How to add ads, affiliates and partners

1 The bottom line

Add ad, affiliate and partner income only where it gives users something useful, and test each step against retention. Most users never pay, so this income matters, but overdoing it destroys trust.

2 Key takeaways

• Give users who won't pay an opt-in way to earn you money. Most installs never convert, so subscriptions alone leave them unmonetized.
  • Steve P. Young, App Masters7:58 says on average 80 to 90% of installs will not pay.
  • David Cremades, From the Bench36:18 puts payers at 2 to 4%, leaving about 95% to monetize with video ads at natural moments.
• Make ads rewarded and place them at moments of gain. Users who ask for the ad feel helped, not interrupted.
  • Matthew Hall, Crossy Road15:25 has no banners or interstitials, only opt-in video for 20 coins, and Hipster Whale earns more from ads than from in-app purchases.
  • Cem Kansu, Duolingo8:28 shows ads only after a lesson, never before or during.
• Pitch affiliates and partners whose offer fits what your users already do. Pay-on-sale deals carry little risk for the brand and no cost for the user.
  • Yves Benchimol, WeWard22:11 approached brands directly and works with Nike, Adidas, Expedia, Booking.com and Amazon.
  • Jake Castillo, Cal AI1:06:27 pays creators $10 once per signup that results in a sale.
• Leave some money on the table to protect trust. Lost trust costs more than the revenue you squeezed out.
  • Tanuj Chatterjee, Super Unlimited20:18 would rather leave 10% of money on the table than bombard people with ads.
• Measure every ad change with tests and complaint counts. Numbers show real backlash and keep one angry review from driving decisions.
  • Cem Kansu, Duolingo10:24 found in 2016 that complaints about missing Finnish were 3x the complaints about ads.

There's a lot of trust that you destroy with overdoing monetization.

Tanuj Chatterjee, Super Unlimited

2.1 Set your trust rules first

Decide what you will not do before you add any revenue line.

• Make "do no harm" the baseline, then add offers that give real value. Source: Giordano Contestabile, Life36043:40
  • Ads must not be intrusive, invasive or unwelcome. Judge this by attention, engagement and user conversations.
  • Use location and context for useful offers. The Uber partnership grew from a landing notification feature built in a hackathon.
  • Life360 started ads only in the last 2 years, in a company about 15 years old.
• Hold back on how often and how widely you monetize. Source: Tanuj Chatterjee, Super Unlimited20:18
  • Give paywalls a big visible X to close. Show no ad on the very first install, so people see the value first.
  • Showing an eSIM ad too broadly taught the team to target only people who will actually benefit.
• Block ads and partners that clash with your promise. Sources: Cem Kansu, Duolingo8:28, Yves Benchimol, WeWard20:37
  • Duolingo blocks gambling, sketchy games and anything not young-audience or brand safe, even at a revenue cost. It also asks networks to drop formats like ads with three X buttons.
  • WeWard rejects brands whose effect on how people act is the opposite of its better-lifestyle mission.
  • Dogfood the product so you feel what each placement does to the experience Cem Kansu, Duolingo8:28.
• Check that your product has a natural place for ads at all. Sources: Peter Sellis, Snap and Discord20:49, Francesco Polizzi, Discord41:05
  • Snap opens to the camera, a creation surface with no natural ad spot. Peter calls his failure to drive camera monetization one of his biggest at Snap.
  • Discord has few inventory spots because most activity is among two to four people. It moved to reward ads called Quests, where users earn orbs for profile customization.
  • Francesco gives the Quests view as an outsider.

2.2 Launch ads simply

In a new app, use one network, learn which formats fit, and keep retention first.

• Put AdMob in and do not overthink it. Source: Steve P. Young, App Masters11:00
  • Ad revenue won't be your biggest winner at first. A banner is fine if it is not in the way.
  • Skip mediation and multiple ad calls at launch. With small audiences, networks need time and volume to learn, and Sid says none beats AdMob on scale or eCPM at this stage Sid, AppBroda7:07.
  • Rafael Kramer9:27 used AdMob alone for a viral game, with an ad after every second death and a 30 second rewarded ad to revive. It earned roughly 10 cents per user and drove all the revenue behind a $100k month.
• Make retention the main goal until the app is stable. Source: Sid, AppBroda2:01
  • Hold off on monetization until retention is in place and the app is fully policy compliant.
• Run one placement of each ad format. Source: Sid, AppBroda8:38
  • Try interstitial, banner, rewarded and native, one variation each. Keep what works for your users.
  • Each format drives a very different kind of revenue and user experience.
• Protect your ad account while you test. Source: Sid Gupta, AdMob3:11
  • Use test ads for every internal session. Taps on live ads from one IP or user can look like invalid activity, and Google may suspend the account.
  • Cap ads per user at one or two with Firebase, and cut frequency sharply during the 30 day review period Sid Gupta, AdMob5:02.

2.3 Add rewarded ads at moments of gain

Let users trade a short ad for something they want, at the point they want it.

• Offer a rewarded ad to users who dismiss the paywall. Source: Steve P. Young, App Masters6:43
  • Cleanup shows onboarding, then a paywall. When the user hits X, they can watch a rewarded ad to delete a batch of photos. Payers get the full product and non-payers still earn revenue.
  • This suits utility and one-time-use apps where you already paid to acquire the user Steve P. Young, App Masters7:58.
  • Tolan tests classic game-style ads on part of its free tier, such as lapsed subscribers who watch ads to keep chatting. The aim is to learn how users react and what ARPU ads produce Ajay Mehta, Tolan26:40.
• Let players double any reward by watching a video. Sources: Steve Young, Fun Run 38:02, Tom Hammond, TheoremReach10:41
  • Fun Run 3 offers 200 daily bonus coins for a tap or 400 for a quick video. Put the option front and center on the home screen.
  • Players see earned coins as theirs, so doubling feels easy. For valuable rewards like Merge Dragons dragons, charge gems to double, which adds purchase monetization.
  • Merge shows 173 coins earned after a lost puzzle and offers to double them Steve P. Young, Duolingo8:35.
  • A 2,000 coin welcome bonus with an optional watch-to-double raised both downloads and ad revenue each time a client game ran it Steve P. Young, App Masters3:49.
• Place offers right after a win, and use many capped placements. Sources: Steve P. Young, Duolingo6:20, David Cremades, From the Bench34:52
  • Show the offer after a level or lesson ends, framed as refilling for the next one. This is Young's hunch, and he has no data to confirm it.
  • Set the cap per placement, not per app. An energy refill video can be watched twice a day, while a pre-match booster has its own cap.
  • Fun Run 3 has video offers on the main menu, daily bonus, arena entry and post-race chest Steve Young, Fun Run 311:31.
  • From the Bench found more users watching fewer videos each earns more than fewer users watching many.
  • A save-me prompt that trades an ad for a continue works best when the player was close to a personal best Steve P. Young, Subway Surfers10:49.
• Give users several ways to earn and let them choose. Source: Tom Hamon, TheoremReach6:05
  • Offer surveys, rewarded video and offer walls side by side. Some users love surveys and others prefer videos, and choice makes them feel empowered.
  • Layer these on top of ads and purchases, not in place of them Tom Hamon, TheoremReach5:10.
  • An indie game with about 200 daily active users added roughly $1,000 a month from surveys, and it took nothing from ads or in-app purchases.

2.4 Add affiliates and partners

Earn from brands and partners by sending them customers, and pay your own promoters only on results.

• Pitch affiliate deals to brands directly, and sell things your users already need. Source: Yves Benchimol, WeWard22:11
  • Affiliate deals are performance based, with tracking already in place, so brands carry little risk. WeWard approached affiliate platforms and brands directly, and the wellness setting makes brands happy to be associated.
  • One app added an Amazon affiliate link in settings for a physical product related to its function Steve P. Young, App Masters16:28.
  • A friend's travel app with millions of downloads sold tours and guides and doubled its top line revenues Chad Mureta, App Empire23:40.
  • Superpower vets supplements and sells them below anywhere else, making trivial money. Purchases feed its AI and doctors, and members who start using the marketplace keep using it Max Marchione, Superpower35:13.
• Pay your own affiliates a flat, one-time amount per paid signup. Source: Jake Castillo, Cal AI1:06:27
  • Cal AI pays $10 for every signup that results in a sale, once. This aligns incentives and avoids long-running revenue obligations that could complicate selling the app.
  • The $10 is roughly 33% of a $30 sale.
• Swap promotion with niche stores through their order confirmation emails. Source: Steve P. Young, App Masters7:56
  • For a yoga app, list yoga stores as partners in your app, and they list you in their confirmation email. Test plain text against banners near the shipping information.
  • Hire a VA to find store emails and send cold outreach with a Loom video. Pay about $5 on Fiverr to edit their email template, and use a tracked Branch link with a custom product page for each store.
  • Offer their customers a deal like 25% off the first year or 60 days free Steve P. Young, Cal AI7:00.
  • Order confirmation emails get about a 65% open rate.
• Take a revenue share from platforms and content partners. Sources: Tibo Sottiaux, OpenAI12:57, Steve P. Young, Gamezop9:18
  • OpenAI plans to pay popular, high-usage plugins part of the revenue when subscribers spend their usage inside them. There are 16 launch partners for sign-in with ChatGPT.
  • Embedding third-party games, quizzes, horoscopes or cricket news lets ads inside that content earn money shared with your app. Gamezop offers 60% through the App Masters link, versus a lower standard share.
  • Partnerships also bring subscribers. Many of Calm's came from partnerships, including B2B deals and unexpected ones like Amex Nick Candler, Calm30:12.

2.5 Test and tune

Measure backlash, frequency and ad quality with real data, then tune one change at a time.

• Count complaints per million daily active users before reacting to backlash. Source: Cem Kansu, Duolingo10:24
  • Compare ad complaints with other complaint topics. Pair this with retention, app closes and review trends from A/B tests.
  • Retention was neutral to slightly down in most Duolingo ad tests.
• A/B test how often each ad shows. Sources: David Cremades, From the Bench32:56, Eado Hofstetter, Fyber7:54
  • Test how many interstitials to show per user and the daily cap for rewarded videos. Caps let users watch ads without farming coins.
  • Test banner refresh by country. A 10 second refresh in one market and 11 seconds in another can both be right.
  • Developers became more engaged in ad work once they saw revenue rise from changing the ad network alone David Cremades, From the Bench32:56.
• Separate match rate from show rate, then fix show rate by band. Sources: Sid Gupta, Appbroda1:55, Sid Gupta, Appbroda3:04
  • Match rate is the share of requests Google has an ad for. Show rate is the share of returned ads you actually display. With 20 requests and 10 filled, match rate is 50%, and if 2 are shown, show rate is 20%.
  • Under 30% is worth fixing now. From 30 to 50% run long experiments when you have time. Above 50%, look for gains elsewhere.
  • Slice show rate by format, platform, app version, ad type and ad source to find the cause Sid Gupta, Appbroda5:25.
  • Add error code listeners to log why ads fail Sid Gupta, Appbroda6:39.
• Once you have user history, build a baseline and test mediation. Sources: Sid Gupta, AdMob2:26, Eado Hofstetter, Fyber5:17
  • Segment historical data by ad format, device and country. AppLovin, ironSource or InMobi can be key in the United States but have far less demand in India.
  • Split test waterfall against bidding and check whether ARPDAU actually rises.
  • Sid has seen many apps where re-enabling competitor ads for about 10% of the audience raised eCPM with little impact on retention. If retention holds, open it up further Sid, AppBroda2:07.

2.6 Where they disagree

Flat fees suit sellable apps, revenue share suits platforms

	Flat fee (Cal AI)	Revenue share (OpenAI)
Payout	Once per paid signup	Ongoing share of revenue
Obligation	Ends at signup	Runs over time
Best for	App that may be sold	Platform growing builders
Upside	Simple, aligned incentives	Distribution becomes builder income

Advice splits on how much to annoy users, whether banners belong, and when to move past a single network.

• Annoying users to push payment can raise revenue, but it risks trust.
  • One indie developer showed a mandatory video ad of about two minutes to anyone who dismissed the paywall, and ad revenue rose 443% Josh with Steve P. Young, App Masters9:17.
  • Steve also suggests a mildly annoying banner so people want to remove it Steve P. Young, Notepad3:21. Moving the remove-ads banner onto the page most users see was one of the changes that took an app from $1 to $17 a day Steve P. Young, App Masters7:40.
  • Super Unlimited and Duolingo go the other way, with big visible X buttons and ads only after a lesson Tanuj Chatterjee, Super Unlimited20:18, Cem Kansu, Duolingo8:28. The annoyance tactics came from one-function utility apps.
• Banners: a good earner or a thing to avoid.
  • For an app people open daily, free with banners can beat a paid download. Ketchapp told Steve banners are the number two ad unit by revenue Steve P. Young and Anna, Caveman7:42.
  • Crossy Road runs no banners or interstitials because many people dislike ads Matthew Hall, Crossy Road15:25. Steve says video ads earn more than banners Steve P. Young, Duolingo2:40.
• Single network or mediation, waterfall or bidding.
  • At launch, skip mediation Sid, AppBroda7:07. In the growth phase, focus on partners who can do bidding, because querying about 10 partners at once keeps latency low Sid, AppBroda6:25.
  • For mature apps, run bidding and waterfall together, because not all partners offer bidding Sid and Steve P. Young, Aproda5:49. Fyber says to test rather than assume bidding wins.