Performance Max Asset Groups Explained

  • Google Ads

This page is updated every two months with a current, evidence-based guide to Performance Max asset groups: what an asset group is, the current specs and limits for every asset type, and how to structure your groups so Google’s AI produces relevant ads instead of mush. The short version is that an asset group is a themed creative kit, and the way you split those kits — by buyer mindset rather than convenience — makes or breaks the campaign. Each update draws on our own build experience plus authoritative industry sources and verified real-time research. Bookmark this page and check back for the latest asset group specs and structuring guidance. Each update includes worked examples with the arithmetic shown.

Last updated: 7 August 2026

In This Guide

  1. Executive Summary
  2. Benchmarks & Numbers at a Glance
  3. What Is an Asset Group?
  4. Specs and Limits
  5. How Google Assembles Ads
  6. Why Structure Matters
  7. Structuring by Buyer Mindset
  8. How Many Asset Groups?
  9. Audience Signals and URLs
  10. Asset Group Checklist
  11. Common Mistakes
  12. What Changed Recently
  13. References

Reading this to build your own campaign? Skip the learning curve — we’ll build the whole thing for you for $129, delivered within 24 hours as a spreadsheet you own. See how it works →

1. Executive Summary

A Performance Max asset group is a themed creative kit that lives inside a single PMax campaign. It bundles every element Google’s AI needs to build ads across Search, Shopping, YouTube, Display, Discover, Gmail and Maps: headlines, long headlines, descriptions, a business name, a call to action, images, logos, videos, a final URL and at least one audience signal.[1] Google’s system mixes and matches those assets automatically, assembling different ad formats for different placements depending on what its models predict will convert.[3]

The central verdict of this article is: structure asset groups by buyer mindset, not by convenience. A single generic asset group forces Google to combine unrelated headlines, images and descriptions into mushy, off-theme ads that serve no one’s intent well.[3] Splitting every product SKU into its own microscopic group is equally damaging: each group competes for the same campaign budget, conversion data thins out, and learning stalls — vendor guidance consistently cites a minimum of 20–50 conversions per asset group before conclusions stabilise.[7][1] The optimal middle ground is a small set of richly populated, thematically coherent groups — typically 3–5 for most accounts — each with copy, imagery, audience signals and a landing page that all point in the same direction.[3]

The framework this article uses — keyword groups, intent groups and competitor groups — organises those themes around buyer mindset. Keyword groups are built around specific product or service search themes. Intent groups are built around where the buyer sits in their decision journey. Competitor groups are built around competitor-brand and alternative-seeking demand. Each group type produces coherent asset combinations, gives Google a clear signal about who to serve, and maps to a landing page that matches the user’s likely expectation at that moment.

Key principles:

  • Fill every asset group to its limits: 15 headlines, 5 long headlines, 5 descriptions, at least 4 landscape and 4 square images, at least one portrait image, and at least 2–3 purpose-made videos.[1][15]
  • Audience signals are a suggestion, not a hard boundary; they accelerate learning but do not confine delivery.[3]
  • Business name and logos are set at the campaign level and should be consistent across every asset group.[17]
  • Allow 3–4 weeks of learning after any structural change before drawing conclusions.[1]
  • Where evidence on a specific threshold is vendor-claimed rather than independently verified, this article says so explicitly.

2. Benchmarks and Numbers at a Glance

Metric Typical range or threshold Applies when Source
ROAS uplift from restructuring to themed asset groups +18% over 30 days (42-account benchmark) Accounts restructured under a new asset-group framework vs unmodified PMax; labelled a study by the publisher but attributed to an internal Tinuiti benchmark — treat as vendor claim pending independent replication [69]
ROAS floor — fashion and apparel 3.5x–4.5x Vendor claim; working target from agencies managing collective Google spend, not a controlled study [69]
ROAS floor — consumables and health 5x–7x Vendor claim; same source as above [69]
ROAS floor — home goods and furniture 2.5x–3.5x Vendor claim; same source as above [69]
CTR — best-seller asset groups 1.8%–2.4% Vendor claim; Q3 2026; applies to well-optimised ecommerce asset groups [69]
CTR — review threshold (flag for action) Below 1.2% Vendor claim; Q3 2026; trigger to audit headlines and imagery in the under-performing group [69]
Minimum conversions before holding structural conclusions 50 conversions per asset group Vendor claim; Q3 2026; do not merge or split groups before this threshold is reached [69]
Minimum conversions per asset group per month (alternate guidance) 20–30 conversions per month Vendor claim; two separate 2026 sources report 20 and 30 respectively; use 30 as the more conservative threshold [7][15]
Learning period after restructure 3–4 weeks Vendor claim; Q3 2026; applies after any structural or creative change to an asset group [69]
Recommended asset groups per campaign — lead generation 3–8 Vendor claim; 2026; assumes sufficient monthly conversion volume per group [72]
Recommended asset groups per campaign — ecommerce 5–15 Vendor claim; 2026; broader range reflects product catalogue depth [72]
Branded impression share — unoptimised campaigns pre-update 23% average Vendor claim; Q3 2026; benchmark for campaigns without brand exclusions or competitor segmentation [69]
Branded impression share — after optimisation Under 8% Vendor claim; Q3 2026; achieved after restructure and brand exclusion settings applied [69]
Weekly conversion volume to sustain efficient learning 15–20 conversions per week Vendor claim; 2026; minimum volume before smart bidding can optimise effectively within a group [74]
Time for duplicate asset-group performance to converge 4–6 weeks Vendor claim; 2026; relevant when A/B testing two structurally similar groups [70]

3. What Is an Asset Group? The Creative Kit

An asset group is the creative layer inside a Performance Max campaign. Where a traditional Search campaign organises keywords, ads and bids inside an ad group, a Performance Max campaign organises creative inputs and audience suggestions inside an asset group.[3][5] The campaign sets the objective, budget and bidding strategy. The asset group provides the raw material — copy, visuals, video and signals — that Google’s AI recombines into finished ads across every eligible Google channel.[1][5]

Structurally, a Performance Max campaign sits above asset groups, and asset groups sit above listing groups in retail campaigns. Asset groups cannot be shared between campaigns; each belongs to exactly one campaign.[1][3] A listing group — the product-selection layer that filters which Merchant Center products are eligible to show — operates alongside an asset group rather than replacing it: the asset group supplies the creative, and the listing group supplies the product inventory.[1][2]

The assets inside the kit are:[5][17]

  • Text assets: headlines, long headlines, descriptions, a business name
  • Action asset: call to action (selected or automated)
  • Destination asset: final URL, with optional display path fields
  • Visual assets: landscape, square and portrait images; square and landscape logos
  • Video assets: YouTube videos in horizontal, square or vertical format
  • Audience asset: at least one audience signal per group

Google’s AI assembles combinations of these assets into ad formats appropriate for each channel. A YouTube pre-roll is built from video plus headline plus description. A Display ad might be built from a landscape image plus a short headline plus logo. A Search-style text ad generated by PMax will draw on headlines and descriptions. Because Google is mixing assets within the group, every asset needs to be coherent with every other asset in the group — a principle that makes buyer-mindset structuring essential rather than optional.[3][1]

Worked example

Understanding What an Asset Group Contains

  • Setup: A Melbourne home-loan broker running a single Performance Max campaign at $4,500 per month wants to understand what to place inside one asset group before deciding how many groups to build.
  • Numbers: One asset group requires a minimum of 3 headlines (×30 characters), 1 long headline (×90 characters), 2 descriptions (first up to 60 characters, remainder up to 90 characters), 1 business name (×25 characters), 1 call to action, 1 final URL, 1 landscape image (minimum 600×314 px), and 1 square image (minimum 300×300 px). The maximums are 15 headlines, 5 long headlines, 5 descriptions, 20 landscape images, 20 square images, 20 portrait images, 5 logos and 15 videos per asset group.[1]
  • Decision: Build one fully-populated asset group first, filling it to the maximum of 15 headlines, 5 long headlines, 5 descriptions, 4 landscape images, 4 square images, 2 portrait images and 2 purpose-made YouTube videos before creating a second group.
  • Why: Google’s guidance states that richer creative coverage within a group produces stronger ad combinations; a group that meets only the minimums gives the AI fewer combinations to test and less signal to optimise against.[1][15]

4. Asset Group Specifications and Limits (Current)

The following table reflects the current asset-group limits as of July 2026. Where a vendor source confirms a limit that matches Google’s API documentation, both are noted. The video maximum of 15 per asset group is confirmed by Google’s official API documentation and represents an increase from the earlier limit of 5.[1][2]

Asset type Minimum Maximum Character limit / specifications
Headline 3 15 Up to 30 characters each; at least one headline at 15 characters or fewer is recommended[1]
Long headline 1 5 Up to 90 characters each[1]
Description 2 (2–3 recommended) 5 First description up to 60 characters; remaining descriptions up to 90 characters each[1]
Business name 1 1 Up to 25 characters; set at campaign level and consistent across all asset groups[17]
Call to action 1 1 One CTA; select from Google’s list or set to automated[1]
Final URL 1 1 One per asset group; optional display path of up to two fields, up to 15 characters each[1]
Landscape image (1.91:1) 1 20 Recommended 1200×628 px; minimum 600×314 px; JPG or PNG; max 5 MB; subject within centre 80%[1]
Square image (1:1) 1 20 Recommended 1200×1200 px; minimum 300×300 px; JPG or PNG; max 5 MB; subject within centre 80%[1]
Portrait image (4:5) 0 20 Recommended 960×1200 px; minimum 480×600 px; JPG or PNG; max 5 MB; subject within centre 80%[1]
Square logo (1:1) 1 (strongly recommended) 5 Recommended 1200×1200 px; minimum 128×128 px; max 5 MB; set at campaign level[1]
Landscape logo (4:1) 0 5 Recommended 1200×300 px; minimum 512×128 px; max 5 MB; set at campaign level[1]
YouTube video 0 (strongly recommended to supply at least 1) 15 Aspect ratios 16:9, 1:1 or 9:16; minimum 10 seconds; at least one vertical (9:16) 10–60 seconds recommended for Shorts[1][2]
Audience signal 1 No documented hard cap A suggestion to the AI, not a hard targeting boundary; accelerates learning[3]

A note on logos and business name: Logos and the business name are configured at the campaign level, not individually within each asset group. Do not attempt to differentiate branding across asset groups — it creates inconsistency and Google consolidates this information at campaign level regardless.[17]

A note on videos: If you supply no video, Google auto-generates one from your other assets. Multiple 2026 vendor sources and Google’s own creative guidance warn that auto-generated videos are consistently weaker than purpose-made assets.[11][15] At minimum, supply one horizontal (16:9) video and one vertical (9:16) video of 10–60 seconds per asset group. The video limit increased from 5 to 15 per asset group during 2026; this higher figure is confirmed by Google’s API documentation.[1][2]

Worked example

Diagnosing an Under-Resourced Asset Group Before Launch

  • Setup: A Brisbane commercial cleaning company is about to launch a Performance Max campaign at $3,000 per month. Their asset group contains 4 headlines, 1 long headline, 2 descriptions, 2 landscape images, 1 square image, no portrait images, no video and one audience signal.
  • Numbers: Maximums available but unused: 11 additional headlines (15 max − 4 supplied = 11 gaps), 4 additional long headlines (5 max − 1 = 4 gaps), 3 additional descriptions (5 max − 2 = 3 gaps), 18 additional landscape images (20 − 2 = 18 gaps), 19 additional square images (20 − 1 = 19 gaps), 20 portrait images (0 supplied), 15 videos (0 supplied). Current state is minimum-compliant but maximally impoverished.[1]
  • Decision: Delay launch by 5 business days to add: 11 more headlines (reaching 15), 4 more long headlines (reaching 5), 3 more descriptions (reaching 5), 3 additional landscape images (reaching 5), 3 additional square images (reaching 4), 2 portrait images (960×1200 px), and upload 2 purpose-made YouTube videos — 1 at 16:9 (minimum 10 seconds) and 1 at 9:16 (maximum 60 seconds for Shorts eligibility).
  • Why: Vendor guidance from a 2026 optimisation playbook identifies 3–5 long headlines, 4–5 descriptions, 15–20 images and 3–5 custom videos as the creative coverage threshold for optimal performance, and Google’s own guidance confirms that more assets produce more testable combinations.[15][1]

5. How Google Assembles Ads From Your Assets

When a user triggers an eligible impression — whether through a search query, a YouTube browse session, a Display placement or a Discover feed — Google’s AI evaluates every asset inside the relevant asset group and selects a combination it predicts will maximise conversion probability for that context.[1][5] The final ad format depends on the channel: a Search-like text ad might use three headlines and two descriptions; a responsive Display ad might use a landscape image, a logo, a single headline and a short description; a YouTube ad will draw on one or more video assets plus supporting copy.[3]

This mixing process is why thematic coherence within a group is not aesthetic preference — it is a structural requirement. If an asset group contains headlines written for first-time buyers alongside imagery designed for repeat purchasers and a landing page for enterprise customers, the AI may assemble combinations that mix those three incompatible signals into a single ad. No individual asset is wrong; the problem is the pool.[3][15]

Asset-level performance reporting — now documented in Google’s API with a dedicated resource updated in June 2026 — lets advertisers see which individual assets are generating conversions, clicks and impressions within each group.[55] Assets are rated; those rated low should be revised or replaced rather than simply accumulated alongside stronger assets, because weak assets dilute the pool even if they are rarely selected.

Google’s AI also uses the audience signal associated with an asset group as a starting point for its targeting model. It does not restrict delivery to that audience — it uses the signal to build a lookalike model and begin serving. Over time, the system expands beyond the signal to find additional converters. This means a well-structured audience signal accelerates early learning without permanently limiting reach.[3][44]

Worked example

Diagnosing Mixed Asset Combinations in a Single Generic Group

  • Setup: A Sydney HR software vendor runs one Performance Max campaign at $8,000 per month with a single asset group containing assets for both their payroll module (targeting finance managers) and their onboarding module (targeting HR managers). CTR is 0.9%, below the 1.2% vendor-flagged review threshold.[69]
  • Numbers: CTR of 0.9% is 0.3 percentage points below the 1.2% flag threshold — a 25% shortfall. The group contains 14 headlines: 7 about payroll accuracy and 7 about onboarding speed. Google’s AI draws from all 14 when building combinations, producing cross-module headline–image pairings (e.g. “Onboarding in 3 clicks” paired with a payroll dashboard screenshot) that match neither buyer segment.[69]
  • Decision: Split the single group into two asset groups at the campaign level — Asset Group 1: Payroll module, final URL = /payroll, audience signal = in-market for payroll software + Customer Match list of payroll contacts; Asset Group 2: Onboarding module, final URL = /onboarding, audience signal = in-market for HR software + Customer Match list of HR contacts. Each group receives 7 of the existing headlines plus 8 new module-specific headlines (reaching the maximum of 15 per group).
  • Why: Google’s guidance states that a single group containing mixed product themes forces the AI to combine unrelated assets, reducing the probability that any given combination aligns cleanly with a user’s query or intent.[3][15]

6. Why Structure Matters: One Generic Group vs Themed Groups

The single most common Performance Max mistake is placing all creative assets into one undifferentiated group. It is the path of least resistance during setup, and it consistently underperforms. The logic of why is mechanical: Google’s AI cannot distinguish which headlines belong with which images belong with which landing page when they are all in the same pool. The combinations it generates will be statistically average — they will serve, but they will not resonate with any specific buyer intent.[3][1]

At the other extreme, splitting every product variant or keyword theme into its own asset group is equally harmful. Each additional asset group shares the same campaign budget. If a campaign spends $6,000 per month across 20 asset groups, the average group receives $300 per month. At a typical conversion rate of 2%, that might generate 2–4 conversions per month per group — well below the 30-conversion-per-month minimum that vendor guidance recommends before conclusions can be drawn.[7][15] The AI cannot optimise what it cannot measure.

The evidence for themed structuring is primarily vendor-claimed rather than independently verified. One Q4 2025 vendor report claims 67% higher ROAS and 89% improvement in conversion rates from category-specific asset optimisation.[73] A Q3 2026 benchmark across 42 accounts found an 18% ROAS uplift from restructuring to a themed framework, though that study is attributed to an internal agency benchmark rather than a controlled experiment.[69] These figures should be treated as directionally plausible rather than empirically settled. What is well established in Google’s own documentation is that themed groups produce more coherent combinations, and coherent combinations are more likely to match user intent.[1][3]

Dimension One generic group Themed asset groups (3–5) Over-split (15+ thin groups)
Creative coherence Low — AI mixes unrelated assets High — each group’s assets share a message Potentially high per group, but many groups are under-resourced
Conversion data per group All data in one bucket; no theme-level insight Sufficient data per group at typical spend levels Often below 20–30 conversions/month per group[7][15]
Learning speed Fast (all budget in one group) 3–4 weeks per group at adequate budget[69] Slow; budget fragmented across 15+ groups
Audience signal alignment One signal for all themes — imprecise One tailored signal per theme — precise Precise per group but many groups starved
Landing page relevance One URL for all queries — weak match One URL per theme — strong match Precise, but AI has insufficient data to exploit it
Reporting granularity Campaign-level only; no theme breakdown Asset group level; theme-level optimisation possible[55] Granular but noisy; low-volume groups produce unreliable data

Worked example

Consolidating an Over-Split Account to Restore Learning

  • Setup: A Perth outdoor furniture retailer is spending $9,000 per month across one Performance Max campaign with 18 asset groups — one per product category (sun loungers, dining sets, umbrellas, fire pits, etc.). Average conversions per group in July 2026: 3.2 per month. Average CTR: 0.8%.
  • Numbers: $9,000 ÷ 18 groups = $500 average monthly spend per group. At a 2% conversion rate and an average order value of $480, each group generates approximately $500 × 0.02 = 10 click-throughs to purchase, or roughly 3–4 orders per group per month — 3.2 conversions confirmed by account data. This is 26.7 conversions below the 30-conversions-per-month threshold before conclusions stabilise.[15] The 0.8% CTR is 0.4 percentage points below the 1.2% review threshold.[69]
  • Decision: Consolidate 18 asset groups into 4: (1) Seating and lounging, final URL = /outdoor-seating; (2) Dining and entertaining, final URL = /outdoor-dining; (3) Shade and shelter, final URL = /shade-structures; (4) Heating and ambience, final URL = /fire-pits-heaters. Each consolidated group receives the combined assets of its 4–5 predecessor groups, edited to remove cross-theme conflicts, targeting a minimum of 15 headlines and 5 descriptions per group. Allow 4 weeks of learning before evaluating performance.
  • Why: Vendor guidance states that groups generating fewer than 20–30 conversions per month do not provide sufficient signal for the AI to optimise, and that consolidation restores learning speed by concentrating budget and data within coherent creative themes.[7][15][69]

7. Structuring by Buyer Mindset: Keyword, Intent and Competitor Groups

The buyer-mindset framework organises asset groups into three types, each targeting a different mental state: what the buyer is searching for (keyword groups), where they are in their decision process (intent groups) and who else they are considering (competitor groups). Within a Performance Max campaign, these map naturally to distinct combinations of copy angle, imagery, audience signal and landing page.[3][8]

Keyword Groups

A keyword group is built around a specific product or service search theme — the core category language a buyer uses when they know what they want but have not yet chosen a supplier. Headlines in a keyword group should mirror that category language. Images should show the product or outcome directly. The final URL should be the most relevant category or product page. The audience signal should include custom segments built from the keywords and URLs that high-intent searchers in that category use.[3][47]

Intent Groups

An intent group is built around a stage in the buyer journey. A top-of-funnel intent group might target users in the awareness or consideration phase with educational language, problem-framing imagery and a landing page featuring a guide or comparison tool. A bottom-of-funnel intent group targets users ready to convert with offer-specific headlines, social proof imagery and a page designed for purchase or enquiry. The audience signal for a bottom-of-funnel intent group should weight heavily toward Customer Match (existing enquirers and past purchasers) and website retargeting audiences.[3][44]

Competitor Groups

A competitor group captures users actively searching for or researching competitor brands and alternatives. The copy angle emphasises differentiation rather than category description — why your product outperforms a named competitor or a widely-searched alternative category. The audience signal should include custom segments built from competitor brand URLs. The landing page should be a dedicated comparison or switching page, not the generic homepage.[8]

Worked example

Keyword Group for a B2B Accounting Software Campaign

  • Setup: A Melbourne-based accounting software vendor launches a Performance Max campaign at $12,000 per month in August 2026 targeting small business owners. Their first asset group is a keyword group for “small business accounting software” and related category terms.
  • Numbers: Asset group contains 15 headlines all using category-direct language (e.g. “Small Business Accounting”, “Tax-Ready Invoicing”, “BAS Lodgement Built In”), 5 long headlines describing specific product workflows (each up to 90 characters), 5 descriptions, 6 landscape images (1200×628 px) showing dashboard screenshots and invoice creation, 4 square images (1200×1200 px), 2 portrait images (960×1200 px), 2 YouTube videos (1 × 16:9 at 45 seconds, 1 × 9:16 at 30 seconds), final URL = /small-business-accounting. Audience signal: custom segment built from keywords “accounting software for small business”, “invoice software Australia”, “BAS software” plus 3 competitor product URLs and 2,400-record Customer Match list of trial sign-ups from the previous 12 months.
  • Decision: Set call to action to “Start free trial”, enable final URL expansion restricted to /small-business-accounting/* via page feed, and set a target ROAS of 4.0x (within the 3.5x–4.5x software-adjacent vendor benchmark).[69]
  • Why: Aligning all 15 headlines, all imagery and the landing page to a single category search theme ensures every AI-assembled combination reflects the same buyer intent — category-directed, product-aware, conversion-ready — rather than mixing category copy with awareness-stage imagery.[3][1]

Worked example

Intent Group Split: Awareness vs Conversion for a Home Loan Campaign

  • Setup: A Sydney mortgage broker runs a Performance Max campaign at $7,500 per month in September 2026. They split intent into two asset groups: Intent Group 1 (top-of-funnel) and Intent Group 2 (bottom-of-funnel).
  • Numbers: Intent Group 1 (awareness): 15 headlines using educational and problem-framing language (“First Home? Start Here”, “How Much Can You Borrow?”, “Fixed vs Variable Explained”), final URL = /home-loan-guide, audience signal = in-market for home loans + users who visited /home-loan-guide but did not complete an enquiry form. Intent Group 2 (conversion): 15 headlines using offer and commitment language (“Get Pre-Approved Today”, “Compare 35 Lenders in 10 Min”, “No Upfront Broker Fee”), final URL = /apply-now, audience signal = Customer Match list of 1,100 past enquirers + website visitors who viewed /apply-now in the last 30 days. Both groups share the campaign’s $7,500 budget. After 4 weeks, Intent Group 2 produces 38 conversions (above the 30/month threshold[15]) and Intent Group 1 produces 14 (below threshold — monitor for a further 2 weeks before actioning).
  • Decision: Maintain both groups; raise tROAS on Intent Group 2 from 3.0x to 3.8x given it has cleared the 30-conversion threshold, and hold Intent Group 1 settings unchanged for a further 14 days to accumulate data.
  • Why: Audience signals built from retargeting and Customer Match data accelerate the AI’s bottom-of-funnel learning; splitting by intent stage prevents awareness-phase copy from diluting the conversion-focused group’s signal.[3][44]

Worked example

Competitor Group for a Cloud Storage Platform

  • Setup: An Australian cloud storage platform runs a Performance Max campaign at $10,000 per month in October 2026 and adds a third asset group specifically targeting competitor-aware demand.
  • Numbers: Competitor Group: 15 headlines written around differentiation themes (“Switch From [Competitor] Today”, “2× Storage, Same Price”, “Australian Data Sovereignty Guaranteed”), 5 long headlines (each up to 90 characters) comparing feature parity, 5 descriptions, 5 landscape images (1200×628 px) showing comparison tables, final URL = /switch-from-cloud-competitor. Audience signal: custom segment built from 4 competitor brand URLs + keyword list of 22 competitor-brand search terms. No Customer Match added (this group targets prospects not yet in CRM). Budget share: this group receives approximately $1,500/month of the $10,000 campaign budget in early weeks; expected conversions at 2% conversion rate and $80 CPC = $1,500 ÷ $80 × 0.02 = approximately 0.375 conversions per day, or ~11 per month — below the 30-conversion threshold for the first 2–3 months.[15]
  • Decision: Set a minimum 60-day evaluation window (not 30 days) before making structural decisions on the competitor group, given projected conversion volume of 11–22 per month over the first 2 months. Do not pause or merge before 60 days.
  • Why: Competitor groups typically have narrower audience pools and lower initial conversion volume than keyword groups; applying the same 30-day evaluation window would generate a premature pause decision based on statistically insufficient data.[15][69]

8. How Many Asset Groups Should You Run?

Google’s platform supports up to 100 asset groups per campaign.[3] In practice, most accounts should stay well below 10 until they have evidence that individual themes can sustain sufficient conversion volume independently. The key constraint is not creative capacity — it is budget and data. All asset groups inside a Performance Max campaign share the campaign budget. More groups means smaller slices per group, which means slower learning per group.[3][5]

Vendor guidance in 2026 suggests 3–8 asset groups for lead generation accounts and 5–15 for ecommerce accounts with large catalogues.[72] A separate 2026 source recommends starting at 1–3 groups for new accounts.[70] These figures are vendor claims, not independently verified benchmarks, and they are subject to the conversion-volume constraint above. The most defensible rule is: do not create an asset group you cannot feed 20–30 conversions per month.[7][15]

A practical decision framework:

  • New account or new campaign (month 1–2): Start with 1–3 asset groups. Build each to maximum creative coverage. Let the AI learn before splitting.
  • Established account with 60+ conversions per month: Split into keyword, intent and competitor groups as described above, ensuring each projected group will receive at least 20 conversions per month at current spend.
  • Large ecommerce account with 200+ conversions per month: Expand to 5–10 groups by product category, each linked to a scoped listing group and page feed. Monitor that no individual group falls below 20 conversions per month for two consecutive months.
  • Any group generating fewer than 20 conversions per month for 8 consecutive weeks: Consolidate into the nearest thematically related group rather than waiting indefinitely.[7][15]

Worked example

Calculating the Right Number of Asset Groups for a Lead Gen Account

  • Setup: A Gold Coast solar installation company is spending $5,000 per month on Performance Max in November 2026, generating 65 conversions per month total (enquiry form completions). They are considering splitting from 2 asset groups to 5.
  • Numbers: Current setup: 2 groups × $2,500 average spend each. Current conversions: 65 ÷ 2 = 32.5 per group per month — above the 30-conversion threshold.[15] Proposed setup: 5 groups × $1,000 average spend each. Projected conversions: 65 ÷ 5 = 13 per group per month — below the 20-conversion minimum vendor threshold.[7] To justify 5 groups, the account would need 5 × 30 = 150 total conversions per month, which would require approximately $5,000 × (150 ÷ 65) = $11,538 per month in budget at current conversion efficiency.
  • Decision: Maintain 2 asset groups. Do not expand to 5 until monthly budget reaches $11,500 or total campaign conversions exceed 150 per month — whichever occurs first. The 2 existing groups are structured as: (1) keyword group targeting “solar panels Gold Coast” and related installation terms, (2) intent group targeting energy-cost-aware homeowners via in-market and interest signals.
  • Why: Adding 3 additional groups would reduce average conversions per group to 13 per month, below the 20-conversion floor needed for the AI to optimise efficiently, increasing the risk of extended learning stalls across all 5 groups simultaneously.[7][15]

9. Audience Signals, Final URLs and Page Feeds

Audience signals, final URLs and page feeds are the three levers that connect each asset group to a specific buyer. Used together coherently, they give Google’s AI a precise starting point and a constrained destination. Used carelessly — or left at defaults — they allow the AI to serve any asset combination to any user on any page of your site, which dilutes all three dimensions of relevance simultaneously.[3][11]

Audience Signals

An audience signal is a suggestion, not a hard boundary. Google may serve an ad to a user outside the signal if it predicts that user will convert. The signal’s primary value is accelerating early learning: it gives the AI a seed population of likely converters from which to build its model.[3][44] Build audience signals from the most specific data available, in descending order of specificity: Customer Match lists (past purchasers, enquirers, email subscribers), website retargeting audiences, custom intent segments (built from keywords and competitor URLs), and then Google-defined in-market or interest audiences as a supplementary layer.[3][2]

Final URLs and Final URL Expansion

Each asset group has exactly one final URL. This is the default landing page for that group’s traffic. If final URL expansion is enabled (the default), Google may also send traffic to other pages on your site when it predicts a better conversion match. This can increase reach but reduces landing page control. For themed asset groups, the recommended approach is to set the final URL to the most relevant category or conversion page, then use a page feed to constrain which other pages the AI can select — rather than allowing full-site expansion.[11][12]

Page Feeds

A page feed is a structured list of URLs you supply to Performance Max to define the set of eligible landing pages for that campaign. When a page feed is attached and URL expansion is enabled, Google selects from the feed’s URLs rather than crawling your entire domain. This is particularly valuable for large ecommerce sites where many pages (e.g. out-of-stock product pages, blog posts, account login pages) should never receive paid traffic.[11]

Worked example

Configuring Audience Signals and a Page Feed for a Retail Ecommerce Group

  • Setup: A Sydney-based homewares ecommerce retailer has a Performance Max campaign at $15,000 per month in December 2026, with 3 asset groups. Asset Group 2 is a keyword group for “premium bed linen” with a final URL of /bed-linen/premium.
  • Numbers: Audience signal composition: (1) Customer Match list — 8,400 past purchasers of bed linen (uploaded July 2026, match rate 62% = 5,208 matched users); (2) website retargeting — users who viewed /bed-linen/* in the last 30 days (estimated 3,200 users per month); (3) custom intent segment — keywords: “premium cotton sheets”, “1000 thread count”, “bamboo bed linen Australia” (22 keywords total) + 3 competitor bedding retailer URLs. Total page feed for this group: 34 URLs restricted to /bed-linen/* and /checkout/* — excluding /blog/*, /about/*, /returns/* and all out-of-stock product pages. Final URL expansion: enabled but restricted to page feed only.
  • Decision: Set audience signal with Customer Match as primary layer, retargeting as secondary, custom intent as tertiary. Upload page feed restricting eligible pages to 34 URLs. Set tROAS at 3.5x (bottom of the vendor-claimed home goods range of 2.5x–3.5x[69]) for the first 4 weeks, then review after the 30-conversion threshold is reached.[15]
  • Why: Combining a Customer Match-weighted signal with a page feed-constrained URL set gives Google the most specific possible seed population and destination set, reducing the probability of spend on irrelevant pages or audiences during the 3–4 week learning period.[3][11][69]

10. Building a Strong Asset Group: A Checklist

The following checklist consolidates the specifications, thresholds and structural principles from this article into an actionable build sequence. Complete items in order — do not launch an asset group that fails a minimum-requirement row.

Creative Assets

  • Headlines: Write 15 headlines (maximum), each up to 30 characters. Vary the angle: include category-direct headlines, benefit headlines and a short punchy headline of 15 characters or fewer. All 15 should be coherent with the group’s single theme.
  • Long headlines: Write 5 long headlines (maximum), each up to 90 characters. Use these to expand on the core value proposition that cannot fit in 30 characters.
  • Descriptions: Write 5 descriptions. First description up to 60 characters; remaining 4 up to 90 characters each. Aim for at least 3 distinct angles (offer, proof, urgency/CTA).
  • Business name: Confirm this is set correctly at campaign level (up to 25 characters). Do not vary it across groups.
  • Call to action: Select the most conversion-aligned CTA from Google’s list. Use automated only if no specific CTA maps well to the group’s conversion goal.

Images

  • Landscape (1.91:1): Supply at least 4, up to 20. Recommended 1200×628 px. JPG or PNG, max 5 MB. Subject within centre 80%.[1]
  • Square (1:1): Supply at least 4, up to 20. Recommended 1200×1200 px. Same file constraints.[1]
  • Portrait (4:5): Supply at least 2, up to 20. Recommended 960×1200 px. Required for full Discover and YouTube Shorts inventory.[1]
  • All images must be on-theme for this specific asset group. Do not reuse generic brand imagery across all groups if it conflicts with a group’s messaging angle.

Videos

Share the Post:

Related Posts