Follow-up (5 working days later)
Hi {first name},
Bumping this in case it got buried. I'd like to sponsor a segment for {app}, and I'm flexible on format and price.
If it's not a fit, a one-word "no" helps me too.
Thanks,
WalterStudy and test-prep creators who could put your apps in front of the exact students using them, sorted by exam, ranked on how likely a sponsorship is to pay for itself. Start here to see the best few for each exam.
Loading…
Open the CPA or ACT list, keep the Pitch first and Strong fit rows with an email, and choose ten. Watch two of each one's recent videos before writing. A first line about a real video is the difference between a reply and silence.
Send the generated email with the opener rewritten. Mark them Contacted. Five working days later, one short follow-up, then Followed up. After that, stop: a creator who wants the deal will answer the second email.
Never pay on subscribers. Ask for their rate, compare it with the ballpark on the row, and buy one integration from the creator with the best views-per-dollar. Measure it before buying a second.
Give each creator their own link and code (a UTM link to the site, an App Store campaign link for iOS). After 30 days you know installs and paid users per dollar. Rebook what paid; drop what didn't.
Run this before agreeing a price. If the break-even view count is above the creator’s typical views, the price is too high, however good the fit.
The defaults are typical for an integrated read in an education niche: 0.5–1.5% of viewers click, a third to a half of those install, and 5–10% of installs pay. Replace them with your own numbers after the first spot. Revenue per paying user is what one subscriber pays you over their whole study period, after store fees.
Price on a creator’s median views, never their subscriber count. These are the market ranges for small and mid-size education creators in 2026, and they are the basis of the ballpark on every row.
| Format | Per 1,000 views | Typical floor | Notes |
|---|---|---|---|
| Long-form YouTube integration (60–90 s) | $15–$35 | $100–$250 | Best for apps. The video keeps getting found in search, and the link stays in the description. |
| Dedicated long-form video | 2–3× an integration | $300+ | Only with a creator whose audience already loves them; too expensive for a first test. |
| YouTube Short | $5–$15 | $50–$100 | Cheap reach; no clickable link in the video. |
| TikTok video | $5–$15 | $50–$150 | Ask for the median views of their last 10 posts; follower counts on TikTok are especially misleading. |
| Instagram Reel + story link | $5–$12 | $50–$150 | The story link is the part that converts. Ask for one with the reel. |
The first email is generated on each creator’s row, with their name, a recent video, and the right format filled in. These are for what comes after.
Hi {first name},
Bumping this in case it got buried. I'd like to sponsor a segment for {app}, and I'm flexible on format and price.
If it's not a fit, a one-word "no" helps me too.
Thanks,
WalterHey {first name}! I built {app}, a {exam} prep app, and your {exam} videos are exactly what my users watch. Could I sponsor a video? Paid, and you'd try it free first. What's the best email for rates?Thanks {first name}, that's helpful.
Your recent videos are averaging about {median views} views, so for a first test I'd like to offer {offer} for one {format}, plus {commission} per paid user from your link for 90 days. If it performs I'd like to book more.
Does that work? I can send a short brief and the app login today.Hi {first name},
You're earlier in your channel than most creators I sponsor, so rather than a flat fee I'd like to offer a commission: {commission} for every paid user who signs up from your link, for as long as they stay subscribed. Free account, your own code, and a monthly report.
Interested?There is no single database of creators, and the useful numbers (how many people actually watch) are not the ones platforms put first. Every platform was asked what it permits before anything was read.
| Source | Status | What it gives, or why not |
|---|---|---|
| YouTube channel pages | Read | Subscribers, video count, the last 30 uploads with their views and age, the Shorts tab, the channel description and header link. robots.txt allows /@handle pages; search, RSS feeds and the internal API are disallowed and are not touched. |
| YouTube "View email address" button | Not read | It sits behind a captcha precisely so it is not collected automatically. When a row says "About tab / DM", open the channel's About tab yourself: many creators have a business email there. |
| TikTok | Not read | tiktok.com/robots.txt disallows Claude's user agents by name. Handles come from links the creator publishes elsewhere; follower counts come from public reporting, dated, and are marked as estimates. |
| Not read | Instagram prohibits automated collection without written permission. Handles are recorded the same way as TikTok's. | |
| Link-in-bio pages (Linktree and similar) | Read | Linktree's robots.txt allows profile pages to Claude-User, which is how this tool identifies itself. Emails and social handles found there are used. |
| Creators' own websites | Read | Home, contact and about pages only, after checking robots.txt. An email is kept only if the site publishes it. |
npm run discover: starting from a few known videos, it reads each video’s related-video list, keeps the ones whose titles match the exam, and counts which channels keep appearing. A channel recommended next to a dozen ACT videos is an ACT channel whether or not any list article knows it.npm run probe and read by hand before going into data/creators.mjs. Channels that only mention an exam once, lifestyle vlogs, and channels that are really about something else were left out.0 creators in total; 0 measured on YouTube; 0 TikTok/Instagram-only.
Out of 100, separately for each exam, answering one question: if this creator said yes, would the sponsorship pay for itself? Every row shows the sentences behind its own number.
| Component | Max | The question |
|---|---|---|
| Exam fit | 30 | Is the channel about this exam (30), shared with one or two others (26), one subject of many (18), an adjacent topic like admissions or accounting careers (10), or general study (4)? |
| Real reach | 30 | Median views of recent posts on their main platform, times the share of the audience likely to be studying for this exam, on a log scale against the size of that exam's market: full marks at 10K relevant views for the CPA and LSAT, 15K for GMAT and GRE, 20K for MCAT, 40K for the ACT and 50K for the SAT. TikTok and Instagram are estimated at a tenth of followers and discounted. |
| Format | 10 | Long-form YouTube (10) beats Shorts and TikTok (6) and Reels (5) for an app: a minute of demo and a lasting link. |
| Activity | 10 | Days since the last long-form upload: two weeks or less scores full marks; over a year scores nothing. |
| Reachability | 10 | A published email (10), a website or link-in-bio (6), or only a DM (3). |
| Free to say yes | 10 | An independent creator (10), a tutor (6), someone who sells competing prep (2), a prep company (0). |
A sum lets strengths hide a fatal weakness, so some things cap the score however well the rest goes.
| At most 40 | Prep companies. They sell what you sell. |
| At most 45 | No long-form upload in a year, or the channel says it is inactive. Also: under 300 views a post. |
| At most 50 | General study audiences, for any one exam. |
| At most 55 | No view or follower count could be found at all. |
| At most 60 | Under 1,000 views a post. |
| At most 65 | Adjacent audiences: admissions, careers, premed life. |
Unknown is not zero: a number that cannot be read scores the neutral middle and says so, rather than burying a creator who simply is not measurable from here.
npm run build:data # re-read every channel (cached for a day) and re-score
npm run build:data -- --fresh # ignore the cache
npm run probe -- @handle ... # measure candidates before adding them
npm run discover -- --match "\bACT\b" --pages 40 @seedchannel
# find new channels from related videosTo add a creator, append them to data/creators.mjs with the exam, kind and any TikTok or Instagram numbers you have seen, then run npm run build:data. Your pipeline notes are stored separately and survive every rebuild.