Thinking · Newsletter
Hook, gameplay, CTA is right. Here is what it leaves out.
Upptic's 2026 creative framework, where we agree, and the two variables it stops short of: the end card, where most clicks happen, and the destination, which has to keep the promise.
TL;DR. Upptic's "Best Ad Creative Strategy for 2026" opens with a line we would put on the wall: "Most teams don't have a creative shortage. They have a learning problem." Its framework is hook, show the experience, give a reason to act; test one or two variables at a time; define the metric before launch; let tests bake; do not kill winners on a schedule; keep a portfolio of proven winners, variations and wild bets. All of it is right and all of it is about the video. It stops at the end of the video, and that is where two of the biggest variables in mobile UA live: the end card, where most of the clicks happen, and the destination the click lands on. Here is the framework, where we agree, and the two steps it leaves out.
Who said it
Upptic is a mobile UA agency; the video is its own creative team on its own channel, published in 2026. We compete with them on some accounts. The framework is good and we are not going to pretend otherwise.
What it gets right
"Stop making more ads." The opening provocation. "More creative doesn't necessarily mean better performance... they're stuck because they don't have a system that turns every test into the smartest next move." Correct. Volume without a read is noise. The nuance we would add is that the networks now require volume to learn from (AppLovin's own June 2026 data put the average top campaign at 182 active creatives), so the choice is not volume or system. It is a system that produces volume on purpose, which is what a concept system is for. See creative operations, defined.
Hook, experience, reason to act. "The first two to three seconds determine whether someone keeps watching or keeps scrolling... The hook isn't entirely a marketing message. It's the pattern interruption that earns attention." Then "why should I play?", communicated as a feeling rather than a feature list, then a CTA tied to the payoff: "Can you beat this level?" beats "Play now." We would only sharpen the first part: a hook is a promise, not an interrupt. An interrupt buys attention the ad may not be able to spend. A promise buys attention with a specific question attached. The distinction matters for what you test next. See a hook is a promise.
"Don't fake your gameplay." "You might win the install, but you'll lose the trust." Agreed, and the industry is in the middle of a live experiment on exactly this point: a mechanic that ran for a year as a fake-ad minigame has just launched as a real game. Our read is in the genre teardowns.
Three questions per test. "What are we testing? ... Test one or two variables at a time. How are we measuring success? Define your metrics before you launch... Let things bake. Stop opening the oven. What happens after the test? Winners won't last forever, but they should multiply." That is a testing discipline most teams do not have, and the "multiply" step is the one most often skipped: decompose the winner, find the mechanism, build the next batch from it.
Do not refresh on a schedule. "Yes, we're looking at you, Halloween ad, still running in April with killer ROAS." Performance says when an ad is done, not the calendar. Especially true on AppLovin, whose own guidance says not to pause creatives.
The portfolio. Proven winners, variations of what works, and new ideas looking for the next breakout, with the honest line that "nine out of 10 ads won't become winners."
What it leaves out: the two steps after the video
The end card. On rewarded and interstitial placements the video ends and a card appears. By the account of advertisers and platform people alike, most of the clicks happen there, not during the video. In most libraries the end card is one asset, made once, never tested. That is a variable with as much variance as the hook, and almost nobody is running the experiment. The practical version: three cards per concept, varying the headline and the button, with the same video. Method in end cards: the guide.
The destination. The video makes a promise and the store page, or for commerce brands the landing page, either repeats it or changes the subject. A fail-state puzzle ad that lands on a store listing full of character art has broken its own promise. For games the fix is store-listing variants matched to the creative concept. For brands it is a presell page per promise. Either way, the ad and the page are one test, and a framework that stops at the CTA has measured half of it. Method in landing page vs product page for paid ads.
The metric question, one step further
Upptic lists "IPM, CPIs, ROAS, retention quality" as candidates. We would be stricter: judge the test on what the account is for (D7 ROAS, retention-weighted installs, or purchases for a brand), and use IPM and CTR only to explain the result afterwards. A library optimised on IPM recruits installers. A library optimised on D7 recruits players. The hidden-object genre is the clearest case of ads that win on the first metric and lose on the second. Our note on this is measure creative on purchases, not on the ad.
Where we land
Take the framework. Add two rows to the test plan: the card after the video and the page after the click. Judge the result on the outcome, not the click. That is the whole difference between a creative strategy and a creative operation.
Questions people ask
What is the hook, gameplay, CTA framework? Upptic's three-part structure for a mobile game ad: earn attention in the first two to three seconds, show the feeling of playing rather than a list of features, and give a reason to act that connects to the payoff.
How many variables should a creative test change? One or two, per Upptic's framework and ours. Change everything and you cannot tell what drove the result.
When should you kill a winning ad? When performance says so, not on a refresh schedule. Rising CPI, falling CTR and IPM, and declining ROAS are the signals.
What does the framework miss? The end card, which collects most of the clicks on rewarded placements and is rarely tested, and the destination, which has to repeat the promise the ad made.
Source: Upptic, "Best Ad Creative Strategy for 2026 (Hook, Gameplay, CTA)", YouTube, 2026. Quotes transcribed from the recording; interpretation is Flixr's.