Performance Marketing skill 07 of 13
Performance Max and Shopping SKILL.md
The Performance Max and Shopping skill makes Claude, Codex or another AI agent set up and improve Performance Max and Standard Shopping. It fixes Merchant Center feeds and custom labels, sets revenue, profit and new-customer goals, structures campaigns by margin and category, builds asset groups with search themes and signals, applies brand, keyword, placement and customer exclusions, reads channel-level reports and tests PMax against alternatives.
performance-max-shopping/SKILL.md, version 1.0.0, 1,584 words.
By Afeez Adebayo, Marketing, SEO & Growth Expert with 10+ years of experience.
Created 10 October 2026. Updated 10 October 2026. Free to use under CC BY 4.0.
Performance Max (PMax) runs one campaign across Search, Shopping, YouTube, Display, Discover, Gmail and Maps, with Google's AI choosing placements, audiences and bids. For years it was a black box. That has changed: Google added channel-level performance reporting (rolling out from late 2025, with data from June 2025 onward), campaign-level negative keywords, brand exclusions and experiments. [Reported from Google Ads announcements] The levers are back; using them well is the job.
This skill sets up, audits and improves Performance Max and Standard Shopping for e-commerce and lead generation: feed quality, goals and values, asset groups, signals, exclusions, channel reporting, new-customer focus and tests against alternatives.
Operating principles
- The feed is the campaign. For retail, product titles, attributes, images, prices and availability decide where you show.
- Value-based goals. Pass revenue or profit; use new-customer goals where acquisition matters.
- Exclude what you don't want to pay for: brand traffic, existing customers (when goals are acquisition), poor placements, irrelevant queries.
- Read channel reports. Know how much goes to Search, Shopping, YouTube and Display.
- Test against alternatives. Standard Shopping, Search and Demand Gen can outperform PMax for some accounts.
Feed essentials (Merchant Center)
| Attribute | Guidance |
|---|---|
| Title | Brand + product type + key attributes (size, colour, material, model) in the words shoppers search |
| Product type and Google category | Accurate and specific |
| Images | Clear main image on plain background; lifestyle images as additional |
| Price and availability | Match the landing page exactly, updated as often as they change |
| GTIN, brand, MPN | Present where they exist |
| Custom labels | Margin band, best-sellers, seasonality, stock level, for campaign splits |
| Shipping and returns | Set in Merchant Center for eligibility and trust |
When to use, and when not to
Use it for retail and e-commerce advertisers with product feeds, lead-generation advertisers using PMax with offline values, when PMax spend looks opaque, when brand or existing customers dominate PMax results, and when deciding between PMax and Standard Shopping.
Don't use it for pure Search accounts (google-search-ads) or for overall budget allocation (media-planning-budgeting).
Inputs to ask for, in one message
- Merchant Center diagnostics and feed export.
- PMax and Shopping campaign exports including asset groups, channel performance, search term insights and placements.
- Conversion goals, values and new-customer settings.
- Margins by product or category; custom label structure.
- Customer lists for exclusions or new-customer goals.
Raw SKILL.md file
This is the exact file agents read, including the credit header and licence notice. Save it as performance-max-shopping/SKILL.md.
--- name: performance-max-shopping description: "Set up and improve Performance Max and Standard Shopping: Merchant Center feed quality and custom labels, revenue, profit and new-customer goals, structure by margin and category, asset groups with search themes and audience signals, brand exclusions, negative keywords and placement and customer exclusions, channel-level reporting, and experiments against Standard Shopping and Search. Use whenever someone asks about Performance Max, PMax, Google Shopping ads, Merchant Center feeds, product titles for ads, PMax brand cannibalisation, channel reporting or new-customer acquisition goals." license: CC-BY-4.0 metadata: author: Afeez Adebayo author_url: https://afeezadebayo.com author_linkedin: https://www.linkedin.com/in/afeez-adebayo/ author_x: https://x.com/za_icon source: https://afeezadebayo.com/skills-md/marketing/performance-marketing/performance-max-shopping/ category: marketing subcategory: performance-marketing version: "1.0.0" created: "2026-10-10" updated: "2026-10-10" --- # Performance Max and Shopping Performance Max (PMax) runs one campaign across Search, Shopping, YouTube, Display, Discover, Gmail and Maps, with Google's AI choosing placements, audiences and bids. For years it was a black box. That has changed: Google added channel-level performance reporting (rolling out from late 2025, with data from June 2025 onward), campaign-level negative keywords, brand exclusions and experiments. [Reported from Google Ads announcements] The levers are back; using them well is the job. This skill sets up, audits and improves Performance Max and Standard Shopping for e-commerce and lead generation: feed quality, goals and values, asset groups, signals, exclusions, channel reporting, new-customer focus and tests against alternatives. ## Operating principles 1. **The feed is the campaign.** For retail, product titles, attributes, images, prices and availability decide where you show. 2. **Value-based goals.** Pass revenue or profit; use new-customer goals where acquisition matters. 3. **Exclude what you don't want to pay for:** brand traffic, existing customers (when goals are acquisition), poor placements, irrelevant queries. 4. **Read channel reports.** Know how much goes to Search, Shopping, YouTube and Display. 5. **Test against alternatives.** Standard Shopping, Search and Demand Gen can outperform PMax for some accounts. ## Feed essentials (Merchant Center) | Attribute | Guidance | |---|---| | Title | Brand + product type + key attributes (size, colour, material, model) in the words shoppers search | | Product type and Google category | Accurate and specific | | Images | Clear main image on plain background; lifestyle images as additional | | Price and availability | Match the landing page exactly, updated as often as they change | | GTIN, brand, MPN | Present where they exist | | Custom labels | Margin band, best-sellers, seasonality, stock level, for campaign splits | | Shipping and returns | Set in Merchant Center for eligibility and trust | ## When to use, and when not to Use it for retail and e-commerce advertisers with product feeds, lead-generation advertisers using PMax with offline values, when PMax spend looks opaque, when brand or existing customers dominate PMax results, and when deciding between PMax and Standard Shopping. Don't use it for pure Search accounts (`google-search-ads`) or for overall budget allocation (`media-planning-budgeting`). ## Inputs to ask for, in one message 1. Merchant Center diagnostics and feed export. 2. PMax and Shopping campaign exports including asset groups, channel performance, search term insights and placements. 3. Conversion goals, values and new-customer settings. 4. Margins by product or category; custom label structure. 5. Customer lists for exclusions or new-customer goals. ## Workflow ### Step 1. Fix the feed Resolve disapprovals and warnings; rewrite titles for top products; add missing attributes; set custom labels for margin and performance. ### Step 2. Set goals and values Revenue or profit values; new-customer acquisition goal (bid higher for new customers or new customers only) where the business wants growth; check conversion duplicates. ### Step 3. Structure campaigns Split by margin band, priority categories or markets using custom labels; avoid too many campaigns that thin out data. ### Step 4. Build asset groups Group by theme or category, with relevant headlines, descriptions, images and videos; add search themes and audience signals (customer lists, website visitors, custom segments) as hints. ### Step 5. Add exclusions Brand exclusions (when brand is covered elsewhere), negative keywords for irrelevant queries, placement exclusions, customer list exclusions for acquisition goals. ### Step 6. Read channel performance Check spend and conversions by channel; question large Display or video spend with weak results; compare Search spend inside PMax with your Search campaigns. ### Step 7. Test Use experiments: PMax versus Standard Shopping, PMax with and without brand, feed changes. Judge on profit and new customers. ### Step 8. Monitor Weekly feed diagnostics, search term insights, channel reports, asset performance and new-customer share. ## Channel report questions | Observation | Question | Possible action | |---|---|---| | Most spend on Search, mostly brand | Is PMax harvesting brand demand? | Brand exclusions; measure brand incrementality | | High Display spend, low conversions | Are placements low quality? | Placement exclusions; asset review | | YouTube spend with view-through credit | Is video adding sales? | Lift or geo test | | Shopping strong, others weak | Would Standard Shopping do as well with more control? | Run an experiment | ## Custom label strategy | Label | Values | Use | |---|---|---| | custom_label_0: margin | high, medium, low | Separate campaigns or ROAS targets by margin | | custom_label_1: performance | hero, steady, zombie (no impressions in 30 days) | Give zombies their own budget to get exposure | | custom_label_2: season | spring, summer, autumn, winter, evergreen | Shift budgets with the season | | custom_label_3: stock | in-stock-deep, low-stock | Avoid pushing items about to sell out | | custom_label_4: price band | under-25, 25-75, over-75 | Different targets for different order values | ## Title formula examples | Category | Formula | Example | |---|---|---| | Apparel | Brand + gender + product + material + colour + size | Acme Women's Waterproof Running Jacket, Recycled Nylon, Navy, M | | Electronics | Brand + product + model + key spec | Acme Wireless Headphones X200, Noise Cancelling, 40h Battery | | Home | Brand + product + material + dimensions | Acme Oak Dining Table, 180 cm, Seats 6 | ## Experiment ideas 1. PMax versus Standard Shopping for a product subset, judged on profit. 2. PMax with brand exclusions versus without, judged on total brand-plus-non-brand profit. 3. New-customer acquisition goal on versus off, judged on new customers and payback. 4. Feed title rewrite for top products versus original (before/after with control products). 5. Asset groups by category versus by audience theme. ## Weekly checklist - Merchant Center disapprovals and price mismatches. - Channel performance shifts by spend and conversions. - Search term insights for irrelevant themes (add negatives). - Asset group and asset performance. - Share of conversions from new customers. - Budget-limited campaigns that should scale. ## Edge cases and judgement calls | Situation | What to do | |---|---| | Small catalogue, small budget | One or two campaigns; strong feed and assets; simple goals | | Very large catalogue | Custom labels for margin and performance tiers; zombie products in their own campaign | | Lead generation PMax | Only with offline qualified values; otherwise it finds cheap low-quality leads | | Brand-heavy results | Brand exclusions plus separate brand Search campaign | | Seasonality and promotions | Use seasonality adjustments for short, predictable conversion rate changes | ## Handoffs to related skills - Tracking and values: `conversion-tracking-capi` - Search campaigns alongside PMax: `google-search-ads` - Customer lists and new-customer goals: `audience-first-party-data` - Video and image assets: `ad-creative-testing` - Incrementality: `incrementality-testing` - Product data for AI shopping surfaces: the AEO & GEO `ai-commerce-product-visibility` skill ## Worked example A homeware retailer put most of its Google budget into one PMax campaign with ROAS above target. Channel reporting showed a large share of spend going to Search, mostly brand queries, and customer list analysis showed many conversions were returning customers. The fix: brand exclusions with a separate brand Search campaign, a new-customer acquisition goal with a higher value for new customers, custom labels splitting high and low margin products, rewritten titles for the top 300 products and an experiment against Standard Shopping for low-margin lines. Reported ROAS fell, but new customers and contribution margin rose. ## Output format ``` # Performance Max and Shopping review: <account>, <date> ## Feed health and fixes ## Goals, values and new-customer settings ## Campaign and asset group structure ## Exclusions (brand, negatives, placements, customers) ## Channel performance findings ## Test plan and monitoring ``` ## Common mistakes - Weak feeds with generic titles. - Letting PMax take credit for brand and existing customers. - Ignoring channel reports. - Too many campaigns with too little data each. - Lead-gen PMax bidding on raw form fills. ## Questions to ask before you change campaigns - Which products make money after ad costs? - Does the business want new customers or total revenue? - Is brand search covered elsewhere? ## Self-check before you answer - Is the feed complete and accurate for top products? - Are values and new-customer goals set correctly? - Are brand, customers, placements and queries governed with exclusions? - Are channel reports read and alternatives tested? ## Official documentation - About Performance Max campaigns: https://support.google.com/google-ads/answer/10724817 - Merchant Center product data specification: https://support.google.com/merchants/answer/7052112 - Customer acquisition goal: https://support.google.com/google-ads/answer/12080169 --- Created by Afeez Adebayo, Marketing, SEO & Growth Expert Website: https://afeezadebayo.com LinkedIn: https://www.linkedin.com/in/afeez-adebayo/ X: https://x.com/za_icon Source: https://afeezadebayo.com/skills-md/marketing/performance-marketing/performance-max-shopping/ Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and adapt this skill, including for commercial use, as long as you credit Afeez Adebayo and link to the source URL above. Keep this notice in any copy or adapted version.
Workflow
Step 1. Fix the feed
Resolve disapprovals and warnings; rewrite titles for top products; add missing attributes; set custom labels for margin and performance.
Step 2. Set goals and values
Revenue or profit values; new-customer acquisition goal (bid higher for new customers or new customers only) where the business wants growth; check conversion duplicates.
Step 3. Structure campaigns
Split by margin band, priority categories or markets using custom labels; avoid too many campaigns that thin out data.
Step 4. Build asset groups
Group by theme or category, with relevant headlines, descriptions, images and videos; add search themes and audience signals (customer lists, website visitors, custom segments) as hints.
Step 5. Add exclusions
Brand exclusions (when brand is covered elsewhere), negative keywords for irrelevant queries, placement exclusions, customer list exclusions for acquisition goals.
Step 6. Read channel performance
Check spend and conversions by channel; question large Display or video spend with weak results; compare Search spend inside PMax with your Search campaigns.
Step 7. Test
Use experiments: PMax versus Standard Shopping, PMax with and without brand, feed changes. Judge on profit and new customers.
Step 8. Monitor
Weekly feed diagnostics, search term insights, channel reports, asset performance and new-customer share.
Channel report questions
| Observation | Question | Possible action |
|---|---|---|
| Most spend on Search, mostly brand | Is PMax harvesting brand demand? | Brand exclusions; measure brand incrementality |
| High Display spend, low conversions | Are placements low quality? | Placement exclusions; asset review |
| YouTube spend with view-through credit | Is video adding sales? | Lift or geo test |
| Shopping strong, others weak | Would Standard Shopping do as well with more control? | Run an experiment |
Custom label strategy
| Label | Values | Use |
|---|---|---|
| custom_label_0: margin | high, medium, low | Separate campaigns or ROAS targets by margin |
| custom_label_1: performance | hero, steady, zombie (no impressions in 30 days) | Give zombies their own budget to get exposure |
| custom_label_2: season | spring, summer, autumn, winter, evergreen | Shift budgets with the season |
| custom_label_3: stock | in-stock-deep, low-stock | Avoid pushing items about to sell out |
| custom_label_4: price band | under-25, 25-75, over-75 | Different targets for different order values |
Title formula examples
| Category | Formula | Example |
|---|---|---|
| Apparel | Brand + gender + product + material + colour + size | Acme Women's Waterproof Running Jacket, Recycled Nylon, Navy, M |
| Electronics | Brand + product + model + key spec | Acme Wireless Headphones X200, Noise Cancelling, 40h Battery |
| Home | Brand + product + material + dimensions | Acme Oak Dining Table, 180 cm, Seats 6 |
Experiment ideas
- PMax versus Standard Shopping for a product subset, judged on profit.
- PMax with brand exclusions versus without, judged on total brand-plus-non-brand profit.
- New-customer acquisition goal on versus off, judged on new customers and payback.
- Feed title rewrite for top products versus original (before/after with control products).
- Asset groups by category versus by audience theme.
Weekly checklist
- Merchant Center disapprovals and price mismatches.
- Channel performance shifts by spend and conversions.
- Search term insights for irrelevant themes (add negatives).
- Asset group and asset performance.
- Share of conversions from new customers.
- Budget-limited campaigns that should scale.
Edge cases and judgement calls
| Situation | What to do |
|---|---|
| Small catalogue, small budget | One or two campaigns; strong feed and assets; simple goals |
| Very large catalogue | Custom labels for margin and performance tiers; zombie products in their own campaign |
| Lead generation PMax | Only with offline qualified values; otherwise it finds cheap low-quality leads |
| Brand-heavy results | Brand exclusions plus separate brand Search campaign |
| Seasonality and promotions | Use seasonality adjustments for short, predictable conversion rate changes |
Handoffs to related skills
- Tracking and values:
conversion-tracking-capi - Search campaigns alongside PMax:
google-search-ads - Customer lists and new-customer goals:
audience-first-party-data - Video and image assets:
ad-creative-testing - Incrementality:
incrementality-testing - Product data for AI shopping surfaces: the AEO & GEO
ai-commerce-product-visibilityskill
Worked example
A homeware retailer put most of its Google budget into one PMax campaign with ROAS above target. Channel reporting showed a large share of spend going to Search, mostly brand queries, and customer list analysis showed many conversions were returning customers. The fix: brand exclusions with a separate brand Search campaign, a new-customer acquisition goal with a higher value for new customers, custom labels splitting high and low margin products, rewritten titles for the top 300 products and an experiment against Standard Shopping for low-margin lines. Reported ROAS fell, but new customers and contribution margin rose.
Output format
# Performance Max and Shopping review: <account>, <date>
## Feed health and fixes
## Goals, values and new-customer settings
## Campaign and asset group structure
## Exclusions (brand, negatives, placements, customers)
## Channel performance findings
## Test plan and monitoringCommon mistakes
- Weak feeds with generic titles.
- Letting PMax take credit for brand and existing customers.
- Ignoring channel reports.
- Too many campaigns with too little data each.
- Lead-gen PMax bidding on raw form fills.
Questions to ask before you change campaigns
- Which products make money after ad costs?
- Does the business want new customers or total revenue?
- Is brand search covered elsewhere?
Self-check before you answer
- Is the feed complete and accurate for top products?
- Are values and new-customer goals set correctly?
- Are brand, customers, placements and queries governed with exclusions?
- Are channel reports read and alternatives tested?
Official documentation
- About Performance Max campaigns: https://support.google.com/google-ads/answer/10724817
- Merchant Center product data specification: https://support.google.com/merchants/answer/7052112
- Customer acquisition goal: https://support.google.com/google-ads/answer/12080169
On this page
- When to use this skill, and when not to
- Example prompts
- Example output
- What the agent needs from you
- How this skill hands off to related skills
- How to run Performance Max and Shopping step by step
- Situations this skill handles
- What makes this skill different
- Practical tips from real projects
- Frequently asked questions
- Changelog
When to use this skill, and when not to
Use it for retailers with product feeds, lead-generation advertisers using PMax with offline values, when PMax spend looks opaque, when brand or existing customers dominate PMax results and when deciding between PMax and Standard Shopping.
Performance Max used to be a black box. Google has since added channel-level performance reporting, campaign-level negative keywords, brand exclusions and experiments, so advertisers have real levers again. Using them well, starting with the feed, is the job.
Don’t use it for pure Search accounts (use Google Search Ads) or overall budget allocation (use Media Planning).
Example prompts
- “PMax spends mostly on brand. What should we change?”
- “Audit our Merchant Center feed and rewrite titles for top products.”
- “Set up a new-customer acquisition goal in PMax.”
- “Should we run Standard Shopping alongside PMax?”
- “Read our PMax channel report and tell me what to fix.”
Example output
The skill returns a fixed structure. A shortened example for an imaginary site:
# PMax and Shopping review: example homeware retailer, 10 October 2026 ## Channel report Search is the largest share of PMax spend, mostly brand queries; many conversions from returning customers ## Fixes 1. Brand exclusions + separate brand Search campaign 2. New-customer acquisition goal, higher value for new customers 3. Custom labels: margin high/medium/low; campaigns split by margin 4. Titles rewritten for top 300 products 5. Experiment: Standard Shopping for low-margin lines
Results are judged on profit and new customers, because PMax can look efficient while harvesting brand demand and existing buyers.
What the agent needs from you
The skill asks for Merchant Center diagnostics and a feed export, PMax and Shopping exports including asset groups, channel performance, search term insights and placements, conversion goals, values and new-customer settings, margins by product or category and customer lists for exclusions or acquisition goals.
How this skill hands off to related skills
PMax depends on several skills. Values come from Conversion Tracking, Search campaigns from Google Search Ads, customer lists from First-Party Audiences, assets from Ad Creative Testing and incrementality from Incrementality Testing. Product data for AI shopping answers continues in the AEO and GEO AI Commerce skill.
How to run Performance Max and Shopping step by step
- Install the skill as
performance-max-shopping/SKILL.md. - Export Merchant Center diagnostics and the feed.
- Export PMax data: asset groups, channel performance, search term insights and placements.
- Share margins and customer lists.
- Review feed fixes, goals, structure, exclusions and channel findings.
- Run experiments and the weekly checklist.
Situations this skill handles
PMax problems follow patterns. The skill handles these situations:
| Situation | What the skill does |
|---|---|
| Mostly brand spend inside PMax | Brand exclusions and a separate brand campaign |
| High Display spend, few sales | Placement exclusions and asset review |
| Returning customers dominate | New-customer goal and customer list exclusions |
| Very large catalogue | Custom labels for margin and performance tiers |
| Products with no impressions | Zombie products in their own campaign |
| Short promotions | Seasonality adjustments for predictable conversion changes |
What makes this skill different
Most PMax advice says to feed it assets and trust it. This skill starts with the feed, sets goals that reflect profit and new customers, uses every exclusion now available, reads the channel report Google added and tests PMax against Standard Shopping and Search instead of assuming it wins.
Reported ROAS often falls when brand and existing customers are excluded; new customers and contribution margin usually rise. The skill judges on the second set of numbers.
Practical tips from real projects
- The feed is the campaign.
- Exclude brand if brand Search covers it.
- Split by margin with custom labels.
- Read the channel report weekly.
A worked example: a homeware retailer put most of its Google budget into one PMax campaign with ROAS above target. Channel reporting showed a large share of spend going to Search, mostly brand queries, and many conversions came from returning customers. Brand exclusions with a separate brand campaign, a new-customer acquisition goal, custom labels splitting high- and low-margin products, rewritten titles for the top 300 products and an experiment against Standard Shopping for low-margin lines lowered reported ROAS but raised new customers and contribution margin.
Feed essentials cover titles (brand, product type and key attributes in shoppers’ words), accurate categories, clear images, prices and availability that match the page, GTINs, custom labels and shipping and returns set in Merchant Center. Title formulas differ by category: apparel adds gender, material, colour and size; electronics adds model and key specification; home adds material and dimensions.
Custom labels turn the feed into a strategy: margin bands, performance tiers (hero, steady, zombie), season, stock depth and price band let campaigns and targets reflect what each product earns, rather than treating the catalogue as one block.
Channel report questions guide action: if most spend is Search on brand terms, PMax may be harvesting existing demand; if Display spend is high with weak results, placements need work; if YouTube claims view-through credit, a lift test should confirm it; and if Shopping alone performs, Standard Shopping deserves an experiment.
Experiment ideas include PMax versus Standard Shopping on a product subset, PMax with and without brand, the new-customer goal on and off, and feed title rewrites against control products.
Asset groups work best organised by theme or category, with headlines, descriptions, images and videos that fit that group. Search themes and audience signals, such as customer lists, website visitors and custom segments, act as hints rather than hard targeting; they help the campaign start in the right place, while creative quality and goals decide where it ends up. The weekly checklist then covers Merchant Center disapprovals and price mismatches, channel shifts, search term insights, asset performance, new-customer share and budget-limited campaigns.
Frequently asked questions
Is Performance Max still a black box?
Less than before. Channel-level reporting, campaign negative keywords, brand exclusions and experiments now give advertisers visibility and control.
How do I stop PMax spending on brand?
Add brand exclusions to PMax and run brand in a separate Search campaign, then measure total brand and non-brand profit.
What is the new-customer acquisition goal?
A setting that bids higher for new customers, or only for new customers, using your customer lists and conversion data.
Why does the feed matter so much?
For retail, product titles, attributes, images, prices and availability decide which searches and placements your products appear in.
Should I run Standard Shopping too?
Test it. For some products and margins, Standard Shopping’s control beats PMax; experiments show which.
Can PMax work for lead generation?
Only with offline imports of qualified outcomes and values; otherwise it tends to find cheap, low-quality leads.
Changelog
- 1.0.0, 10 October 2026: first release.
Use, adapt and share this skill under CC BY 4.0: credit Afeez Adebayo and link to this page.
All Performance Marketing skills
- Paid Media Audit
- Media Planning and Budget Allocation
- Conversion Tracking, Consent Mode and Conversions API
- Incrementality Testing: Lift and Geo Experiments
- Marketing Mix Modelling
- Google Search Ads and AI Max
- Meta Ads (Facebook and Instagram)
- TikTok, LinkedIn and Other Paid Social
- Ad Creative Strategy and Testing
- Landing Pages and Conversion Rate Optimisation for Paid Traffic
- Audiences and First-Party Data for Paid Media
- Paid Media Reporting and Attribution
