Industry Insights
Subscription advertising should not be judged on installs or trials.
An install, a free-trial start and a paid subscription are three different events worth three different amounts. Accounts often optimise toward whichever one fires most often, and that is usually the install.
What Makes This Industry Different
One account can carry three conversions worth completely different amounts.
A subscription app usually measures an app install, a free-trial start, an in-app subscription and sometimes a website purchase inside the same account. The install fires far more often than the rest and costs the least, so it dominates any conversion set it shares with them.
Smart Bidding and Meta's optimisation both follow volume. Left on the default conversion set, the account keeps buying users who install and never pay, and the reported CPA looks healthy the whole time.
Whether the account works comes down to install-to-paid, and that only becomes readable once a cohort has finished its trial and passed the first billing date. Until then CPI is the only number available, and CPI does not answer the question.
What we usually see
What we look at in a subscription account
Installs, trials and subscriptions usually share one conversion column
The install fires far more often than the subscription, so an account CPA is mostly an install cost under another name. We separate the events, mark only the paid ones as primary, and give each one a value the business agrees with.
Install-to-paid decides whether the account works
Two campaigns can buy installs at the same cost and end up with completely different numbers of paying users. We report installs, trial starts and paid subscriptions campaign by campaign, so a cheap install source with no paid conversions stops receiving budget.
A free-trial cohort cannot be judged inside the trial period
A trial period plus a billing delay means this week's spend only shows its real cost per paying user weeks later. We set the review window to the trial length plus the billing lag, and hold campaign changes until a cohort has run through it.
The store listing caps what any campaign can deliver
Everything after the ad click happens on a page the ad account cannot change: the icon, the first two screenshots, the rating and the recent reviews. We look at the store conversion rate before recommending more install budget, because a listing problem raises the cost of every campaign at once.
App events and website purchases sit in two different systems
In-app events land in an SDK or a measurement partner while web purchases land in GA4 and the pixel, so no single report shows a cost per paying customer. We reconcile the two before setting a budget split between app and web campaigns.
At a Glance
Kick Ads manages paid media for subscription apps and digital services across Google Ads and Meta Ads. Our work can include the conversion set and event values, app campaign and Search structure, free-trial and subscription tracking, store listing review, and reporting on install-to-paid alongside install volume.
Diagnostics
What we check first
Conversion set
Which events are primary, which are secondary, and what value each one carries
Install-to-paid
Install, trial start, paid subscription and cancellation, broken down by campaign and source
Store listing
Icon, screenshots, description, rating and recent reviews, plus the store conversion rate
Tracking
SDK or measurement partner setup, GA4 and pixel on the web side, and whether the two can be read together
Channels
The campaign mix we usually run for subscription apps
Before they search
Meta Awareness
SupportBuilds the audience the install and trial campaigns later draw on, particularly around a feature launch
While they are searching
App campaigns for installs
CoreVolume comes from here, which is exactly why the campaign has to be optimised toward a paid event rather than the install itself
Brand and category Search
CoreCatches people looking for the app by name, and people looking for what it does
Meta conversion campaigns for trials
CoreOptimised on the trial start, then checked against how many of those trials convert to paid
While they are comparing
App re-engagement
SupportReaches people who installed and stopped, which is usually cheaper than buying another install
What the results tell us
Install-to-paid tracking
Installs and trials are easy to buy. The number that decides the budget is how many of them start paying.
The calendar
The subscription calendar we plan against
Subscription demand is driven by moments when people decide to change something, and by the discount windows the platforms have trained everyone to wait for.
- Jan
New-year resolutions. Fitness, learning and habit apps see their highest intake of the year.
We plan the annual-plan offer for January rather than a monthly one, because the intent to commit is highest here and lowest in March.
- Feb to Apr
The churn window. January's signups reach the end of their first paid period.
We watch install-to-paid retention here rather than install volume, because this is where January's numbers get their real result.
- Jul to Aug
Summer. Usage patterns change for entertainment and learning apps in opposite directions.
We do not run one budget shape across both. Which way the account moves depends on what the app is for.
- Sep
Term starts. Learning, productivity and study apps pick up again.
We reopen the campaigns paused over summer instead of leaving the summer setup running into the autumn.
- Nov to Dec
11.11 and Black Friday. Annual-plan discounting is now expected rather than optional.
We plan the discount depth against lifetime value before the window, because a deep annual discount is a decision about next year's revenue.
This calendar comes from running these accounts, not from a measurement. Chinese New Year moves between January and February, so the dates around it shift each year.
Common problems
- Optimising to app installs by default
- One conversion action holding installs, trials and subscriptions
- Judging a cohort before the trial period has finished
- No in-app event values, so every conversion counts the same
- Adding install budget while the store listing is the bottleneck
- App and web campaigns reported separately and never compared
- Leaving cancellations and refunds out of the CPA calculation
Related services
Other Industries
Explore other industry insights
FAQ
Common questions
It depends on how much paid volume the account has. Installs can be a reasonable starting target while data is thin, and the account should move to a trial or subscription event once there are enough of them each week for bidding to learn from.
Long enough for a cohort to finish its free trial and pass the first billing date. CPA read before that point will always flatter the cheapest install source.
Yes. We work with whatever SDK or measurement partner is already in place, and focus on which events are sent back to Google and Meta and what value each one carries.
Yes. A subscription bought on a website has the same problem in a different place. A free signup, a trial and a paid plan are separate events, and the account usually optimises to the easiest one.
Want to know what your app budget is actually buying?
Send us the account and the current conversion setup. We can look at the conversion set, install-to-paid by source, the store listing and tracking before recommending what to change.
Talk to us