trade-off. And the platform vets the data quality — they require at least ten tests per category before a creator can sell. Luna: That makes sense. You don't want a brand paying for a single test that might be an outlier. Lucas: Exactly. And now that we're on the subject of value — I think it's a good time to mention that if these kinds of marketing conversations have sparked something you've actually used in your own strategy, well, that's exactly why we keep this podcast ad-free. Luna: Yeah, it's important to us. We believe the content should stand on its own, without sponsor breaks or product placements. Lucas: If you've found value in what we do, the simplest way to keep it going is supporting us at buy me a coffee dot com slash fexingo. No pressure, just if it's made a difference. Luna: And now back to the testing library trend — Lucas, you mentioned TestStack. Are there any other platforms trying this? Lucas: Yes, a couple. One called InsightPool takes a different approach: they don't host the data directly, but they connect brands to creators for one-off custom tests. So a brand pays a creator to run a specific test — say, 'show me how this product performs with a storytelling hook vs. a direct sell hook' — and the creator designs and executes the test, then reports back. Luna: That's more like a research service than a data library. But it's still based on the same principle: the creator's knowledge and audience are the asset. Lucas: Right. And the pricing varies wildly. A micro-creator with 10,000 followers might charge $200 for a custom test, while a macro-creator with a million followers could charge $5,000 or more. Luna: Because the larger audience gives more statistically significant data. Makes sense. Lucas: Exactly. And that brings up an interesting point: should we start thinking of creators as research and development partners? Not just publishers. Luna: I think that's exactly where the industry is heading. Especially as brands get more sophisticated about measuring ROI. They want to know not just 'did this video get views?' but 'why did it get views?' The testing data answers the 'why'. Lucas: And creators who build that muscle — who document their experiments and treat their failures as valuable — they're going to have a massive competitive advantage in negotiations. Luna: So for our listeners who are creators: start tracking everything. Even the tests that flop. Lucas: And for brand marketers: start asking creators about their testing data. You might be sitting on a gold mine of insights that never made it to the feed. Luna: I think we'll see dedicated marketplaces for creator testing data within the next year. It's just too valuable to stay in spreadsheets. Lucas: Agreed. And that wraps up today's episode. Until next time, keep testing.