TikTok ads tactics

Rank candidates by comment sentiment (e.g. 'what is that app?') first, then engagement rate, then average views. Spark when engagement is above ~7.5-10% or a video does 3-5x the account's average views. Let organic climb first: if still gaining (e.g. 100K/day) wait 1-2 days. Start with ~$20 (SideShift runs $50-100/day per video for a week); keep running while CPMs don't rise; CPMs usually rise when engagement falls from ~15% to ~5%, then kill it. Exclude sparked views from creator view bonuses. Can add an app-install anchor to track CPI. Spark can also be targeted by country/language/age.

Level 1: every video in its own ad set with $10-20; shut off after $5-10 if clearly off-target; after 1-2 days and $20-40 of promising spend it graduates. Level 2: ad sets (or a CBO campaign in parallel) with 4-5 videos and $100-500, tested against the same lower-funnel or download targets at bigger spend. Level 3: the 1-2 that rise to the top become evergreen campaigns taking ~90% of spend. Never turn off working level 1/2 ads, so one video may run at all three levels at different spend.

Dec 25 is dead. Dec 26-28: start testing (e.g. $5-10K/day) to find the best ~10 videos, expecting ~2x better performance on Jan 1-2. Dec 28-30: conversion quality jumps, raise daily spend across videos. If Jan 1 falls on a Thursday/Friday, expect a smaller jump; the peak is the first Sunday (Saturday night into Sunday morning), when you 10x spend if early signals are good. Year-round, Sunday-Tuesday are the best days for health/fitness/productivity ads and the 1st-3rd of each month converts subscriptions better; time launches accordingly.

Audit starts in Meta Events Manager / Google Ads Goals, not Ads Manager. For the last 7 or 30 days, compare total installs, free trials and paid subscriptions the platform received (not attributed ones) against RevenueCat, Amplitude or Mixpanel before sending. ~5% gap is fine; above 10% must be fixed (change source, parameters or the event itself). Watch for pitfalls: double sources that Meta deduplicates but Events Manager shows twice; missing parameters so only opted-in (IDFA) users are attributed (one change left only 20% of conversions reported). Once fixed, it can run 6-12 months but must be rechecked.

When trial-to-paid from paid traffic trails other sources or its own past, run a regression on non-converting trials. Hard factors: device, OS version, language. Soft factors: onboarding answers (stated goal, intent, e.g. 'just browsing'). Keep the raw trial event unchanged; engineers add a new event (e.g. mapped as 'subscribe' or 'purchase') fired via the MMP (Adjust, AppsFlyer, Singular) or platform SDK only when a trial meets the criteria, and campaigns optimize to that. Most predictive factors across apps: age, goals, and consumer vs prosumer/solopreneur vs business. Advanced: add onboarding questions designed to create this variance. Exclusion is a last resort; skip it if trial quality is already fine.

Make many organic creator videos (multiple creators, multiple hook styles), then put the platform minimum behind each one rather than judging by organic views, since good videos sometimes flop on timing or the algorithm. TikTok: ~20 creatives at the $20 minimum each, can switch off after ~$5; Meta: 30-40 creatives at ~$1 each since it is easier to duplicate. Judge within hours on top-of-funnel clicks only: most get 0-2 clicks, one gets 5-20. Keep spending on that one to see whether it was a fluke, then track it down to purchases and make variations (change the first or last second). Mid-virality videos (top 25%) often convert better than the most viral, clickbaity ones. For a $5K starting budget he'd spread it over as many hypotheses as possible, aiming for 100 variations, and compress '500 videos' of learning into a week. Test the same creative across Google, YouTube, Meta and TikTok; one platform can give 2x results.

Every TikTok ad click goes to the 15-question web quiz rather than the App Store. Pixels fire on every question and answer combination, and answers that indicate a 'white-hot' user (already working out, not casual beginners, the most expensive and worst-converting lead) are used as TikTok conversion events, so the algorithm learns from early on-web signals instead of in-app events it can't see. Answers flow into Amplitude and a Google Sheet, which the founder read daily while iterating. Knows the cost per answer that is 'in the money' and what it means downstream.

Short AI-generated UGC ads built in MaxFusion with three parts: a hook that stops the scroll (take ideas from current TikTok trends), a plain explanation of the use case, and a call to action to download. Example 1: an AI creator says 'Everyone keeps asking how people make those viral AI character videos. You don't need editing skills... open it, pick a character, type what you want them to say, hit generate... Download Cantina.' Example 2 (more obviously AI): an 'emotional support brick' that got a product launch video made in Cantina, ending 'Try it. Download Cantina.' The point is installs, so no cinematic production; cheap to make, so many angles can be tested on Meta.

Scroll TikTok in your niche, find videos that already went viral organically and remake them yourself. He ran five formats, one post each (five ads total), with about $50 of spend per ad. Keep ads running if they convert or are only slightly unprofitable, since early on the ad account needs signal to learn whom to target (lose money on the account, not on the app); stop ads that are very unprofitable. Decide up front how much you are willing to lose to find a winner.

Turn on goal-based budget increase with a target cost per acquisition ('I want customers at price X'). TikTok then controls spend: instead of manual increases every few days, it raises the budget automatically for as long as the ads hit the target CPA.

Hook is a third-party clip (e.g. a convoluted AI-looking workout); it cuts to the creator saying 'This is just ridiculous. If you just download this app, it'll give you the exact workouts you need'. In the top ad the hook runs ~9 seconds before he talks; aim to land the CTA within 15 seconds. Same format on Wrestle AI: Cayden stitched a coach calling out moves. Made by AbMax's one in-house creator (Calvin); George and Lino run them.

Use the 'How did you hear about us?' onboarding answer (or the Superwall + Appstack integration) to route users: answers like Instagram, Facebook, TikTok or other paid channels get a hard paywall with no X; 'searched the App Store' gets a soft paywall with a close X. Every source still sees a paywall. Aim: avoid 1-star reviews from organic users.