The feature list
Sync, share, export, integrate. A list of features sells to someone who already wants the product; everyone else needs to see it do its job, once, fast.
Ad creative for apps and SaaSInstalls that never activate start in the ad.
Consumer apps and self-serve software, sold through a trial or a sign-up. A caveat first: trials take longer to read than sales, so the weekly loop runs a step slower here, and we say so before you buy.
What we see in app and software ads.
Software ads tend to lose people in the first seconds, before the useful part appears.
Value that arrives too late
The ad opens on a logo, a tagline or a slow pan over the interface. The moment the product solves the problem is at second eight, and most viewers left at second two.
Installs that cost more than they're worth
An ad that sells a benefit the product does not deliver on the first screen earns installs that never activate. Cheap installs are the expensive kind.
What we'd make first
Lead with the screen, inside three seconds.
The first frame is the product doing its job: the result on screen before the viewer can scroll. Then a before and after, or a carousel that answers one objection per card. Several demos run side by side, and the first screen that earns attention is the one we build on.
- Speed-to-value demo, motion: the first result on screen in three seconds
- Before and after: the spreadsheet or the mess, then the app
- Objection carousel: price, time to set up, what happens to my data, one card each
- Founder to camera: the problem you built it to solve, in a sentence
A first drop for an app, ad by ad.
Growth's thirty ads a month give several first screens a fair test. The first week looks like this.
- First result on screen, motionspeed to value
- Before and after, staticcontrast
- Objection carousel, staticobjection handling
- First result on screen, motion: speed to value (the job done before second three)
- Before and after, static: contrast (the old way beside the app)
- Objection carousel, static: objection handling (price, set-up, data, one per card)
- Founder to camera, UGC-style: authority (why it exists, from the person who built it)
- One feature, one moment, motion: specificity (the single thing a user does most)
- Free trial, no card, static: loss aversion (nothing to lose by trying)
- What users say, static: social proof (real quotes, with permission)
- The Sunday-night problem, UGC-style: problem first (the moment they go looking for you)
Illustration: the shape of a first drop, not a client's ads. Every ad is remade to your brief.
The numbers we watch for an app.
Cost per activated user, not cost per install. Hook rate tells us whether a first screen earns attention; the activation rate behind each ad tells us whether it earned the right attention. An ad with cheap installs and poor activation is retired, whatever its click-through rate says.
Which activation depends on what you sell. A consumer app is judged first on cost per install, then on the share of those installs that reach the moment the app is for: the first workout logged, the first photo edited. Self-serve B2B software has no install, so the first number is cost per trial or sign-up, and activation is the first real piece of work done in the product, with a paid plan following weeks later. The same demo can win one test and lose the other, so we never borrow an app's verdict for a SaaS product or the reverse.
Because activation takes days and a trial takes weeks, we read these numbers over two to four weeks rather than one, and we agree the metric and its window on the kickoff call. That is the slower loop we mention above, and it is why apps are a second fit for us rather than the first. The platforms face the same arithmetic: Meta's help center explains it. Meta's help center says an ad set is Learning limited when it is unlikely to get around 50 optimization events in the week after its last significant edit, and that the status is not a penalty. A trial event that fires a handful of times a week gives the system, and us, little to read in seven days.
What an app or SaaS drop holds, and what stays yours.
Real screens, in every size
Every screen in the ads is recorded from the test account you give us, so nothing on screen is mocked up. Each ad comes in 9:16 for Reels, Stories and TikTok and in 4:5 and 1:1 for the feed, edited to make sense with the sound off, because an interface demo is usually watched in silence.
Approved by you, owned by you
Nothing runs until you approve it, and any ad that misses the brief is remade free within seven days of delivery. Once paid for, you hold a license to run each ad on any channel with no time limit, so a demo that activates users can also play on your homepage. Your growth team or media buyer runs the campaigns; we read the results you share. Each week a three-line note says what we are testing; on Growth and Scale a monthly video report follows, and Scale adds a quarterly review.
How it works for an app or SaaS product.
The same Monday drop, read over a longer window so a good first screen gets a fair hearing.
Agree the event that counts
On the kickoff call we agree what activation means for your product and how many weeks to read it over, and you share a test account.
Several first screens at once
The first drop, within seven days, puts different opening moments side by side: the result, the before and after, the objection.
Read installs or trials, then activation
You run the ads and share cost per install or trial each week, and the activation behind each ad as it comes in.
Build on what activates
The first screen that brings users who stay gets new variants; cheap installs that never activate are retired.
Where most apps start.
Growth: £2,950 a month for 30 ads. Testing which first screen wins needs several distinct demos running at once, which is what thirty ads a month is for. A single-feature app with a small budget can start on Starter and add volume once a demo wins.
Questions we get from apps and saas.
Do you need access to the app?
A test account, yes, so every screen in the ads is real. We record and recreate the interface from it; we never mock up a feature the product does not have.
Can you make app store screenshots or onboarding screens too?
Not as part of a plan: we make ads. Landing page variants that carry a winning ad's hook onto the page are an add-on, quoted in writing.
Why is the loop slower for apps?
A store sees a purchase the same day; an app sees an activation days later and a paid conversion weeks later. We read each ad over a longer window so a good first screen is not retired before its trials convert.
Concept work: every ad in these teardowns is a concept we drew, not a real advertiser's or a client's ad.
More teardowns for this kind of business:
Not your kind of business?
Each of these pages speaks to one kind of customer, in its own words.
Last updated
Send us your ads and a link to the app.
Get my free ad reviewFree. 48 hours. No call needed.