Back to Explore
GuideVibeFix editorialUpdated Nov 14, 2026

Can Ads Keep Your Free Vibe Project Alive? In-Product Ad Monetisation That Doesn't Drive Users Away

A free product that charges users nothing has to charge advertisers. This guide does the ad math first, then covers four formats, frequency rules, and privacy lines. All example projects, people, and money figures in this guide are fictional.

A Times Square street scene with a huge digital billboard showing a portrait ad as pedestrians walk below

A-Nan's "South Wind Weather" is a free weather app: check the weather, get rain alerts, put a nice widget on the home screen — all free, no subscription, no in-app purchases. Six months after launch it reached 10,000 daily active users, and the server bill arrives every month like clockwork. What pays it? The few small ad slots inside the app. The week the first ads went live, A-Nan watched the dashboard all night: someone opened the weather details page and a banner flickered at the bottom; someone checked the air quality and an interstitial popped up and was dismissed in a second; someone else willingly watched a 30-second video ad to unlock the 24-hour forecast. Ads are neither a monster nor a money printer — they are one way for a free product to stay alive, provided you do the math first, then decide what ads look like, where they sit, and how many times a day they appear.

Two things to make clear first. First, the walkthrough project "South Wind Weather", the operator "A-Nan", and every time, quantity, and money figure in this guide are fictional, invented only to explain the process. Second, boundaries: this site already published a guide on freemium strategy, which covers how free features lead to paid conversion; and guides on donations and voluntary payment, pricing page design, and referral programs, covering voluntary payment, pricing pages, and referral growth. This guide repeats none of them. It covers one thing only: not charging users, but charging advertisers — how in-product ads work, how far to take them, and when not to do them. Freemium gets some users to pay; ads let other users' attention pay for the product. The two sets of books must be kept separate.

A Times Square street scene with a huge digital billboard showing a portrait ad as pedestrians walk below

1. Do the math first: how much money can ads actually make

Before talking about ad monetisation, take out the calculator. Ad revenue has a one-line formula: revenue = impressions ÷ 1,000 × eCPM (earnings per thousand impressions). Impressions depend on daily active users, how many of them can see an ad slot, and how many times each person sees one per day; eCPM depends on the ad format, the users' region, and what advertisers bid. The formula is simple, but every number in it is a choice: if impressions never grow, a high eCPM is useless; if eCPM is too low, no amount of impressions will pay for the servers.

The fictional example numbers in this guide work out like this: "South Wind Weather" has 10,000 daily active users, 85% of whom see an ad slot at least once a day — 8,500 people. Those 8,500 people see an ad 3 times a day on average, for 25,500 impressions a day. The blended eCPM is $1.20, meaning every 1,000 impressions earn $1.20. Daily revenue is 25,500 ÷ 1,000 × 1.20 = $30.60, and a month (counted as 30 days) is $918. All of these numbers are fictional; the point is not $918 itself but the order of magnitude: a free tool with 10,000 daily users makes under $1,000 a month from ads — comfortably enough for a mid-range server and a domain name, but not enough to pay one full-time person.

Run the math backwards and you get three tiers of judgement. The first tier is "cannot even pay for the servers": under 2,000 daily users, or an eCPM below $0.50, monthly revenue stays under $100, and maintaining ad code costs more than it earns — better to skip ads for now and spend the energy on growing daily users. The second tier is "subsidy": 5,000 to 20,000 daily users, a few hundred to a couple of thousand dollars a month, enough to cover servers, domains, and third-party services so the product stops losing money. The third tier is "material": over 100,000 daily users, or a category with an unusually high eCPM (finance or education users, for example), with monthly revenue in the tens of thousands — only then do ads truly become a business line. Most vibe projects stay in the second tier for life. That is not failure; subsidy is the normal place for ads in a small product.

One more number belongs in the same calculation: the click-through rate. The fictional example numbers in this guide are that of 25,500 daily impressions, a 2% click-through rate means about 510 clicks a day. These numbers are also fictional, and the point is that click-through rate is not the revenue target: ad revenue settles per impression, not per click. Optimising for clicks leads to clickbait ad placements and accidental-tap designs — clicks go up in the short term, and users all leave in the long term. Click-through rate is only useful for judging whether the ads and the users are a good match: a rate stuck below 0.5% for a long time means the advertisers and the users are a bad fit, and the fix is to change the ad source — not to make the button bigger.

2. Three ad formats: banner, interstitial, and rewarded video, with an 8x price gap

More ad slots are not better, and more expensive ones are not better either. The four common formats, from least to most interruptive, are: banner, native, interstitial, and rewarded video. The more interruptive the format, the more advertisers will pay — and the less patience users have. Choosing a format means finding the balance between "earn more" and "drive fewer people away", and the balance differs by product: tools with short, purposeful sessions tolerate little interruption; content products where users linger and scroll tolerate relatively more.

The fictional example eCPM numbers in this guide are: banner ads $0.80 per thousand impressions, native ads $1.60, interstitial ads $2.40, and rewarded video $6.50. All of these numbers are fictional; the point is the relative gap: rewarded video costs advertisers about 8 times what a banner costs, native about 2 times, interstitial about 3 times. The gap is that large because advertisers buy different things: a banner buys "a glance while passing by", while rewarded video buys "a user who willingly watches 30 seconds and confirms". Something a user watches willingly is naturally worth much more.

Where do banners belong? At the edge of the user's sight, where they never block the core action: the bottom of a weather details page, the top or bottom of a list page, the bottom of the settings page. The iron rule for banners: never place one next to a button the user is about to tap, and never in a spot where it suddenly jumps into place while scrolling — clicks from accidental taps are judged invalid traffic by ad platforms, and at best they are not paid for, at worst the account is banned. Where do interstitials belong? At natural breaks between tasks: after checking the weather once, after finishing an article, after finishing a game round — never in the middle of a task. Users who just finished something are relatively relaxed, and that is when an interstitial with a one-tap close is tolerated best.

Rewarded video fits only one scenario: the user actively chooses "watch an ad in exchange for a benefit". For example: "Watch a 30-second video to unlock the 24-hour forecast", or "Watch a video for one more AI outfit suggestion". The benefit must be concrete, delivered immediately, and proportionate to the ad's length: what 30 seconds of video buys should be worth at least 30 seconds of the user's time. A vague benefit ("watch videos to earn points" — with no explanation of what points do), a delayed reward ("rewards arrive within 24 hours"), or a benefit out of proportion to the time (30 seconds of video for a useless badge) means users will never tap it a second time. Rewarded video is the only one of the four formats that users tap on purpose, and its reputation depends on how fast the reward arrives.

Native ads are the easiest format to get wrong: they look like content, mixed into a feed, and users cannot tell which item is content and which is an ad. Native ads must carry a clear "Ad" label, as visible as the content title — not small light-grey text hidden in a corner. Native ads with unclear labels get high short-term click rates and lose trust much faster in the long term: once users realise they were tricked into tapping an ad, their trust in the whole feed drops. The ad label is not decoration; it is the baseline that lets native ads survive long-term.

3. Frequency and interruption: how many times a day belongs in the rules

The most damaging thing about ads is not that they exist, but that they never stop. The same user seeing 3 interstitials a day and seeing 30 a day feels completely different; an app that shows 2 pop-ups in 5 minutes and one that shows 10 has very different retention. Frequency control is not "show a few less by feel" — it is numbers written into the rules: how many times each ad slot may appear per user per day, the minimum gap between two appearances, and whether there is a cooldown after consecutive appearances.

The fictional example frequency rules in this guide are: banners stay on screen permanently with no limit; interstitials appear at most 3 times per user per day, with at least 30 minutes between two appearances; rewarded video has no limit — because users tap it themselves, however many times they watch is their own choice. These numbers are fictional; the point is to "set rules per format": a permanent banner never interrupts the flow, so it needs no limit; an interruptive interstitial needs a cap and a gap; an opt-in rewarded video needs no cap, but it does need fraud protection — if one person watches 20 rewarded videos in an hour to farm rewards, the reward system is drained, so fraud rules must be written together with reward rules.

Beyond frequency rules, leave an exit: a paid option to remove ads. Removing ads is the most natural, least annoying paid feature: users dislike ads but will pay for "no ads". The site's freemium guide already covers paid conversion design as a whole, so this guide adds only one ad-side point: the price of removing ads should line up with the ad revenue. If one user brings in $0.09 of ad revenue a month (a fictional example number, derived from the math above: $918 ÷ 10,000 people), charging $1 a month to remove ads is an 11x markup, and users will feel it is not worth it; charging $0.99 a year, or a one-time $4.99 purchase, makes users feel "this money is well spent". Before pricing, work out how much ad revenue one user brings in per month, then set the remove-ads price so users feel it beats putting up with ads.

One frequency scenario is easy to miss: a new user's first three days. Users who just downloaded the app have no affection for the product yet, and their tolerance for ads is at its lowest. The fictional example approach in this guide is to show no interstitials for the first 3 days after a new user signs up, keeping only the banner at the bottom, and to return to normal frequency from day 4. These numbers are fictional; the point is "let users fall in love with the product first, then show them ads" — the other way round, users will remember "this app has a lot of ads" before they even work out what the product does.

4. Where ads come from: ad networks, direct sales, and the privacy line

With ad slots built, where do the ads come from? For a vibe project, there are realistically only two choices: join an ad network, or find advertisers yourself and sell directly. An ad network hands the slot to a platform (such as Google AdMob for mobile ads), which fills ads automatically and settles per impression. The upside is zero sales cost — connect today and earn tomorrow. The downside is low eCPM, no control over ad content, and the constant risk of an ad that clashes with your product's tone. Direct sales means approaching relevant brands yourself: a weather app talks to an umbrella brand, a sports app talks to a sports drink brand. The upside is high eCPM, relevant ad content, and brands willing to commit long-term. The downside is that sales is itself a full-time job, and one person cannot close many deals.

The test is simple: under 50,000 daily users, do not touch direct sales — advertisers do not care about that scale, and the price you close will not cover the time you spent negotiating. Past 50,000 daily users, you can reserve one slot as a direct-sales experiment: one slot, one brand, one month. If the deal closes, expand it; if not, fall back to network fill. Networks and direct sales are not either-or; the model is "network as the base, direct sales as the markup": the network guarantees the slot always has something to show, and direct sales replaces network fill whenever a brand will pay more, with the two channels backing each other up.

Privacy is the line that must never be blurred when taking ads. An ad network's SDK collects device information for ad targeting, and three things must happen before it goes in. First, read the SDK's privacy statement and confirm what it collects, what it is used for, and where the data lives — do not read only the integration docs and skip the privacy terms. Second, give users a real choice in the app: whether to allow personalised ads, with non-personalised ads shown when they decline. The rules differ by region: users in Europe fall under GDPR consent requirements, users in California under CCPA opt-out requirements, and users in mainland China under the separate-consent requirements of the Personal Information Protection Law — the names and exact requirements follow the local law itself, and this guide does not expand on them. The only point here is that consent for ad targeting is not done by "one sentence in the privacy policy"; there must be a switch users can actually tap in the product. Third, do not run personalised ads on child-related content: products aimed at children, or products with a large share of underage users, must switch off ad targeting — this is a hard rule on most ad platforms and one of the easiest places to get penalised.

Two engineering reminders for SDK integration. First, load ad SDKs lazily: do not initialise every ad SDK at app launch and make startup speed pay for ads; initialise the relevant SDK only when the user reaches the page that holds the slot. Second, a failed ad load must never break the page: an ad request timeout, a fill failure, or an SDK crash must not freeze the user's main flow. Reserve a fixed height for the ad container, and leave it blank or collapse it when loading fails — do not let the page jump around as ads load and fail. Every time the page jumps, users trust the product a little less.

5. Fill rate and ad blocking: an empty ad slot looks worse than no ads

With slots built and a network connected, does every open show an ad? No. That is the "fill rate": of 100 ad requests sent, how many actually return an ad. The fictional example numbers in this guide are that "South Wind Weather" has a banner fill rate of 72%, interstitials at 61%, and rewarded video at 88%. All of these numbers are fictional; the point is that the fill rate is never 100%: niche regions, late-night hours, and newly launched apps all have lower fill rates. When the fill rate is low, an empty ad slot looks worse than having no slot at all — users see a blank area and assume the product is broken.

Low fill rates have three remedies. First, run a waterfall across two or more networks: if the first network does not fill, ask the second automatically, then the third, down the chain until someone fills it. Order the waterfall by eCPM from high to low — do not put the lowest bidder first just to raise the fill rate. Second, prepare fallback content: when no ad arrives, do not leave the slot blank — show your own content instead, such as a new-feature introduction, the paid remove-ads entry, or a recommendation for a sister product. Fallback content is itself a free self-promotion, a hundred times better than blank space. Third, accept reality: low fill rates late at night and in remote regions are normal. Do not lower your ad quality bar to fill the gap and let in vulgar, exaggerated, or even fraudulent ads — one bad ad damages the brand more than a week of empty slots.

Ad blocking is a different matter: some users run ad-blocking tools, or use the system's ad-tracking limits, so ad requests never go out at all. Do not try to "fight back" against these users: popping up a message demanding they turn off blocking, or hiding content behind ads to force them to watch, is fighting your own users. The right approach is to accept that this group generates no ad revenue and treat them as "free users without ads" — they do not watch ads, but they use the product, recommend it to friends, and rate it in the app store. Count ad revenue only from the users who actually see ads, never from everyone.

6. Measuring: four metrics, checked once a week

Once ads are live, watch four metrics, once a week. Checking more often causes anxiety; checking less lets problems grow. The first is eCPM: revenue per thousand impressions, which reflects how efficiently a slot earns. If eCPM suddenly halves, first check whether the ad source changed or whether a region's fill broke — not whether users suddenly dislike ads. The second is fill rate: the share of requests that return an ad, which reflects how completely a slot monetises. A fill rate stuck below 50% for a long time means the slot's position, size, or audience is wrong — change the position or size, rather than adding more slots.

The third is impression frequency: how many times each user sees an ad per day on average. If frequency rises, revenue rises, and retention falls, ads are driving people away; if frequency is flat and revenue falls, fill or eCPM is the problem. The fourth, and most important, is retention before and after ads go live. In the first week after ads launch, day-1 retention falling by 2 percentage points is normal fluctuation; falling by more than 5 points means ads are genuinely driving people away, and you should immediately lower frequency or remove the most interruptive slot. The fictional example numbers in this guide are that in the first week after "South Wind Weather" launched interstitials, day-1 retention fell from 41% to 39% — a drop of 2 percentage points — and the team judged it acceptable and kept the interstitials. These numbers are fictional; the point is to "let retention vote on ads": however good the revenue numbers look, falling retention is a loss — users who leave shrink the pool of impressions, and next month's revenue will only be smaller.

Read the four metrics in one table, never separately. eCPM up, fill rate down, total revenue flat means you are trading fill rate for unit price — neither a gain nor a loss. Frequency up, retention down, revenue up means you are trading retention for revenue — a short-term gain and a long-term loss. eCPM and fill rate both up, retention unchanged, means you found a better ad source — the only kind of "free win". A/B testing can tell you which change genuinely works: split one group of users in half, show one half the new slot and the other half the old one, and compare a week of revenue and retention before deciding. This site already published an A/B testing guide; the splitting and statistics follow that guide's rules and are not repeated here.

7. Anti-patterns: six ad practices that earn complaints

First, stacking pop-ups at launch: a user opens the app, a splash ad appears, and after closing it an interstitial appears — blocked twice before getting in. Choose either a splash ad or an interstitial, never stack them. Second, fake close buttons: a close button made tiny, or one that opens the ad's landing page before closing, is judged as induced clicking by ad platforms — at best unpaid, at worst a banned account — and users will simply uninstall. Third, banners hugging action buttons: a banner pressed against high-frequency buttons like "Save", "Submit", or "Next page" means one shaky hand taps the ad. Accidental-tap rates spike, the ad platform judges it invalid traffic, and both account safety and user experience lose.

Fourth, rewards that never arrive: a rewarded video is watched, but the reward is not delivered, or arrives hours late. Users feel cheated and will never tap a rewarded video again. Rewards must arrive within 5 seconds of the video ending (a fictional example number), with a clear confirmation: "24-hour forecast unlocked". Fifth, uncontrolled ad content: joining a network and never checking what actually appears in the slot, until a user screenshots it and complains "how did this ad get into your app". Check network ads regularly — open your own app at least once a week and look at what is playing in the slot, and block unsuitable categories in the network dashboard. Sixth, treating ad revenue as the only goal: rushing to add ads before reaching 2,000 daily users, squeezing core features into a corner to fit more slots, or refusing to remove ads after retention falls. Ads are one way for a free product to survive, not the purpose of the product. A product needs users first; only then do ads mean anything.

8. Pre-launch checklist

Go through this list item by item before shipping in-product ads. First, do the math: daily users, the share of users who see an ad slot, impressions per user per day, blended eCPM, and monthly revenue — work out all five numbers before deciding whether to take ads at all. With under 2,000 daily users and monthly revenue under $100 (a fictional example threshold), skip ads and grow daily users first. Second, price the four formats: banner eCPM $0.80, native $1.60, interstitial $2.40, rewarded video $6.50 (all fictional example numbers), with rewarded video about 8 times the banner; banners sit at the edge of sight, interstitials at natural task breaks, rewarded video only as "watch an ad for a benefit", and native ads must carry a visible "Ad" label. Third, write frequency into the rules: banners stay permanently, interstitials at most 3 times per user per day with gaps of 30 minutes or more, rewarded video unlimited but fraud-protected; no interstitials for new users in their first 3 days (a fictional example number); keep a paid remove-ads exit, and work out one user's monthly ad revenue before pricing it. Fourth, networks as the base and direct sales as the markup: do not touch direct sales under 50,000 daily users; on privacy, give users a switch for personalised ads, read the SDK's privacy statement, and run no personalised ads on child-related content; load SDKs lazily, and never let a failed load break the page. Fifth, have a fallback for fill rate: banners 72%, interstitials 61%, rewarded video 88% (all fictional example numbers); run a waterfall across two or more networks, fill empty slots with your own content, and never accept vulgar or exaggerated ads; do not fight ad-blocking users — treat them as free users. Sixth, check four metrics weekly: eCPM, fill rate, impression frequency, and retention before and after launch; a day-1 retention drop of up to 2 percentage points in the first week after interstitials launch is acceptable (a fictional example number), and a drop over 5 points means lowering frequency or removing the slot immediately; test changes with A/B testing before rolling them out to everyone.

Six pitfalls, once more for emphasis. First, do not take ads before doing the math: when daily users and eCPM cannot support real revenue, maintaining ad code costs more than it earns. Second, do not put a banner next to an action button: accidental taps are not paid for and carry a ban risk. Third, do not build fake close buttons: users will close an ad they cannot close by uninstalling. Fourth, rewarded-video benefits must arrive immediately: watch it and get nothing, and users will never tap again. Fifth, do not let network ad content run uncontrolled: open your own app once a week and look at the slot, and block unsuitable categories. Sixth, do not trade retention for revenue: a retention drop over 5 percentage points means a loss, however good the revenue looks.

eCPM by ad format (fictional sample): banner $0.80, native $1.60, interstitial $2.40, rewarded video $6.50 — about 8x from lowest to highest

Back to A-Nan at the start of this guide. Three months after ads launched, the books for "South Wind Weather" look like this: 10,000 daily users, 25,500 impressions a day, a blended eCPM of $1.20, $30.60 a day, $918 a month — all fictional example numbers from this guide. The server bill is covered, and A-Nan no longer pays for it out of salary every month. On the users' side: the banner sits quietly at the bottom, interstitials appear at most 3 times a day, rewarded video only opens for people who actively want to unlock the 24-hour forecast, and day-1 retention fell by only 2 percentage points. Ads did not make the product better, and they did not make it worse — they simply let a free product survive without subsidy. For a free product, ads done to this standard are enough. When daily users grow tenfold, then it is time to discuss whether ads are a business line; for now, they are simply one road that keeps the product alive.

Views 4Comments 0

Comments (0)

ME
0/1000
Loading comments...
Browse projectsPublish your project

Related articles

Real photo: a developer pointing at code on a laptop screen, with two more monitors showing code behind it
Guide
When the program crashes, catch the error first: Python exception handling

An exception is not a broken program: it is Python telling you what went wrong, with a name you can catch. Following Chapter 8 of the official Python Tutorial: syntax errors vs exceptions, try/except, matching by inheritance, which errors to catch and which to leave alone, raise, else, finally, with, user-defined exceptions, chaining, and ExceptionGroup — plus a 100-accident breakdown from a fictional bookstore script.

Project BuildingPythonLearning & Career
Real photo: the aisle of an independent bookstore, wooden shelves packed with colorful book spines under warm light
Guide
Turn a book into an object and its rules into a class: Python classes and objects

A class bundles data and its rules together: one Book class gives every book its own state, methods let objects do the work themselves, and inheritance lets an e-book stand on a paper book's shoulders. Following Chapter 9 of the official Python Tutorial: class and __init__, self, methods, class vs instance variables, attribute lookup, inheritance and overriding, super(), the underscore-private convention, dataclasses, __iter__ and __next__ — plus a 100-accident chart.

Project BuildingPythonLearning & Career
Real photo: hands typing on a laptop showing Python code with def statements on screen, a blue notebook and eyeglasses beside it
Guide
Pack repeated work into functions and split files into modules: Python functions and modules

A function packs repeated work into a name; a module packs definitions into a file you take out with import. Following Chapter 4 sections 4.8-4.9 and Chapter 6 of the official Python Tutorial: def and return, local and global scope, defaults evaluated once, keyword arguments, *args and **kwargs, / and *, lambda, docstrings and annotations, the ways to import, the __main__ guard, packages and the search path — plus a 100-accident breakdown from a fictional bookstore script.

Project BuildingPythonLearning & Career