Google App Campaigns: Setup and Optimization

App campaigns hand Google nearly every lever. What you still control is the goal, the bid, the creative mix and how cleanly installs and in-app events are measured, and that is where results are decided.

Last updated: November 8, 2026

In This Guide

  1. Executive Summary
  2. Benchmarks & Numbers at a Glance
  3. When to Use App Campaigns
  4. Goals & Bidding
  5. Creative Assets
  6. Measurement & Attribution
  7. In-App Events & Value
  8. iOS & SKAdNetwork
  9. Learning Period & Budgets
  10. Optimization & Scaling
  11. Common Mistakes to Avoid
  12. What Changed Recently
  13. References

1. Executive Summary

Google Ads App campaigns in November 2026 are a fully automated, cross-network user acquisition and engagement system. Five principles define best-in-class management:

  • Optimize to the deepest measurable outcome. Installs are a proxy. The campaign goal should be the highest-value event that has enough reliable conversion volume—registration, purchase, or subscription—not the cheapest measurable action.
  • Budget to the formula, not to intuition. Install campaigns require a daily budget of at least 50× the target CPI. In-app-action campaigns require 10–15× the target CPA. Underfunding a campaign starves the learning algorithm and produces misleading performance data.[1]
  • Creative breadth is a delivery variable. Google’s automation requires a wide, format-diverse asset library—up to 20 images, 20 videos, and 4 strong text assets—to access every eligible placement and to run statistically meaningful creative tests.[2]
  • Measurement quality determines bid quality. Target CPA and target ROAS are only as accurate as the conversion signals feeding them. Firebase SDK 12.12.1 or later, a correctly configured MMP, and clean event schemas are prerequisites, not optional upgrades.[3]
  • Respect the learning period. Material changes to bids, budgets, creative sets, or conversion events within 7–14 days of launch or a prior change reset the algorithm, extend volatility, and make performance evaluation meaningless. Changes should be incremental—generally within 15–20% of recent averages—and spaced at least one to two weeks apart.[4]

2. Benchmarks and Numbers at a Glance

Metric Typical range or threshold Applies when Source
U.S. iOS CPI, non-gaming apps $3.60 average; top-quartile accounts below $1.80 Google App campaigns, U.S. market, non-gaming vertical [5]
Android CPI vs. iOS CPI 30–40% lower than iOS equivalent Cross-platform campaigns; scope not fully specified in source [5]
U.S. finance-app iOS CPI $7.40 Finance vertical, U.S. iOS; use as directional benchmark only [5]
U.S. lifestyle-app CPI $2.10 Lifestyle vertical, U.S.; platform not specified in source [5]
U.S. B2B productivity-app CPI $5.80 B2B productivity vertical, U.S. iOS [5]
Casual-game CPI $1.20 Casual games; geography not specified in source [5]
North American finance-app CPI $4.13 Finance vertical, North America; not Google-only [6]
North American e-commerce CPI $2.49 E-commerce vertical, North America; not Google-only [6]
Minimum daily budget — install campaign (target CPI) 50× target CPI Google App campaign for installs; Google vendor guidance [1]
Minimum daily budget — in-app-action campaign (target CPA) 10–15× target CPA Google App campaign for actions or engagement; Google vendor guidance [1]
Minimum action volume for action optimization ≥10 distinct users completing the event per day Google App campaigns targeting a post-install action; Google vendor guidance [1]
Target CPI premium — “users likely to perform an in-app action” vs. “all users” ≥20% higher than the all-users campaign CPI Split install campaigns by audience quality; Google vendor guidance [1]
Recommended iOS install campaigns per account 8 or fewer iOS App campaigns; ensures SKAdNetwork campaign IDs are not fragmented [7]
Recommended minimum iOS conversions per day for model stability ≥30 conversions per day iOS App campaigns; Google guidance, volume will vary by event and budget [3]
Bid or budget change limit during stabilization ≤15–20% of current setting per change All App campaign types; changes beyond this threshold risk resetting learning [8]
Learning period after a material change 7–14 days All App campaign types following changes to bid, budget, or conversion event [8]
Google Ads estimated app-install CPA $6.92 Mobile app installs; methodology and scope not fully specified — vendor claim, use directionally [9]

3. When to Use App Campaigns

Google Ads App campaigns are the correct tool when the primary objective is driving installs, post-install actions, or in-app revenue at scale across Google’s owned and partner inventory—Search, Google Play, YouTube, Discover, and the Display Network—from a single automated campaign structure. They are not the right tool for driving website traffic, promoting a mobile website without an app, or acquiring users through channels outside Google’s ecosystem.

Use App campaigns when at least one of the following conditions is true:

  • The app is live (or approaching launch) on Google Play, the Apple App Store, or both, and paid user acquisition is a funded growth lever.
  • Post-install conversion data—registrations, purchases, subscriptions—is being tracked through Firebase or an approved MMP, and the account has or will shortly achieve the minimum event volume needed for action optimization.
  • The marketing team can commit to maintaining a stable campaign structure, a fully populated asset library, and consistent conversion tracking for at least 30 days before evaluating results against LTV-adjusted targets.

Do not launch App campaigns when tracking is unvalidated, when the daily budget cannot meet the 50× CPI or 10–15× CPA thresholds, or when the app store listing is materially incomplete—poor store assets reduce the value of the store-listing assets Google pulls automatically.[1]

Worked example (illustrative)

Deciding whether to launch an App campaign for a new fintech app

  • Setup: A Chicago-based personal finance app has just launched on iOS and Android. The team has Firebase installed but has recorded only 47 purchases in the first 14 days across both platforms—roughly 3 purchases per day. The proposed daily budget is $500.
  • Numbers: Finance-app iOS CPI benchmark = $7.40[5]. Minimum daily budget for install campaign = 50 × $7.40 = $370/day—the $500 budget clears this threshold. However, minimum action volume for action optimization = 10 distinct users completing the target event per day[1]; the app is currently recording only 3 purchases/day. Action (CPA) optimization is therefore not yet viable.
  • Decision: Launch an App campaign for installs using target CPI = $7.40, daily budget = $500, conversion event = first open. Do not set purchase as the optimization event until purchase volume reaches ≥10/day.
  • Why: Google requires ≥10 distinct users completing the target event per day before in-app-action optimization has sufficient signal[1]; launching CPA optimization on 3 purchases/day would restrict delivery and produce unreliable bids.

Worked example (illustrative)

Choosing between install and engagement campaigns for a mature gaming app

  • Setup: A mobile casual game account spending $8,000/month has been live for six months. Firebase reports an average of 42 “level_5_complete” events per day at an observed CPA of $1.90. The team wants to shift optimization from installs to level completions.
  • Numbers: Casual-game CPI benchmark = $1.20[5]. Current CPA for level_5_complete = $1.90. Minimum daily budget for action campaign = 10 × $1.90 = $19/day at the low end, 15 × $1.90 = $28.50/day at the high end[1]. Monthly budget = $8,000 ÷ 30 = $266.67/day—far above the minimum. Daily action volume = 42 events, well above the ≥10/day threshold.
  • Decision: Migrate to an App campaign for engagement, set target CPA = $1.90, daily budget = $267, optimization event = level_5_complete.
  • Why: Both the daily event volume (42 ≥ 10) and the budget-to-CPA ratio (140× > 10×) exceed the thresholds required for stable in-app-action optimization[1], making engagement optimization viable.

4. Campaign Goals and Bidding (tCPI, tCPA, tROAS)

Google App campaigns offer four bidding approaches mapped to four campaign objectives. Selecting the wrong pairing—for example, running target ROAS before purchase data is mature—is one of the most common and costly configuration errors in App campaign management.[4]

Objective Campaign type Bidding strategy Minimum viable condition
Maximize install volume App campaign for installs (ACi) — All users Target CPI App live; daily budget ≥ 50× tCPI
Installs likely to take an action ACi — Users likely to perform an in-app action Target CPI (≥20% above all-users tCPI) Some post-install event data; insufficient volume for direct CPA opt.
Specific post-install action App campaign for engagement (ACe) Target CPA ≥10 users/day completing the event; daily budget ≥ 10–15× tCPA
Revenue or conversion value ACe or ACi with value signals Target ROAS or Maximize conversion value Frequent, accurate revenue events; Firebase or MMP value passing configured
Pre-launch registrations App campaign for pre-registration Target cost per pre-registration App available for pre-registration in targeted countries; daily budget ≥ 50× bid

Setting target CPI

The target CPI should reflect the expected economic value of a new user, not the lowest price at which installs can be purchased. Setting the target too low restricts delivery; setting it above the profitable ceiling acquires volume that does not pay back. Use 30-, 60-, or 90-day cohort contribution margin where available as the ceiling, then work down for margin.[4]

When running parallel install campaigns—one targeting all users, one targeting users likely to perform an in-app action—set the quality-focused campaign’s CPI at least 20% above the broad campaign’s CPI, because the audience is narrower and more valuable.[1]

Setting target CPA

Target CPA should be set at or below the expected contribution value of a user who completes the target event. If the event is a leading indicator rather than a revenue event, build in a haircut for the percentage of users who continue to purchase. Avoid optimizing toward a low-intent event indefinitely; plan to migrate to value-based optimization once revenue data is reliable.[4]

Setting target ROAS

Target ROAS requires accurate, timely conversion values passed through Firebase or an MMP. A 300% target ROAS means seeking $3.00 of tracked conversion value per $1.00 of ad spend. Set the target close to recent achievable performance—Google’s guidance is to review the previous 30 days of actual conversion-value data before setting or changing a ROAS target, and to avoid targets substantially above historical averages.[8] Overly aggressive ROAS targets reduce eligible auction participation and can cause delivery to stall.

Worked example (illustrative)

Calculating a viable target CPI for a parallel all-users / quality-users campaign split

  • Setup: A San Francisco e-commerce app is launching two parallel ACi campaigns: one targeting all users, one targeting users likely to make a purchase. The app team’s 30-day cohort data shows each purchasing user generates $18.00 in gross margin. The team wants a 3-month payback and tolerates a maximum CPI of $6.00 on the quality campaign.
  • Numbers: North American e-commerce CPI benchmark = $2.49[6]. All-users campaign target CPI = $2.49 (benchmark). Quality campaign minimum CPI = $2.49 × 1.20 = $2.99. The team’s economics allow up to $6.00, so target CPI for the quality campaign is set at $3.50 to remain within the economic ceiling while exceeding the 20% premium. Daily budget for all-users campaign = 50 × $2.49 = $124.50/day. Daily budget for quality campaign = 50 × $3.50 = $175.00/day.
  • Decision: Set all-users ACi target CPI = $2.49, daily budget = $125. Set quality ACi target CPI = $3.50 (+40.6% above all-users), daily budget = $175.
  • Why: Google requires the quality-focused install campaign CPI to be set at least 20% above the all-users campaign CPI because the audience is narrower and more valuable[1]; $3.50 satisfies that floor and remains within the $6.00 economic ceiling.

Worked example (illustrative)

Setting target ROAS for a subscription app with reliable revenue data

  • Setup: A health-and-fitness subscription app has 90 days of Firebase purchase data showing an average in-app revenue of $9.50 per converted user and an observed CPI of $3.60.[5] The team wants to shift from target CPA to target ROAS.
  • Numbers: Observed ROAS over the last 30 days = ($9.50 revenue ÷ $3.60 cost) × 100 = 263.9%. To set a viable tROAS target close to historical performance, the team sets tROAS = 250%—approximately 5% below the 30-day observed figure to allow delivery headroom. Daily spend = $1,200. Expected daily revenue at 250% ROAS = $1,200 × 2.50 = $3,000.
  • Decision: Set target ROAS = 250%, daily budget = $1,200. Do not change the target for at least 14 days.
  • Why: Google’s guidance is to set ROAS targets close to the previous 30 days of actual conversion-value data and avoid targets substantially above historical averages[8]; 250% is within 5% of the observed 263.9%, preserving delivery while targeting efficiency.

Managed for you

Everything on this page, handled every month

You are reading how to do it properly. Omologist does it, inside a written mandate, for 15% of your budget.

✓  Conversion tracking built and tested before any ad runs

✓  Search campaigns built to what your budget can actually feed

✓  Changes made inside a written mandate, not whenever

✓  A report on the 1st that says what was done and why

See the service and prices

15% of budget plus GST · minimum A$330 incl GST · accounts under about A$10,000 a month

5. Creative Assets and Asset Variety

Google App campaigns assemble ads automatically from the asset library an advertiser provides. The breadth, quality, and format diversity of that library directly determine how many placement types the campaign can enter, how many creative combinations are tested, and how efficiently the algorithm finds the strongest performers.[2] A thin library limits inventory access and creative learning simultaneously.

Asset limits and targets

Asset type Maximum allowed Practical recommendation
Text (headlines/descriptions) Up to 10 (interface); dedicated guidance recommends 4 strong, independent ideas 4 meaningfully different texts covering user problem, primary benefit, differentiator, CTA
Images Up to 20 20 distinct visual concepts across square, landscape, and portrait orientations
Videos Up to 20 Multiple videos differing in concept, opening, length, and orientation
HTML5 / playable Up to 20 (where eligible) Use for mobile games and interactive previews; pair with static and video assets

Asset variety has three dimensions that must all be addressed: concept variety (different user needs, benefits, proof points), format variety (text, image, video, HTML5), and placement variety (portrait, square, landscape; multiple video lengths).[10] Uploading 20 near-identical image crops satisfies the count but fails all three dimensions.

After launch, monitor asset-level ratings. Add assets similar to those rated “Best” and replace those rated “Low.” Allow 2–3 weeks after any asset change before judging its effect on performance.[11] Refresh incrementally rather than replacing the entire library at once to minimize disruption to the algorithm’s creative model.[2]

The AppTopCombinationView resource, introduced in Google Ads API v22, now surfaces the best-performing asset combinations rather than only individual asset ratings. Use this data to identify which headline, image, and video pairings drive results, then build additional variants around those concepts.[12]

Worked example (illustrative)

Diagnosing a thin creative library and calculating the minimum viable asset build

  • Setup: A Boston meal-kit delivery app launches an ACi campaign with 2 text assets, 5 images (all landscape, all showing the same dish), and 1 horizontal video. Ad Strength is rated “Poor.” The campaign has been live for 21 days with a daily budget of $300 at a target CPI of $3.60.[5]
  • Numbers: Current asset count: 2 texts, 5 images (1 concept, 1 orientation), 1 video (1 orientation). Target counts per Google guidance: 4 texts, 20 images (multiple concepts and orientations), up to 20 videos (multiple orientations and lengths)[2]. Gap: 2 additional texts, 15 additional images including square and portrait versions, at least 4 additional videos in portrait and square formats. Minimum daily budget check: 50 × $3.60 = $180/day—$300 clears the threshold, so delivery limitation is creative, not budget.
  • Decision: Produce and upload 2 additional text assets, 15 additional images (5 portrait, 5 square, 5 landscape of distinct concepts), and 4 additional videos (2 portrait 15-second cuts, 2 square 30-second cuts). Do not change budget or bid during the asset expansion.
  • Why: Google’s creative guidance requires distinct visual differences across image assets and multiple video orientations to access all eligible placements[10]; a single orientation and concept restricts inventory and creative learning simultaneously.

6. Measurement and Attribution (Firebase, MMPs)

Bidding quality in an App campaign is bounded by measurement quality. Target CPA and target ROAS are mathematical functions of the conversion signals Google receives; if those signals are delayed, duplicated, missing, or misdefined, the algorithm bids against a distorted objective. Building a reliable measurement stack is therefore a prerequisite to profitable automation, not an optional enhancement.[3]

Firebase vs. MMP: roles and responsibilities

Component Primary role Key 2026 requirement
Firebase / Google Analytics for Firebase Native Google measurement, event collection, on-device conversion measurement, audience creation SDK version 12.12.1 or later for on-device measurement[3]
AppsFlyer Cross-network attribution, cohort/LTV reporting, fraud prevention, postback management Latest SDK; Advanced Data Sharing enabled for Google Ads integration[3]
Adjust MMP attribution, SKAN/AdAttributionKit workflows, fraud controls, cross-channel reporting Consistent event taxonomy; postback and privacy settings configured for Apple frameworks[3]

An MMP does not replace Firebase if Firebase is used for Google’s on-device measurement or product analytics. Firebase alone may not meet requirements for sophisticated cross-network attribution, fraud detection, or unified cohort reporting. The two systems serve complementary roles; define clearly which is used for bidding signals, executive reporting, cohort analysis, and reconciliation before launch.[3]

Validate the full attribution path in a test environment: app event → SDK receipt → Firebase or MMP processing → Google Ads import or postback → campaign reporting. Check for duplicate events, currency mismatches, time-zone differences, reinstall handling, consent states, and delayed postbacks before the campaign goes live.[3]

Set post-install conversion windows longer than the install window—ideally long enough to capture at least 90% of eventual conversions. Evaluating CPA or ROAS on a 1-day or 3-day window for events that typically materialize over 14–30 days produces systematically inflated cost figures that lead to incorrect target adjustments.[7]

Worked example (illustrative)

Diagnosing a conversion-window mismatch inflating reported CPA

  • Setup: A Seattle travel booking app runs a target CPA campaign optimized toward “booking_complete” with a 7-day post-install conversion window. After 14 days the team reports a CPA of $38.00 and considers raising the target CPA. Historical cohort data from Firebase shows that 65% of bookings occur between day 8 and day 21 post-install.
  • Numbers: Observed 7-day CPA = $38.00. If 65% of bookings are outside the 7-day window, only 35% of eventual conversions are being counted. Adjusted expected CPA = $38.00 × 0.35 = $13.30 (the fraction of conversions visible at day 7). True CPA once the full cohort matures = approximately $13.30—64.9% lower than the dashboard figure. The team was considering raising target CPA to $45, which would have overpaid for installs by approximately $31.70 per booking.
  • Decision: Extend the post-install conversion window from 7 days to 21 days in Firebase and Google Ads settings. Hold target CPA at the current setting of $20 and re-evaluate after 30 days using the extended window.
  • Why: Google recommends setting the post-install conversion window long enough to capture at least 90% of eventual conversions[7]; a 7-day window captures only 35% of this app’s bookings, making every reported CPA figure a severe overstatement.

Not ready to change anything today?

We post the sharpest bits of these guides — what a change is worth, how long it takes to show up, and when to leave it alone. Short enough to read on your phone.

Follow on LinkedIn Follow on Instagram

7. In-App Events and Value Optimization

Defining the right optimization event is the highest-leverage decision in App campaign management. An event that is too shallow—such as app open—trains the algorithm to find cheap openers, not valuable customers. An event that is too rare—such as a platinum subscription that 2% of users ever complete—gives the algorithm insufficient signal to learn effectively. The practical target is the deepest event that reliably generates ≥10 completions per day across the campaign’s current delivery level.[1]

Event hierarchy

Build a stable hierarchy of tracked events rather than instrumenting every interaction. Mark only business-relevant events as primary conversions in Google Ads. Keep diagnostic or low-intent events secondary so bidding does not optimize toward cheap but unprofitable activity.[3]

  • Install and first open: required baseline; use for install campaigns and delivery diagnostics.
  • Activation: registration, tutorial completion, account verification—first meaningful milestone of user value.
  • Leading indicators: add-to-cart, trial start, loan application, level 5 completion—use only when they reliably predict downstream revenue.
  • Revenue events: purchase, subscription start, renewal—the preferred optimization target once volume is sufficient.
  • Retention events: repeat purchase, subscription renewal—useful for cohort evaluation and long-term LTV modeling.

Pass value and currency on every revenue event. Validate refund handling, subscription renewal deduplication, and consistent transaction IDs. Without accurate value passing, target ROAS optimization bids against an incorrect objective.[3]

When migrating from one optimization event to a deeper one—for example, from first_open to purchase—do not change the event mid-campaign. Create a new campaign with the target event, let it accumulate 7–14 days of learning data, and then evaluate whether it can sustain delivery before deprecating the original campaign.[4]

Worked example (illustrative)

Migrating the optimization event from activation to purchase for a productivity app

  • Setup: A B2B productivity app has been running an ACe campaign optimized toward “account_created” (activation) at target CPA = $12.00, daily budget = $180. After 60 days the Firebase dashboard shows 28 activations per day and 11 purchases per day. The team wants to migrate optimization to “purchase.”
  • Numbers: Current daily activation volume = 28 (above ≥10 threshold). Current daily purchase volume = 11 (above ≥10 threshold[1]). Observed purchase CPA based on existing spend: $180/day ÷ 11 purchases/day = $16.36/purchase. B2B productivity-app CPI benchmark = $5.80[5]. Minimum daily budget for new purchase-optimized campaign = 10 × $16.36 = $163.60/day—the existing $180 budget clears the minimum. Learning period required after launch = 7–14 days[8].
  • Decision: Launch a new ACe campaign with optimization event = purchase, target CPA = $16.36, daily budget = $180. Run both campaigns in parallel for 14 days before pausing the activation campaign.
  • Why: Google advises against switching the conversion event on a live campaign[4]; launching a new campaign preserves the existing activation campaign’s learning while the purchase campaign accumulates its own conversion history during the mandatory 7–14 day learning period.

Worked example (illustrative)

Deciding whether a leading-indicator event is a valid optimization proxy

  • Setup: A Dallas insurance quote app wants to optimize toward “quote_started” because only 4 policy purchases per day are being recorded—below the ≥10/day threshold. The team has 90-day Firebase data showing that 31% of users who start a quote complete a purchase within 30 days, and the average policy generates $42.00 in gross margin.
  • Numbers: Purchase volume = 4/day (below ≥10 threshold[1]). Quote_started volume = 29/day (above threshold). Implied CPA for a purchaser via quote_started: if 31% of quote_starters purchase, each quote_started event represents $42.00 × 0.31 = $13.02 in expected value. Set target CPA for quote_started = $13.02 × 0.70 (30% margin haircut) = $9.11. Minimum daily budget = 10 × $9.11 = $91.10/day.
  • Decision: Set optimization event = quote_started, target CPA = $9.00, daily budget = $100. Plan to migrate to purchase optimization once purchase volume reaches ≥10/day for 14 consecutive days.
  • Why: Google requires ≥10 distinct users completing the target event per day for action optimization[1]; quote_started at 29/day satisfies the threshold while purchase at 4/day does not, making quote_started a valid interim proxy provided its correlation with purchase (31%) is empirically established.

8. iOS, SKAdNetwork, and Privacy

iOS measurement for Google App campaigns operates under Apple’s privacy framework, which provides aggregated, delayed, threshold-gated attribution rather than the user-level, near-real-time visibility available on Android. No single report on iOS should be treated as a complete view of performance; accurate evaluation requires reconciling multiple signals.[7]

How SKAdNetwork and AdAttributionKit work in practice

Apple’s SKAdNetwork provides privacy-preserving attribution for impressions, clicks, installs, and limited post-install signals. Apple’s AdAttributionKit, its successor framework, is described as supporting attribution for app installations and re-engagement across multiple advertising formats and channels.[13] Advertisers should maintain an MMP or Firebase implementation that can support Apple’s evolving framework rather than hard-coding analysis around one legacy report format.

Practical constraints include:

  • Postbacks are delayed and aggregated; same-day performance dashboards are incomplete for iOS.
  • Conversion values must be mapped to meaningful outcome tiers—activation, trial, revenue bands—before launch; retroactive changes disrupt historical comparability.
  • Privacy thresholds can suppress postbacks for smaller campaigns or less common conversion events.
  • Google recommends running 8 or fewer iOS install campaigns so SKAdNetwork campaign identifiers are not fragmented across too many campaigns, which reduces the statistical usefulness of each postback.[7]

Google’s recommended iOS measurement stack

  • Present the App Tracking Transparency (ATT) prompt at a contextually appropriate moment and optimize the prompt copy for opt-in rate.
  • Integrate Firebase SDK 12.12.1 or later for on-device conversion measurement.[3]
  • Use an approved MMP with its latest SDK and configure Advanced Data Sharing (AppsFlyer) or equivalent settings to pass Google Ads signals.[3]
  • Enable probabilistic attribution where the MMP and account support it.[3]
  • Reconcile SKAdNetwork or AdAttributionKit postbacks, Google-reported conversions, MMP totals, and modeled data. Use geo-holdout or incrementality tests for strategic decisions.
  • Evaluate iOS performance by cohort and over longer windows (D7, D14, D30), not by same-day dashboard metrics.

Google’s guidance also recommends maintaining at least 30 iOS conversions per day to give its models sufficient signal for stable optimization, though the required volume varies by event rarity, geography, and budget.[3]

Worked example (illustrative)

Structuring iOS campaigns to stay within the 8-campaign SKAdNetwork limit

  • Setup: A New York media streaming app is managing 12 separate iOS ACi campaigns segmented by genre interest (sports, news, entertainment, kids, true crime, comedy, drama, reality, international, documentary, music, gaming). SKAdNetwork postbacks from campaigns 9–12 are returning no data due to privacy threshold suppression.
  • Numbers: Current iOS campaign count = 12. Google recommended maximum = 8[7]. Campaigns exceeding the limit = 4. Suppressed postback campaigns = 4 (campaigns 9–12). Daily iOS budget across all 12 campaigns = $2,400 ($200/campaign average). Consolidating to 8 campaigns: $2,400 ÷ 8 = $300/campaign/day average—above the 50 × $3.60 (health/fitness benchmark[5]) minimum of $180/day, confirming each consolidated campaign is adequately funded.
  • Decision: Consolidate from 12 iOS campaigns to 8 by merging the 4 lowest-volume genre segments into broader thematic groups (e.g., documentary + international, music + gaming). Maintain total iOS daily budget at $2,400.
  • Why: Google recommends 8 or fewer iOS install campaigns to prevent SKAdNetwork campaign identifier fragmentation that suppresses postbacks below privacy thresholds[7]; 4 of the current 12 campaigns are already receiving no attribution data.

“So what do I actually have to do?”

Tell us the business and give us the account. We build the conversion tracking and prove it works with a test conversion before any ad runs, then launch the campaigns your budget can feed.

After that, changes happen inside a written mandate, so you know in advance what gets changed without asking and what never gets touched. A report on the 1st, a budget email around the 23rd, and nothing else unless it needs your yes or no.

How it works

9. The Learning Period and Budgets

Google App campaigns use machine learning to identify the users, placements, creatives, and bidding levels that best achieve the specified goal. This learning requires a minimum volume of conversion data and a stable account environment. Performance during the learning period—typically 7–14 days after launch or a material change—is not representative of steady-state performance and should not trigger bid or budget reactions.[8]

The learning period is extended or restarted by material changes to: target CPI, CPA, or ROAS; daily budget (changes above 15–20%); conversion event definition; campaign objective; geographic targeting structure; or the creative asset library (mass replacement).[4]

Judge readiness to exit learning by conversion volume and stability, not by elapsed days alone. A campaign on a very low budget or optimizing toward a rare event may still be learning after 21 days. The practical test is whether the algorithm is producing stable CPAs or CPIs within an acceptable range of the target over a 7-day rolling window.[3]

Budget minimums

The budget-to-bid ratio is a delivery constraint, not a profitability guarantee. A campaign set below the minimum will typically be budget-limited before it accumulates enough conversion volume to exit learning, creating a feedback loop of under-delivery and poor optimization.[1]

  • Install campaigns (target CPI): daily budget ≥ 50× target CPI.[1]
  • In-app-action campaigns (target CPA): daily budget ≥ 10–15× target CPA.[1]
  • Pre-registration campaigns: daily budget ≥ 50× the target cost per pre-registration.[1]
  • Target ROAS campaigns: budget must be sufficient to generate regular value conversions; Google warns that small budgets may prevent these campaigns from producing conversions.[4]

Worked example (illustrative)

Calculating the minimum viable budget for a pre-registration campaign before a Q4 2026 launch

  • Setup: A mobile puzzle game is targeting a November 15, 2026 launch on the Google Play Store. The team wants to run a pre-registration campaign starting November 1, 2026 and has set a target cost per pre-registration of $1.20.
  • Numbers: Minimum daily budget for pre-registration campaign = 50 × $1.20 = $60.00/day[1]. Campaign duration = November 1 to November 14 = 14 days. Total minimum budget for the campaign = 14 × $60.00 = $840. The team has $1,500 available, which is 1.79× the minimum and provides headroom for learning volatility. Expected pre-registrations at $1.20 CPA with $1,500 budget = $1,500 ÷ $1.20 = 1,250 pre-registrations.
  • Decision: Set pre-registration campaign daily budget = $107.14 ($1,500 ÷ 14 days), target cost per pre-registration = $1.20, start date = November 1, 2026.
  • Why: Google requires a daily budget of at least 50× the pre-registration bid to give the campaign sufficient auction participation during the learning period[1]; $107.14/day is 89.3× the $1.20 target, well above the minimum.

Worked example (illustrative)

Scaling budget incrementally to avoid resetting campaign learning

  • Setup: A Miami food delivery app is running a stable ACe campaign at $400/day, target CPA = $22.00 for “first_order_placed.” The campaign exited learning on day 12 and has averaged $21.80 CPA over the following 7 days. The growth team wants to increase daily budget to $800 as quickly as possible.
  • Numbers: Current daily budget = $400. Target daily budget = $800 (100% increase). Google’s guidance: keep budget changes within approximately 15–20% to reduce the risk of resetting learning[8]. Maximum safe single increase = $400 × 0.20 = $80. Step 1 budget = $480. Wait 7 days. Step 2 budget = $480 × 1.20 = $576. Wait 7 days. Step 3 budget = $576 × 1.20 = $691.20. Wait 7 days. Step 4 budget = $691.20 × 1.20 = $829.44 (rounds to $800 target). Total elapsed time for safe ramp = 28 days (4 steps × 7 days each).
  • Decision: Increase daily budget in four 20% increments over 28 days: $480 → $576 → $691 → $800. Do not change target CPA during the ramp.
  • Why: Google’s optimization guidance states that keeping budget adjustments within roughly 10–20% reduces volatility and avoids resetting the learning period[8]; a single 100% increase from $400 to $800 would restart the 7–14 day learning clock and disrupt a campaign already achieving its $22.00 CPA target.

10. Optimization and Scaling

Once a Google App campaign has exited its learning period and is producing stable conversion volumes and costs near the target, the focus shifts to systematic scaling and performance improvement. Scaling without structure—raising bids aggressively, fragmenting campaigns, or rotating creative too fast—restarts learning and erases the signal the algorithm has accumulated.[8]

Optimization cadence

  • Daily: check spend, delivery status, tracking health, rejected events, anomalous CPI or CPA spikes, and significant MMP-to-Google Ads discrepancies.
  • Weekly: review conversion volume, event quality, cohort D7 retention, creative asset ratings, geographic and platform splits, and delayed conversion adjustments.
  • Every 2–4 weeks: adjust targets or budgets based on sufficiently mature cohort data. Use a 30-day window for diagnosis, not day-over-day comparisons.[8]
  • Quarterly: audit event schema, revenue logic, SDK versions, Apple framework configuration, attribution windows, privacy settings, and incrementality assumptions.[3]

Scaling levers

Scale budget before bid. Increasing daily budget within the 20% incremental threshold grows delivery without changing the auction signal. Raising tCPA or lowering tROAS targets should follow, not precede, evidence from mature cohorts that the current target is too conservative. When scaling geographically, add markets as separate campaigns rather than broadening an existing campaign’s geo targeting mid-flight, to preserve the existing campaign’s conversion history.[4]

Use the AppTopCombinationView (available via Google Ads API v22[12]) to identify which asset combinations are driving conversions, then produce additional creative variants around the strongest concepts. Avoid making mass asset replacements; refresh incrementally to minimize disruption to the creative model.[2]

Worked example (illustrative)

Using AppTopCombinationView data to prioritize the next creative iteration

  • Setup: An Atlanta retail fashion app is running an ACe campaign at $600/day, target CPA = $18.00 for “add_to_cart.” The team pulls AppTopCombinationView data for the period October 1–31, 2026 and finds that 3 of the top 5 asset combinations feature the same portrait-format video (15 seconds, showing a styling tutorial) paired with any of 2 specific text assets. The other 17 video assets in the library appear in fewer than 8% of top combinations.
  • Numbers: Top-performing video appears in 3 of 5 top combinations = 60% of top combinations. That video’s pairing with text asset A drives an estimated 38% of total campaign conversions. The team has 4 additional portrait-format video concepts in production. Cost to produce 4 additional portrait videos at $400 each = $1,600 total. Expected improvement: if the best-performing concept drives 38% of conversions, adding 4 variants of the same concept type could increase conversion coverage. Current daily conversions = $600 ÷ $18.00 = 33.3 conversions/day.
  • Decision: Prioritize production and upload of 4 portrait-format styling-tutorial videos (15–30 seconds each) in the next creative sprint. Upload 2 at a time, spaced 14 days apart, to allow learning between additions. Do not remove the existing top-performing video during the refresh.
  • Why: Google’s creative guidance recommends adding assets similar to those rated “Best” and refreshing incrementally rather than replacing the entire library[2]; AppTopCombinationView data confirms the portrait styling-tutorial concept is the strongest performer, making it the highest-ROI creative investment.

11. Common Mistakes to Avoid

The following errors are consistently associated with underperforming Google App campaigns. Each is actionable and has a specific corrective threshold or rule.

  • Setting target CPI below market clearing price. A target CPI that is materially below the vertical benchmark restricts delivery before the campaign can accumulate enough data to learn. For a U.S. finance app, setting target CPI at $2.00 when the vertical benchmark is $7.40[5] will result in minimal delivery. Correct by setting target CPI at or above the vertical benchmark, then optimize down from evidence.
  • Underfunding relative to the bid. Any campaign with a daily budget below 10× target CPA (for action campaigns) or 50× target CPI (for install campaigns) is structurally unable to exit learning at scale.[1] This is the most common cause of campaigns stuck in perpetual learning.
  • Optimizing toward an event with insufficient volume. Running target CPA optimization against an event that generates fewer than 10 completions per day produces unstable bids and delivery gaps.[1] Use a higher-volume leading indicator until the target event reaches ≥10/day.
  • Making changes faster than the learning period allows. Changing bids, budgets, creative sets, or conversion events within 7–14 days of a previous change restarts the learning clock and makes performance evaluation meaningless.[8] Space material changes at least one to two weeks apart.
  • Exceeding 8 iOS campaigns per account. Fragmenting iOS campaigns across more than 8 campaigns causes SKAdNetwork campaign identifiers to suppress postbacks below privacy thresholds, eliminating attribution data for the marginal campaigns.[7]
  • Using a 7-day conversion window for events that materialize over 30 days. This systematically overstates CPA and leads teams to raise bids unnecessarily. Set windows to capture at least 90% of eventual conversions.[7]
  • Uploading format-monotone creative libraries. A library with only landscape images and horizontal videos cannot access portrait or square placements. Distinct orientations—portrait, square, landscape—are required for full inventory access.[10]
  • Enabling target ROAS before value tracking is validated. If revenue values passed to Google are incorrect, delayed, or inconsistently defined, tROAS bids against a corrupted signal. Validate the full value-passing pipeline through Firebase or MMP before activating value-based bidding.[3]
  • Treating day-over-day performance as a signal. Google’s own troubleshooting guidance recommends using a 30-day comparison window when diagnosing App campaign performance, since daily auction and learning fluctuations are misleading.[8]
  • Switching campaign objective on a live campaign. Google advises against switching from install volume to in-app-action optimization on an existing campaign. Create a new campaign for the new objective and run both in parallel during the transition period.[4]

Worked example (illustrative)

Diagnosing and correcting a tCPA campaign stuck in perpetual learning

  • Setup: A Phoenix on-demand service app has been running an ACe campaign optimized toward “booking_confirmed” for 30 days. Daily budget = $95. Target CPA = $14.00. The campaign has been in “Learning” status for all 30 days and averages 4.2 conversions per day. The team is considering lowering target CPA to $10.00 to “force better performance.”
  • Numbers: Minimum daily budget for action campaign = 10 × $14.00 = $140.00/day[1]. Current daily budget = $95—below the $140 minimum by $45/day (32.1% shortfall). Daily conversions = 4.2 (below the ≥10/day threshold[1]). Lowering tCPA to $10.00 would increase the minimum required budget to 10 × $10.00 = $100/day—still potentially insufficient and would restart the learning period. Correct minimum budget at $14.00 tCPA = $140/day.
  • Decision: Increase daily budget from $95 to $140 (a 47.4% increase—above the 20% incremental guideline, but necessary to meet the structural minimum). Do not change target CPA. Allow 14 days for the campaign to restart learning at the correct budget level before evaluating CPA.
  • Why: The campaign is below the 10× CPA daily budget minimum[1], which is preventing it from generating the ≥10 daily conversions needed for the learning algorithm to function; lowering the target CPA without fixing the budget would worsen delivery, not improve it.

12. What Changed Recently (Last 30 Days)

The most significant App campaign developments from October–November 2026 fall into four areas: expanded bidding flexibility via the Google Ads API, a new asset-combination reporting resource, AI-assisted asset generation infrastructure, and updated troubleshooting guidance around conversion-signal health. No new App-campaign-exclusive ad format was confirmed in Google’s published announcements during this window; treat any vendor claims about new placements with caution until verified in your own account.[12]

View-Through Conversion Optimization

Google Ads API v26 introduced Campaign.view_through_conversion_optimization_enabled, a field that allows advertisers to signal that an App campaign should factor view-through conversions into its optimization target.[12] The default value is false, meaning existing campaigns are unaffected unless the field is explicitly set. This capability is API-level first and may not yet surface in the standard Google Ads UI for all accounts.

Before enabling this setting, audit your view-through conversion quality. If your measurement stack attributes a view-through conversion every time a user sees an impression and later installs through any channel, enabling this optimization can cause the bidding system to over-weight impressions that played no causal role in the install. A geo-holdout or incrementality test is the most reliable way to quantify the incremental value of view-through events before committing budget to optimizing toward them.[12]

Worked example (illustrative)

Evaluating View-Through Conversion Optimization for a Finance App

  • Setup: A Chicago-based personal finance app running an ACi campaign with target CPI of $4.13 (the North American finance-app benchmark[6]) is considering enabling view-through conversion optimization after a Google Ads API v26 upgrade. The team wants to understand whether enabling it will improve or distort performance.
  • Numbers: Current daily installs = 38, all click-attributed. View-through conversions reported in Google Ads over the same 30-day window = 210 (5.5 per day), representing an additional 14.5% of total reported conversions. If view-through conversions are included in the optimization signal, the effective CPI would appear to drop from $4.13 to approximately $4.13 ÷ 1.145 = $3.61—a 12.6% reduction that may be arithmetically misleading rather than a real efficiency gain. A geo-holdout test across two equally sized U.S. DMAs for 14 days would cost approximately 14 × $157 (daily spend at $4.13 CPI × 38 installs) = $2,198 in the test DMA to generate a statistically interpretable result.
  • Decision: Do not set Campaign.view_through_conversion_optimization_enabled to true until a 14-day geo-holdout test quantifies incrementality. Keep the field at its default value of false during the test period.
  • Why: The API default is false[12], and enabling optimization toward view-through conversions without an incrementality baseline risks inflating reported volume while reducing actual install quality, because the bidding system will allocate budget toward impressions that may not be causally driving installs.

New Bidding Goals Without a Target CPA or ROAS

Google Ads API v22 added App-campaign bidding goals that allow optimization for installs or total conversion value without requiring the advertiser to specify a target CPA or target ROAS value.[12] This mirrors the Maximize Conversions and Maximize Conversion Value strategies available in other campaign types. The practical utility is for accounts in one of two situations: first, a newly launched app where there is insufficient historical conversion data to set a defensible CPA or ROAS target; second, a testing phase where the advertiser wants to discover the market-clearing cost before committing to a fixed target.

Use these unconstrained goals with a firm daily budget cap and a predefined evaluation window—typically 14 days and at least 30 conversions—before switching to target CPA or target ROAS. Running unconstrained bidding indefinitely without a value ceiling is not a strategy; it is a measurement phase. As soon as enough cohort data exists to calculate a profitable target, transition to tCPA or tROAS.[1]

Worked example (illustrative)

Using Maximize Conversions to Discover Market-Clearing CPI Before Setting tCPA

  • Setup: A Seattle fitness app launching on Android in January 2026 has zero historical Google App campaign data. The category benchmark for health-and-fitness CPI is $3.60[5]. The team does not yet know whether this app will clear at $3.60 or higher, so they cannot set a defensible tCPA. They choose the new Maximize Conversions goal (no target specified) with a hard daily budget of $180.
  • Numbers: Daily budget = $180. Benchmark CPI = $3.60[5]. Expected installs per day at benchmark = $180 ÷ $3.60 = 50 installs/day. Android CPI discount versus iOS = 30–40%[5]; estimated Android CPI = $3.60 × (1 − 0.35) = $2.34. Expected installs at Android rate = $180 ÷ $2.34 ≈ 77 installs/day. The team will run Maximize Conversions for 14 days, targeting ≥30 total conversions (minimum for a stable tCPA read), then calculate 14-day average CPA and set tCPA at that figure ± 10%.
  • Decision: Launch with Maximize Conversions (no target CPA set) at $180/day for exactly 14 days ending January 28, 2026. After 14 days, calculate the actual average CPI from the campaign, set tCPA to that figure rounded to the nearest $0.50, and transition to target CPA bidding.
  • Why: The new unconstrained bidding goal[12] is designed for discovery phases where historical data is insufficient to set a defensible target; a 14-day window with a firm daily budget cap bounds the financial exposure while generating enough signal (targeting ≥30 conversions) for the first tCPA setting.

Top Asset-Combination Reporting via AppTopCombinationView

The AppTopCombinationView resource, available through the Google Ads API, surfaces which pairings of text, image, and video assets are performing best within a campaign.[12] Previously, asset-level reporting showed individual asset ratings—Good, Best, Low—but could not reveal which specific combinations were driving the strongest downstream performance. The combination view changes this, allowing creative teams to identify, for example, that a particular headline paired with a specific 15-second vertical video consistently produces lower CPA than other pairings.

Use the combination view to inform, not replace, creative testing. The data reflects historical auction outcomes and is subject to selection bias: an asset combination may appear in the top view because it received the most delivery rather than because it is inherently superior. Validate strong combinations by producing new variants built around the same concept and observing whether performance holds. Do not remove assets that are absent from the top-combination report without first checking their individual asset rating; an asset rated “Best” in isolation may contribute to combinations that the report does not surface prominently.[12][14]

Worked example (illustrative)

Using AppTopCombinationView to Prioritize the Next Creative Sprint

  • Setup: A Miami e-commerce app running an ACi campaign has 20 image assets, 4 text assets, and 8 video assets uploaded. The campaign has been live for 21 days. The creative team pulls the AppTopCombinationView report via the Google Ads API and finds that 3 of the top 5 combinations share one specific 10-second square video (asset ID: V-04) paired with a headline emphasizing free-shipping (“Free shipping on every order”).
  • Numbers: V-04 appears in 3 of the top 5 combinations. The 3 combinations containing V-04 account for a combined install volume of 410 installs out of 980 total campaign installs in the reporting window (41.8% of volume). The CPI for those combinations = $2.49 (North American e-commerce benchmark[6]). The remaining 570 installs from other combinations average $3.10 CPI—a $0.61 (24.5%) premium per install. Producing 3 new square videos built around the free-shipping concept costs an estimated $900 in production.
  • Decision: Commission 3 new square video variants (10–12 seconds each) built around the free-shipping concept by March 15, 2026. Upload all 3 as additional assets alongside V-04; do not remove any existing assets. Allow 21 days before re-pulling the combination report to evaluate whether the new variants enter the top combinations.
  • Why: The AppTopCombinationView data[12] shows that the free-shipping concept is generating 41.8% of installs at a CPI $0.61 below the remaining creative pool; producing concept variants (not copies) around the strongest theme is the highest-ROI next creative action, while keeping existing assets preserves the diversity the algorithm needs to continue testing.

AI-Assisted Asset Generation: AssetGenerationService

Google Ads API v22 introduced AssetGenerationService, which supports AI-generated text and image assets.[12] This is not an App-campaign-only feature—it applies across campaign types where the service is enabled—but it is relevant to App campaign workflows because the volume of asset variants required for effective creative coverage has historically been a production bottleneck. The service can accelerate first-draft generation of image and text variants, particularly when scaling from one market or language to another.

Treat AI-generated assets as a starting point, not a final product. Apply the same quality criteria that apply to human-produced assets: distinct visual concepts, independently meaningful headlines, and accurate claims consistent with the app’s store listing. AI-generated images in particular should be reviewed for brand consistency, accuracy, and policy compliance before upload. Google’s image asset guidance still requires that the main subject be prominent, that blank space not exceed 80% of the image, and that images be undistorted and clearly composed.[15]

Updated Troubleshooting and Stability Guidance

Google’s recently refreshed troubleshooting documentation reinforces several practices that are worth treating as standing operating procedure rather than one-time advice.[14]

  • Evaluate performance over 30-day windows, not day-to-day. Daily auction fluctuations and the conversion-reporting lag for post-install events make single-day CPI or CPA readings unreliable for decision-making. Use 30-day rolling windows when assessing whether a target or budget change is warranted.[14]
  • Monitor conversion-signal health continuously. A sudden drop in the global ping volume for the conversion action used by an App campaign—caused by an SDK update, a consent-mode misconfiguration, or a Firebase/MMP integration break—will cause Google’s bidding system to lose signal and can produce erratic delivery, inflated CPI, or campaign suspension. Add a daily alert for conversion count drops greater than 20% from the 7-day average.[14]
  • Keep bid and budget changes within 10–20% per adjustment. Google’s guidance specifically recommends this range to reduce learning volatility. Changes larger than 20% can reset the learning period and produce a 7–14 day window of degraded performance.[14]
  • Set targets within reach of recent actual performance. Review the previous 30 days of conversion volume or value data before changing tCPA or tROAS. A target set more than 20% above or below recent actuals is likely to either restrict delivery (if too aggressive) or overspend without profitable return (if too loose).[14]

Worked example (illustrative)

Diagnosing a Conversion-Signal Drop After an SDK Update

  • Setup: An Atlanta on-demand grocery app runs an ACe campaign optimized toward “first_purchase” events. On November 3, 2026, the engineering team releases app version 4.1.0, which updates the Firebase SDK. Within 48 hours, the Google Ads conversion count drops from an average of 22 “first_purchase” events per day to 6 per day—a 72.7% decline. Spend continues at $310/day (target CPA = $14.00; previous daily budget = $14.00 × 15 = $210, recently increased to $310 to scale).
  • Numbers: Pre-update daily conversions = 22. Post-update daily conversions = 6. Decline = 16 conversions/day = 72.7% drop. At $310/day spend and 6 conversions, effective CPA = $310 ÷ 6 = $51.67—3.7× the $14.00 target CPA. The ≥10 conversions/day minimum threshold[1] is now missed. If unaddressed for 7 days, cumulative overspend at effective CPA versus target = ($51.67 − $14.00) × 6 conversions × 7 days = $1,582.14 in spend above target-CPA economics.
  • Decision: Immediately pause the campaign on November 5, 2026 (day 2 post-update). Roll back Firebase SDK to the previous verified version in a hotfix release. Validate the full conversion path in a test device before reactivating. Reactivate the campaign at the original daily budget of $210 (not the scaled $310) and allow 14 days of restabilization before re-evaluating scaling.
  • Why: Google’s updated guidance states that a sudden fall in global pings for the conversion action causes delivery and cost changes[14] because the bidding system is optimizing toward a signal that has effectively disappeared; pausing immediately limits financial exposure while the root cause is resolved, and returning to the pre-scale budget reduces the risk of compounding the signal-loss impact during the restabilization window.

Practical Actions for November 2026

The following table summarizes the actionable changes from the last 30 days, their status, and the recommended immediate action for a U.S.-focused App campaign account.

Change Status / availability Recommended action Source
View-through conversion optimization (Campaign.view_through_conversion_optimization_enabled) API v26; UI availability varies by account Do not enable without an incrementality or geo-holdout test to verify view-through conversion quality [12]
Bidding goals without tCPA or tROAS (Maximize Conversions / Maximize Conversion Value) API v22; general availability Use for 14-day discovery phases on new campaigns; set a firm daily budget cap and transition to tCPA/tROAS after ≥30 conversions [12]
AppTopCombinationView asset-combination reporting API; not yet confirmed in standard UI Pull monthly via API; use top combinations to brief the next creative sprint—do not use to remove non-appearing assets without checking individual ratings [12]
AssetGenerationService (AI text and image generation) API v22; not App-campaign-exclusive Use for rapid variant generation; review all AI-generated assets for brand accuracy, policy compliance, and visual quality before upload [12]
Updated stability and troubleshooting guidance Published November 2026; applies immediately Evaluate performance on 30-day windows; set a daily alert for conversion-count drops >20%; keep bid/budget changes within 10–20% [14]
Asset Studio and AI video enhancements (Veo / Imagen) Broader Google Ads rollout; not App-campaign-exclusive Confirm availability in account before incorporating into App campaign workflows; do not substitute for human-produced creative strategy [16]

Worked example (illustrative)

Applying the Full November 2026 Update Checklist to an Existing Account

  • Setup: A Dallas productivity app account spends $4,200/month across two ACi campaigns and one ACe campaign. The account manager runs a monthly audit on November 10, 2026 to incorporate all changes from the last 30 days. Target CPI for ACi campaigns = $5.80 (U.S. B2B productivity benchmark[5]). Target CPA for ACe campaign = $29.00. Both ACi campaigns have been live for 45 days; the ACe campaign has been live for 22 days.
  • Numbers: ACi campaign 1 daily budget = $290 (50 × $5.80 = $290—exactly at minimum[1]). ACi campaign 2 daily budget = $310. ACe daily budget = $290 (15 × $29.00 = $435 recommended minimum[1]; current budget is $145 below minimum—33.3% shortfall). AppTopCombinationView pulled via API: top 3 combinations share one landscape video asset (V-02) and the headline “Built for teams.” 30-day conversion count for ACe campaign = 148 conversions (6.7/day—below the ≥10/day threshold[1]). No view-through conversion optimization is currently enabled. No recent SDK updates in the last 30 days.
  • Decision: (1) Increase ACe daily budget from $290 to $435 (a 50% increase; above the 20% incremental guideline but required to meet the structural minimum—implement in a single step and document). (2) Pull AppTopCombinationView data and brief creative team to produce 2 new landscape videos built around the “Built for teams” concept by November 24, 2026. (3) Do not enable view-through conversion optimization—no incrementality test has been run. (4) Set a Google Ads automated alert for daily conversion count drops >20% from the 7-day average on all three campaigns. (5) Make no tCPA or tROAS changes until the ACe campaign has been at the corrected budget for 14 days (re-evaluate December 1, 2026).
  • Why: The ACe campaign’s budget is 33.3% below the 15× CPA minimum[1], which is suppressing daily conversions to 6.7 (below the ≥10/day learning threshold[1]); fixing the budget is the highest-priority action because all other optimizations—creative testing, bid adjustments, and scaling—depend on the algorithm having sufficient signal to function correctly, and the updated Google guidance[14] explicitly links conversion-signal health to delivery stability.

References

  1. [1] https://support.google.com/google-ads/answer/6167156?hl=en support.google.com
  2. [2] https://support.google.com/google-ads/answer/6167158?hl=en-Gb support.google.com
  3. [3] https://business.google.com/us/accelerate/resources/articles/drive-better-performance-and-… business.google.com
  4. [4] https://support.google.com/google-ads/answer/7100895?hl=en support.google.com
  5. [5] https://apsteq.com/blog/mobile-app-install-ads-case-study/ apsteq.com
  6. [6] https://www.digitalapplied.com/blog/mobile-app-marketing-statistics-2026-install-data www.digitalapplied.com
  7. [7] https://support.google.com/google-ads/answer/10384955 support.google.com
  8. [8] https://support.google.com/google-ads/answer/12265334?hl=en-BD&ref_topic=18851185 support.google.com
  9. [9] https://adplus.com/tools/ad-benchmarks/google-ads-benchmarks-mobile-app adplus.com
  10. [10] https://services.google.com/fh/files/misc/creative_playbook.pdf services.google.com
  11. [11] https://ads.google.com/intl/en_us/intl/en_us/home/grow-my-app/take-action/ ads.google.com
  12. [12] https://developers.google.com/google-ads/api/docs/release-notes developers.google.com
  13. [13] https://developer.apple.com/app-store/ad-attribution/ developer.apple.com
  14. [14] https://business.google.com/us/accelerate/resources/articles/best-practices-guide-setting-… business.google.com
  15. [15] https://business.google.com/es-all/accelerate/resources/articles/make-better-creative-asse… business.google.com
  16. [16] https://business.google.com/us/accelerate/announcements/ business.google.com
Share the Post:

Related Posts