Sid Gupta
CEO and co-founder of AppBroda (ad monetization)
Where to hear and read about Sid Gupta3 episodes and tests
App MastersDo This to Increase Your AdMob Show Rate
App MastersAdMob Mediation
App MastersHow to Fix AdMob Invalid Traffic / ActivitySid Gupta is the CEO and co-founder of AppBroda, an ad monetization company that helps app developers earn more from Google AdMob and mediation partners. His most useful lesson is to treat ad revenue as a diagnostic problem: segment the data, find the exact slice that is failing, and fix it with one tested change at a time.
if it's above 50 don't even think about touching it
Do This to Increase Your AdMob Show Rate
Sid Gupta: the story and the lessons
1Sid Gupta and his work at AppBroda
Gupta runs AppBroda, which helps app developers grow and protect ad revenue. As CEO and co-founder of AppBroda, he works with many developers on AdMob show rate, mediation setup and account suspensions.
His benchmarks come from that hands-on work rather than official Google guidance, and he says so openly: "please don't quote me on it."
He treats show rate, mediation and policy as linked revenue levers. Across his talks he covers getting more matched ads displayed, making networks compete for inventory, and keeping the AdMob account alive in the first place.
2How Sid Gupta decides whether a low show rate is worth fixing
Gupta separates match rate from show rate before optimizing either. Match rate is the share of requests Google fills; show rate is the share of those filled ads actually displayed. Of 20 requests with 10 filled, match rate is 50%, and if 2 are shown, show rate is 20%.
He says many developers confuse the two, and they need different fixes.
He triages show rate in three bands: under 30, 30 to 50, above 50. Under 30% is low and worth fixing now because it drags down CPMs and revenue. From 30 to 50% is mediocre, worth long experiments when nothing else is in the pipeline. Above 50% rarely improves, so look for revenue elsewhere.
Google gives no official target; these bands are his rule of thumb from working with multiple developers.
3Gupta's drill-down for finding where ads fail to show
First rule out a broken integration with Google's test tools. Add a test device in the ad inspector and the mediation test tool before deeper analysis.
He admits it is obvious, but "sometimes when you are doing an update you might have goofed it up."
Slice the network report to locate the failing traffic. Break show rate down by format, platform, app version, ad type and ad source using the ads activity report.
This exposes a single bad build, a sizing issue with one ad type, or a mediation partner that fails to render ads.
Read error codes with listeners to learn why requests fail. Google answers each request with an ad or one of about 15 or 16 error codes, and listeners log which one came back.
The code separates a configuration fault from an ad delivery fault.
Confirm every fix with Firebase A/B tests and patience. Use Firebase to test replacing weak placements or disabling ads in a bad version. He says as few as 10,000 ad requests can show which variant works better.
He warns the process is long and tedious and calls for many experiments: "Firebase is your answer for running a couple of ab tests."
4The common show rate causes Gupta fixes
Adaptive banners solve ads that don't fit some devices. Fixed-size banners or native ads can lack space on certain devices; adaptive banners resize width and height automatically.
Ads on screens users never reach load but never show. In a hard game where most players quit after level three, ads on levels four and five drag down show rate. Move placements or lower difficulty.
Fast-moving users also leave a screen before the request returns, so request timing must give ads time to load.
Stale cached ads and over-preloading waste impressions. A cached ad held through a 30-minute session may no longer count, so call the is-ready function and move to the next ad if it fails. Preload fewer ads.
He slows native and banner refreshes without hurting eCPM. Developers tend to refresh native ads very quickly, which lowers show rate; reduce frequency until it helps without cutting revenue.
He ignores drops concentrated on low-end devices. Poor internet, overheating and low memory cause those failures, and little can be done about them.
5How Sid Gupta builds an ad mediation stack
Mediation earns more because it makes networks compete. With a single demand source you accept its price; more networks in the bidding stack push prices up.
Start from a baseline segmented by format, device and country. Use analytics such as Firebase to cluster audiences, since each combination may need its own strategy. AppLovin, ironSource and InMobi can matter in the US but have less demand in India.
Then list relevant partners, describe your inventory, and let them say where their demand is strongest. Match formats to partners, such as Unity for high-CPM rewarded video.
Begin with Google optimized flows, then set your own floors. Optimized flows set floors from geography, traffic and history and suit beginners. Manual flows let you set floors by country, such as one dollar.
In Ad Manager, a target eCPM lets Google build dynamic flows to hit a price, and bid distribution data shows bid counts and average prices for setting floors.
Multi-call and network optimization squeeze more from each partner. Multi-call requests the same placement from a partner several times; network optimization switches waterfall priority in real time as partner eCPMs swing with seasonality.
Check with each partner what multi-call allows; most permit it.
6Why Gupta moved toward open bidding and a hybrid
Long waterfalls add latency and cap how many partners you can use. A waterfall of seven to eight partners calls each in turn, slowing loads and hurting show rate, so developers kept stacks short.
Open bidding runs one live auction across many networks. Bids run in parallel, so 10, 15 or 20 partners can compete with no latency penalty, and the highest bid wins.
Server-level integrations cut most SDKs, making apps lighter, less crash-prone and easier to maintain, and dashboard numbers now match.
Running mediation through AdMob closes the cash flow gap. Google collects from all partners and pays on the 21st of each month, instead of developers chasing each network's cycle.
He calls the gap between paying up front for user acquisition and waiting for ad revenue the place developers struggle most.
He calls choosing one method over the other a bias. Put partners that bid well into open bidding and keep the rest in the waterfall stack.
"the world that is moving today is actually moving towards a hybrid strategy"
7Sid Gupta's playbook for an AdMob invalid activity suspension
An invalid activity suspension gives 30 days to fix things, with no appeal. Google reviews the account daily during that window, watching clicks, downloads and ad requests for patterns like very high CTR or repeat clickers.
Test with test ads only, never live ads. In a new app with low traffic, a team tapping live ads from one IP creates a click spike Google reads as invalid activity.
"please, please ensure you use test ads, always when you're doing this"
During review, cut ads and traffic until nothing looks unusual. Cap ads at one or two per user with Firebase, shrink acquisition budget to a few trusted sources, and drop tier three and four countries.
Frequency capping also blunts competitors who click ads to get an app banned.
Remove placements where users click ads by accident. New users tap the wrong spot, land on an ad and go back; Google treats this pattern as invalid traffic.
Transparent backgrounds overlapping content are a very common cause, so keep ad and content boundaries clearly separated.
8The content and placement rules Gupta says keep accounts safe
Google Play approval does not protect your AdMob account. The policies are related but separate: "Google might block your ad mob and might never remove you from Google play."
You are also responsible for iframe content, so adult, violent or hate content pulled from elsewhere can trigger a ban.
Declare the target age and justify any data you collect. Collect no data from users under 13. For any age, state why you need phone numbers, emails or location, or delay the request until the feature needs it.
Never tie ads to navigation or the start of an action. Next, back and icon buttons must not lead to an ad, and interstitial, rewarded or video ads at the start of a game or scan break policy.
Replace a launch interstitial with the App Open format, which Google built for that moment.
Ads must look like ads and every screen must have a way back. Native ads, dialogs or notifications a reviewer can't tell from content are violations, and so is a broken back button or endless loop.
He calls the back button rule a very understated policy.
Do This to Increase Your AdMob Show Rate
Explains the difference between AdMob match rate and show rate, what show rate to aim for, and a step-by-step process to diagnose and fix a low one. The benchmark bands (under 30, 30 to 50, above 50) tell you when it is worth the effort.
Gupta runs AppBroda, which helps app developers grow and protect ad revenue. As CEO and co-founder of AppBroda, he works with many developers on AdMob show rate, mediation setup and account suspensions.
His benchmarks come from that hands-on work rather than official Google guidance, and he says so openly: "please don't quote me on it."
He treats show rate, mediation and policy as linked revenue levers. Across his talks he covers getting more matched ads displayed, making networks compete for inventory, and keeping the AdMob account alive in the first place.
Gupta separates match rate from show rate before optimizing either. Match rate is the share of requests Google fills; show rate is the share of those filled ads actually displayed. Of 20 requests with 10 filled, match rate is 50%, and if 2 are shown, show rate is 20%.
He says many developers confuse the two, and they need different fixes.
He triages show rate in three bands: under 30, 30 to 50, above 50. Under 30% is low and worth fixing now because it drags down CPMs and revenue. From 30 to 50% is mediocre, worth long experiments when nothing else is in the pipeline. Above 50% rarely improves, so look for revenue elsewhere.
Google gives no official target; these bands are his rule of thumb from working with multiple developers.
First rule out a broken integration with Google's test tools. Add a test device in the ad inspector and the mediation test tool before deeper analysis.
He admits it is obvious, but "sometimes when you are doing an update you might have goofed it up."
Slice the network report to locate the failing traffic. Break show rate down by format, platform, app version, ad type and ad source using the ads activity report.
This exposes a single bad build, a sizing issue with one ad type, or a mediation partner that fails to render ads.
Read error codes with listeners to learn why requests fail. Google answers each request with an ad or one of about 15 or 16 error codes, and listeners log which one came back.
The code separates a configuration fault from an ad delivery fault.
Confirm every fix with Firebase A/B tests and patience. Use Firebase to test replacing weak placements or disabling ads in a bad version. He says as few as 10,000 ad requests can show which variant works better.
He warns the process is long and tedious and calls for many experiments: "Firebase is your answer for running a couple of ab tests."
Adaptive banners solve ads that don't fit some devices. Fixed-size banners or native ads can lack space on certain devices; adaptive banners resize width and height automatically.
Ads on screens users never reach load but never show. In a hard game where most players quit after level three, ads on levels four and five drag down show rate. Move placements or lower difficulty.
Fast-moving users also leave a screen before the request returns, so request timing must give ads time to load.
Stale cached ads and over-preloading waste impressions. A cached ad held through a 30-minute session may no longer count, so call the is-ready function and move to the next ad if it fails. Preload fewer ads.
He slows native and banner refreshes without hurting eCPM. Developers tend to refresh native ads very quickly, which lowers show rate; reduce frequency until it helps without cutting revenue.
He ignores drops concentrated on low-end devices. Poor internet, overheating and low memory cause those failures, and little can be done about them.
AdMob Mediation
Ad mediation turns a single demand source into a competition between many ad networks, raising what each impression sells for. The episode walks through baselines, waterfall setup, open bidding and a hybrid approach, and why running it through AdMob also solves payment delays.
He treats show rate, mediation and policy as linked revenue levers. Across his talks he covers getting more matched ads displayed, making networks compete for inventory, and keeping the AdMob account alive in the first place.
Mediation earns more because it makes networks compete. With a single demand source you accept its price; more networks in the bidding stack push prices up.
Start from a baseline segmented by format, device and country. Use analytics such as Firebase to cluster audiences, since each combination may need its own strategy. AppLovin, ironSource and InMobi can matter in the US but have less demand in India.
Then list relevant partners, describe your inventory, and let them say where their demand is strongest. Match formats to partners, such as Unity for high-CPM rewarded video.
Begin with Google optimized flows, then set your own floors. Optimized flows set floors from geography, traffic and history and suit beginners. Manual flows let you set floors by country, such as one dollar.
In Ad Manager, a target eCPM lets Google build dynamic flows to hit a price, and bid distribution data shows bid counts and average prices for setting floors.
Multi-call and network optimization squeeze more from each partner. Multi-call requests the same placement from a partner several times; network optimization switches waterfall priority in real time as partner eCPMs swing with seasonality.
Check with each partner what multi-call allows; most permit it.
Long waterfalls add latency and cap how many partners you can use. A waterfall of seven to eight partners calls each in turn, slowing loads and hurting show rate, so developers kept stacks short.
Open bidding runs one live auction across many networks. Bids run in parallel, so 10, 15 or 20 partners can compete with no latency penalty, and the highest bid wins.
Server-level integrations cut most SDKs, making apps lighter, less crash-prone and easier to maintain, and dashboard numbers now match.
Running mediation through AdMob closes the cash flow gap. Google collects from all partners and pays on the 21st of each month, instead of developers chasing each network's cycle.
He calls the gap between paying up front for user acquisition and waiting for ad revenue the place developers struggle most.
He calls choosing one method over the other a bias. Put partners that bid well into open bidding and keep the rest in the waterfall stack.
"the world that is moving today is actually moving towards a hybrid strategy"
How to Fix AdMob Invalid Traffic / Activity
Explains how AdMob detects invalid traffic and policy violations, and what to change in an app during the 30 day window after a suspension. Covers click patterns, content and age rules, and ad placement mistakes that get accounts banned.
Gupta runs AppBroda, which helps app developers grow and protect ad revenue. As CEO and co-founder of AppBroda, he works with many developers on AdMob show rate, mediation setup and account suspensions.
His benchmarks come from that hands-on work rather than official Google guidance, and he says so openly: "please don't quote me on it."
He treats show rate, mediation and policy as linked revenue levers. Across his talks he covers getting more matched ads displayed, making networks compete for inventory, and keeping the AdMob account alive in the first place.
An invalid activity suspension gives 30 days to fix things, with no appeal. Google reviews the account daily during that window, watching clicks, downloads and ad requests for patterns like very high CTR or repeat clickers.
Test with test ads only, never live ads. In a new app with low traffic, a team tapping live ads from one IP creates a click spike Google reads as invalid activity.
"please, please ensure you use test ads, always when you're doing this"
During review, cut ads and traffic until nothing looks unusual. Cap ads at one or two per user with Firebase, shrink acquisition budget to a few trusted sources, and drop tier three and four countries.
Frequency capping also blunts competitors who click ads to get an app banned.
Remove placements where users click ads by accident. New users tap the wrong spot, land on an ad and go back; Google treats this pattern as invalid traffic.
Transparent backgrounds overlapping content are a very common cause, so keep ad and content boundaries clearly separated.
Google Play approval does not protect your AdMob account. The policies are related but separate: "Google might block your ad mob and might never remove you from Google play."
You are also responsible for iframe content, so adult, violent or hate content pulled from elsewhere can trigger a ban.
Declare the target age and justify any data you collect. Collect no data from users under 13. For any age, state why you need phone numbers, emails or location, or delay the request until the feature needs it.
Never tie ads to navigation or the start of an action. Next, back and icon buttons must not lead to an ad, and interstitial, rewarded or video ads at the start of a game or scan break policy.
Replace a launch interstitial with the App Open format, which Google built for that moment.
Ads must look like ads and every screen must have a way back. Native ads, dialogs or notifications a reviewer can't tell from content are violations, and so is a broken back button or endless loop.
He calls the back button rule a very understated policy.