your ad plays inside a game. make it the one they don't skip.
6 short guides + a script that edits your ads for you. the setup, the 35-second video, the end card, the first 30 days. numbers with sources, not vibes.
35 s median watchtype 200, not 2d7 by defaultnever add to an old set+30% every 3 daysmost clicks happen on screens 2 & 335 s median watchtype 200, not 2d7 by defaultnever add to an old set+30% every 3 daysmost clicks happen on screens 2 & 335 s median watchtype 200, not 2d7 by defaultnever add to an old set+30% every 3 daysmost clicks happen on screens 2 & 335 s median watchtype 200, not 2d7 by defaultnever add to an old set+30% every 3 daysmost clicks happen on screens 2 & 3
real ad · real data
this mascara ad has been running for 533 days.
#1 of 184 active ads for that brand. 68 seconds. no studio. a girl puts it on one eye, stops, shows you the difference, then does the other eye.
nobody pays for an ad for a year and a half unless it sells. that's the format i'd copy first for applovin, where people actually watch 35 seconds.
the bonus guide breaks down 10 ads like this one, second by second.
data: brandsearch, oct 23 2026. we're not affiliated with this brand or with brandsearch.
#1of 184 active ads for that brand
533days live
68 slong
×2versions
695ads on that funnel
ad view · brandsearch · oct 23 2026
why first campaigns die
it's almost never the channel.
01
the "2".
a 2× roas target is typed 200. type 2 and you've asked axon for 2%. it's in the official tutorial, and it's the first thing guide 1 fixes.
02
one video, one image.
the #1 launch mistake according to pep hufen, who runs applovin accounts for dtc brands. the algorithm needs a portfolio to choose from.
03
touching it on day 2.
every practitioner says the same thing: 7 days hands off, unless spend is zero, tracking is broken or an ad got rejected.
04
reading axon like meta.
axon counts clicks only, so it can look worse than your other dashboards. set your target off the wrong number and the model barely spends.
05
a dead end card.
most clicks happen on screens 2 and 3, not during the video. most brands leave those screens on default.
what's inside
6 guides. one job: a campaign that spends.
the 60-minute setup
account, pixel, brand settings
the campaign, screen by screen
roas · universal · d7 and why
12 end-card lines under 75 characters
35 seconds they can't skip
the specs that actually spend
the 35-second skeleton
30 first lines that open on a person
8 scripts with blanks
half your clicks happen here
the 3 screens of every axon ad
brand settings vs dpa
18 end card concepts
a one-week end card test
the clone machine
find ads that have run for months
3 prompts: blueprint, rewrite, hooks
ai scenes with your real product
1 blueprint → 12 ads
scale without breaking it
day −7 to day 30
+30% steps and the next-day dip
two scoreboards that never match
break-even roas in one division
claude motion
describe the edit, get the mp4
re-cut meta winners to 35 s
slow zooms, 5-second loops, captions
motion.py script included
bonus
10 ads that refuse to die
10 shopify brands' ads live for 105 to 533 days
what happens, second by second
why each one lasts
how i'd rebuild each one for applovin
bonus 2
the field reference
16 rules to follow, 16 mistakes to avoid
every contradiction between sources, settled
the 4 checklists: account, creatives, campaign, weekly
the full glossary
guide 06 · claude motion
stop editing. describe it.
tell claude "take my best meta ad, cut it to 35 seconds, hook at the top, captions from the voiceover, end on a dark card". it writes the timeline, the script renders the vertical mp4. twelve variations in one go.
most ads die in a week. a few run for 500 days. i'm interested in the second kind.
applovin is the first channel in years where people actually watch the whole thing. most shopify stores still haven't touched it, or tried it once with a 15-second meta cut and quit.
this is the playbook i wish existed: short, numbers first, every figure credited to the person who said it.
secure checkout · instant download · digital product, all sales final
questions.
what do i get, exactly?
8 pdfs (6 guides + 2 bonuses, 84 pages) and the claude motion files: motion.py, a readme and an example timeline. you get the download link right after checkout and by email.
do i need to know how to code for claude motion?
no. you install 3 free tools once (the guide walks you through it), then you describe the edit to claude in plain english. claude writes the file and runs the render.
is this made by applovin?
no. it's an independent guide, not affiliated with or endorsed by applovin or axon. every number comes from public videos by practitioners and the official axon channel, credited in the guides.
will this work in my country?
per the 2026 sources, axon targets everywhere except western europe, and canada is available. if you only sell in france or germany, this isn't for you yet. check the countries in your own account.
is there a guaranteed result?
no, and be careful with anyone who promises one. results depend on your product, offer, creative and budget. the playbook gives you the setup and the method; the numbers inside are examples and sourced figures, not promises.
can i get a refund?
it's a digital download, so all sales are final. if a link doesn't work or a file is missing, write to us and we'll fix it right away. read the "what's inside" section above before you buy.