case study 00 ✨ the one that went everywhere
658.3K views on a six-second video ✨ $0 ad spend
insight: a love island usa storyline was peaking and every fan was catching up on a flood of reposts at once, the shared, slightly overwhelmed feeling was the content.
idea: a 6-second reaction video built around a single on-screen hook.
the hook: ‘when u finally catching up on trinity’s reposts but there’s a lot so u gotta lock tf in’
hypothesis: if the hook lands in the first second and the clip loops, the for you feed will carry it to non-followers.
execution: filmed and posted native on @valleyof_thegolden, timed to the fandom moment. no boost, no spend.
results: 658.3K views · 521.4K unique viewers · 158K likes · 2,593 saves · 468 shares · 199 comments · 100% non-followers · 81.3% for you traffic · average watch 7.2 seconds on a 6-second video (it looped) · likes peaked at 0:01 · 1,346 total hours watched · 172 new followers.
what i learned: the first second is the whole ad. a loopable format turns a trending conversation into watch time, and watch time is the receipt.
case study 01 ✨ creative testing that found a winner
one variable changed: 27.9K views on an account under 1K followers
insight: successful tiktok shop affiliates were answering one question in public, ‘what products made you money, and how much?’, and the answers were getting saved like research.
idea: run that exact format on my testing account, @valleygirl4lyfe, and treat the comments as the deliverable.
the hook: ‘what’s ur best seller n how much u make? i’m nosey’
hypothesis: the format works, but the overlay wording decides who stops, so i changed only the wording between versions.
execution: three iterations. versions one and two were close. version three changed exactly one variable, the overlay text, and i seeded the comments by answering my own question first.
results: 27.9k views · 1,421 likes · 290 comments · 563 saves · 101 shares, on an account under 1k followers.
what i learned: one variable at a time, or you learn nothing. the comments doubled as free product research from successful affiliates, and aspiring affiliates saved it and stayed. one post, two audiences, both served on purpose.
case study 02 ✨ commerce creative that sells
287.1K organic views that sold product
insight: product videos that open with the pitch get skipped. entertainment earns the click before the ask.
idea: a 12-second demo where the product is the payoff, not the opening.
the creative: the Octobuddy MagSafe mount, filmed in-app, posted native, product linked in TikTok Shop.
hypothesis: one product, one use case, zero fluff would out-convert a feature list.
results: 287.1k views · 3,593 likes · 396 saves · 125 shares · 100% organic, and it drove tiktok shop sales and commission.
what i iterated: the video now anchors a cross-platform system, the same creative reposts to instagram and facebook with amazon affiliate links, so one filming session monetizes three platforms.
the commerce receipt ✨ tiktok shop, october 2026: $427.38 attributed gmv · 23 items sold · 110.1k product impressions · 5.6k product clicks (+147% vs the month before). octobuddy alone did $113.89 gmv on 6 units that month, still selling six months after the video posted.
case study 03 ✨ content with a job to do
the nonprofit video that turned views into sign-ups
insight: moms trust moms. for a kids’ music therapy nonprofit, a share from another parent beats any targeting setting.
idea: one video with one job, get parents to the sign-up form, with the next step impossible to miss.
execution: the ‘sign up here!’ video for heart & soul music therapy, whose social i run end to end.
results: the page’s most-viewed video of the last 90 days, 1,364 views, 1,100 unique viewers, and its most-shared post (16 shares), with clicks through to the sign-up form.
what i learned: give the right audience something worth passing on and distribution takes care of itself. same instincts, different conversion.
the lab ✨ my testing system
most of my main-feed winners started as experiments in the lab: my testing account @valleygirl4lyfe. the loop never changes: 1) watch ✨ comments, saves, and other creators’ winners are the research. 2) hypothesize ✨ name the variable before i film. 3) test ✨ run it in the lab, change one thing at a time. 4) grade ✨ saves, comments, and watch time decide, not vibes. 5) graduate or kill ✨ winners move to the main feed and become series; losers get buried with the lesson attached. that’s the whole system: strategy you can watch happen in public.
what i do, in agency words
research the brand, the audience, and the competitors → write the hooks, scripts, and storyboards → produce the creative, my own, or ai-generated under my direction → review the cuts and send what doesn’t pass back for another take → read the results and brief the next round. performance ugc is the format i already live in: short, native, hook-first, graded on data.