How to tell if paid acquisition is profitable

1 The bottom line

Paid acquisition is profitable when your cost per acquisition sits below a break-even you calculated from what each customer pays back. Without that number, cheap installs and fast growth can hide a steady loss.

2 Key takeaways

• Calculate break-even CPA first, then judge every channel against it. A cost means little until you know the line it must stay under.
  • A $10 CPA against a $40 break-even is out of the norm George, Wrestle AI, Brawler and AdMax14:43.
• Judge the quality of customers, not the cheapness of the trial. Cheap trials often convert badly further down the funnel.
  • Someone 65 plus might cost 100% more per trial but convert much better Marcus Burke42:50.
• Track how fast cash comes back, because fast payback makes scaling safe. Money that returns quickly can be spent again.
  • Visible stayed cash flow positive through its growth and reached about $7M revenue before its Series A Luke Martin-Fuller, Visible43:00.
• Fix trial-to-paid and retention before you buy traffic. Paid traffic converts worse than organic, so weak economics show up fast.
  • Built With Science then took Meta from zero to $100,000 a month in ad spend profitably Ethan Ethier, Built With Science17:22.
  • One app made about $1 revenue per install on Meta against about $3 on a healthier app Steve Young, App Masters34:15.
• Scale when a test shows a clear multiple, and stop when the channel does not pay. Conviction should come from numbers, not hope.
  • Halo AI took a $100k personal loan after early math looked strong, then reached roughly 50% margin Dillion Verma, Halo AI51:02.
  • Coconote tried paid ads and never became first-purchase profitable Brett Bauman and Zack Hargett, Coconote35:00.

Making sure that you put a dollar in, you get a dollar out that day will keep you a really efficient growth machine.

Luke Martin-Fuller, Visible

2.1 Get the product ready before you spend

Paid traffic converts worse than organic, so lift conversion and retention first.

• Raise trial-to-paid and retention before you add paid channels. Source: Ethan Ethier, Built With Science17:22
  • Organic traffic is warm and converts well, so paid traffic will always convert lower.
  • Improving trial-to-paid and retention lifts the floor for every channel you add later.
  • They launched in January 2025 and ran no Meta ads until January 2026, spending the year on the product.
• Learn your revenue per install from organic and Apple Ads before you try Meta. Source: Steve Young, App Masters34:15
  • He wants a $10 to $15 cost per trial, or about $3 revenue per install, before considering Meta.
  • Many apps with big MRR only break even on ad spend.
  • His app made about $3,000 a month organically, and a hard paywall did not fix the weak Meta numbers.
  • An acquired app doing $500,000 a year has about $3 revenue per install, which is why he plans to add Meta there.
• Start paid only when you can own it. Source: Brett Bauman and Zack Hargett, Coconote35:00
  • Ads are hard to halfass, and an agency has weaker incentives than you.
  • If you do not know where your customers hang out online, ads are a bad first tool.
  • Neither founder spent half their time on ads. Clips cut by an outside agency from raw creator footage beat their organic-style creative, but ads still did not work.
• Match organic or paid to your ticket price, novelty and time. Source: Yannis, Superwall20:45
  • Organic suits impulse installs, novelty apps and low budgets, but it trades dollars for your own time.
  • If you have budget, run some paid early to see whether the product converts.
  • He has seen novelty apps overspend on ads with too low a ticket, and teams go reposter without the skills to manage creative.

2.2 Calculate break-even and payback

Put a number on what you can afford to pay, and on how soon the money returns.

• Compute your break-even CPA and watch whether your CPA rises or falls. Source: George, Wrestle AI, Brawler and AdMax14:43
  • Healthy CPAs vary by platform and by app, so use your own break-even as the line.
  • On Meta, a $24 CPA would be near 2x ROAS.
  • AdMax averages about $15 per purchase on TikTok.
• Use retention to show that spend pays back. Source: Jacob Eiting and David Barnard, RevenueCat33:30
  • With median annual retention near 30%, a $1M spend in March 2026 returns about $300,000 of free cash flow in March 2027.
  • Apps with a first-year payback profit on spend and compound from there.
  • Churned users remain a list to market to, which makes subscription apps a good target for growth financing.
• Pull payment forward to shorten payback. Source: Michael Segal and Mark Ungerer, Skylight29:47
  • For a software subscription, push a roughly $60 to $70 annual payment up front to fix CPA and payback.
  • Skylight has paid back every order in its history on day zero because the device is sold up front. It is bootstrapped and profitable, with no VC needed for marketing.

2.3 Check the quality of what you buy

A low cost per trial means little if those users never pay.

• Break cost per trial down by age against trial-to-paid. Source: Marcus Burke42:50
  • Collect age during onboarding so you can see trial-to-paid by age bucket.
  • Map each bucket against cost per target, and do not let Meta just buy the cheapest trials.
  • Young traffic gives cheap trials but converts worst. In his example chart, 25 to 34 was the most expensive bucket.
• Watch who fast Meta growth brings in. Source: Melissa Cash, Pok Pok30:40
  • Meta can grow quickly but bring in customers who are a worse fit.
  • Pok Pok now invests in brand and partnerships to bring in more organic and better customers.
  • That includes a collab with Calm and a brand department started about six months before the interview.
• Pick channels knowing whether they capture demand or create it. Source: Dmitri, YourMove AI20:15
  • Google ads and SEO capture latent demand from people already searching, but are capped by search volume.
  • Meta and TikTok create induced demand with more upside, but you compete with every app that could monetize that user.
  • Your app must monetize as well as those other apps to win the auction.

2.4 Track it while it runs

You will never get perfect attribution, so aim for enough truth to place the next dollar.

• Keep marketing cash ROI positive as you scale. Source: Luke Martin-Fuller, Visible43:00
  • Visible uses one acquisition metric, essentially ROAS: a dollar in, a dollar out that day.
  • It depends on the blend of annual and monthly plans, how much you spend and how fast you scale.
  • Roughly 50% of its acquisition comes via paid, and the rest through word of mouth.
• Test cannibalization by switching campaigns off in separate geos. Source: Yoav Sharon, Simply11:11
  • Simply turned off app campaigns in one geo and web campaigns in another, then watched organic traffic and cross-app effects.
  • His view is that it is never about perfect attribution. The goal is enough truth in CAC modeling to decide where the next dollar goes.
• Find winning creators by matching post timing to download spikes. Source: Evan Yadegari, Locked17:50
  • Instagram does not allow direct link attribution, so compare each post against download and revenue spikes.
  • If a creator is profitable, renew the same deal, then renegotiate for a bigger one.
• Cap paid search at a ceiling, even when you rank first. Source: Tanuj Chatterjee, Super Unlimited24:00
  • Run search ads only on selected keywords and markets, watch ROAS carefully, and watch for cannibalization of organic traffic.
  • On iOS they are number one for the VPN keyword in 67 of 71 tracked countries and number two in the rest. Android is weaker, number one only in the US, so extra push goes where rank is not yet first.

2.5 Decide whether to spend more or stop

Scale when the numbers show a multiple, and stop when the economics or the channel do not fit.

• Scale hard once a test shows a clear multiple on spend. Source: Dillion Verma, Halo AI51:02
  • Post videos and measure. One video at about a million views showed constant conversion.
  • Early math showed LTV to CAC or ROAS of about 5 to 1 or 6 to 1, which gave conviction to put everything into creator payments.
  • He ran at break even for four to five weeks before margin appeared.
• Scale a novel idea before the cheap-download window closes. Source: Rafael Kramer, Superwall16:01
  • TikTok ads for the debloat app cost a couple of cents per download at first.
  • That saturated within days, then stayed profitable at about 30% profit on ad spend for the next few weeks.
  • He thinks they should have scaled it more before selling, since copycats arrived.
• Do not lean on paid ads when revenue per user is low. Source: Jay, Nomad Table36:05
  • With a much lower revenue per user, even a very cheap cost per download is hard to justify.
  • Paid may help at the very start to solve the cold start, but not as a long-term strategy.
  • Word of mouth already brings 10 to 15% of new users, and the rest is almost all UGC. High-ARPU apps such as betting are a different case.
• Give a channel enough time before you call it a failure. Source: Dmitri, YourMove AI25:00
  • No channel can be spun up in two weeks. SEO takes about 3 months of intentional work before anything shows.
  • Invest enough that a failure means the channel is not viable, not that execution was poor.
  • People who do not know what they are doing give up after 3 to 6 months. His flyers in Washington Square Park did not work well.

2.6 Let earned media lower your paid cost

Viral moments and launches make every paid dollar work harder.

• Fund top-of-funnel views, because going viral cuts paid CAC. Source: George, Wrestle AI, Brawler and AdMax34:24
  • After the Rork Max launch got 10 million views on Twitter, Facebook paid CAC fell by about a quarter for a month.
  • That led them to run clipping campaigns for AdMax, using slideshow formats at about a 30 cent CPM. They accept a loss on the clips to improve paid metrics.
• Treat earned media as an asset that compounds. Source: Jack McDermott and Tanmay Jain, ElevenLabs16:27
  • Each launch makes the next one land bigger and fills branded search, which is where they start paid first.
  • People who see ads are more likely to convert because they already know the brand.
  • Tanmay says they have not measured it, but is confident it lowers CPA and CPI.
• Run launches and steady paid as two parallel tracks. Source: Jack McDermott and Tanmay Jain, ElevenLabs13:30
  • Plan viral upside moments and also a forecastable month-to-month paid engine.
  • For paid, pick the channel, product and message, and model it as a sustainable channel rather than a launch tool.
  • Launch cadence is roughly monthly per team, with paid on Meta, TikTok, Apple Search Ads, Google Play and Google Ads.

2.7 Where they disagree

How fast payback must be depends on your brand

	Visible	Skylight	ElevenLabs
Payback target	$1 back same day	Day zero if humanly possible	Not needed in week one
How	Single ROAS metric	Pull payment forward	Accept elevated CAC while testing
Relies on	Annual/monthly plan blend	Upfront payment	Brand strength, earned media
Suits	Cash-flow-positive growth	Hardware + subscription	Strong brands

The right payback speed, test size and start date depend on your product and your brand.

• How fast payback must be. Visible wants a dollar back the same day, and Skylight29:47 says to do whatever you humanly can to pay back on day zero.
  • ElevenLabs does not need to nail week-one payback, because brand strength lets it test creatives and accept elevated CAC while it finds the right payback Jack McDermott and Tanmay Jain, ElevenLabs15:30.
  • Slower payback fits teams with strong earned media. Without one, the same-day rule is safer.
• When to start paid. Yannis20:45 says run some paid early to see if the product converts.
  • Ethan Ethier, Built With Science17:22 waited a full year, and Coconote35:00 says not to start until you own the channel.
  • An early paid test reads your product's conversion. A full scale-up is the step the others say to delay.
• How much to test before judging a channel. Evan Yadegari, Locked18:40 says $500 on a creator either wins or teaches you something.
  • Dmitri, YourMove AI25:00 says SEO needs about 3 months before anything shows.
  • Small bets suit creators, where one post gives a result. Slow channels like SEO need longer.