We spent over 1 million euros on TikTok Ads in 100 days, distributed across Mondly, Motorola, Creative Tim and Blindspot. Here are the 9 lessons that mattered most, with no “best practices” recycled from Twitter.
1. Creative is 80% of performance, the rest is 20%
We tested the same setup on 6 accounts: same audience, same offer, same budget. Only the creative differed. The difference between the worst and the best CPA was 4.2x. Not targeting, not bidding, not landing page. Creative.
2. Volume of creative beats individual quality
At Motorola we ran for months with 3 “perfect” creatives. Average results. Then we switched to 15 creatives/week, lower production value. CPA dropped 34%. Hits come from probability, not genius.
3. Short hooks beat clever hooks
We A/B tested “clever” hooks (4-6 second setup) vs. direct hooks (0-1.5s straight to the point). Direct hooks won 23 out of 25 tests. The audience has no patience, no matter how good the payoff is.
4. Smart Performance Campaign works, but not from zero
At Mondly, SPC beat manual CBO by 18-22% on CPA. But only after the account had 3 months of data and 8+ creatives tested. Launched from zero on a new account, SPC is inferior. The order matters.
5. Conversion API isn’t optional
An account without CAPI loses 15-25% of reported conversions. The algorithm optimizes on incomplete data. We moved two accounts from pixel-only to CAPI + pixel in a week: reported CPA dropped 19% with no other change.
6. Authentic UGC beats studio production, but not always
For DTC and app installs: UGC wins clearly. For premium brands (Motorola), UGC alone undermines perception. Optimal mix for premium: 60% UGC, 40% quality production. For DTC: 85% UGC, 15% branded.
7. Scaling too fast kills good accounts
We lost 40k in a week on an account by raising the budget 100% after a good weekend. The algorithm got lost, CPA tripled. The rule we apply now: max 20%/48h, regardless of performance.
8. Creative fatigue is real and comes fast
A winning creative on TikTok has an average lifespan of 10-14 days before CTR drops 30%+. Without constant refresh, good accounts become bad in 3 weeks. Fixed schedule of 3 new creatives/week, non-negotiable.
9. Competitor research saves you months of learning
At Blindspot we imported 30 angles from TikTok Ad Libraryfrom 5 competitors. Of the first 10 hooks we launched, 6 came from patterns we saw there. 4 scaled. Without research, we’d have discovered them in 2-3 months, with budget burned.
What DIDN’T change in 100 days
Fundamentals: good hook, clear offer, product that solves something real. Without these, any tactic is a delay. You spend 1M on a mediocre product, you get a mediocre CPA and an expensive lesson about PMF.
100 days and 1M+ euros. The most important thing learned: process discipline beats any hack or trend. Repeat weekly: research, production velocity, rotation, disciplined scaling, postmortem. The rest are details.
