Google Ads Campaign Structure in 2026: The Catalog-Size Playbook for Shopping, Performance Max, AI Max, and Demand Gen (Plus LLM-Shopping Optimization)

By /

Updated: May 2026 · Author: Ankit Agarwal, Founder, iComchain LLC · Reading time: 19 minutes · Audience: E-commerce founders, DTC brands, and Google Ads operators spending $5K-$200K/month who want maximum visibility at minimum cost across both Google and the new LLM shopping engines

TL;DR. Google Ads in 2026 has four campaign types that matter for e-commerce — Standard Shopping, Performance Max, AI Max for Search, and Demand Gen — and the right structure depends on three things: catalog size, margin profile, and whether your products are LLM-ready. The biggest shift this year: ChatGPT pulls 83% of its Shopping carousel data directly from Google Merchant Center feeds, Perplexity Shop and Gemini AI Mode pull from similar pipelines, and Google’s own AI Mode uses Shopping signals to power product results. That means your GMC feed quality now determines visibility on five AI shopping engines simultaneously — but only if your feed has the new AI-critical fields most operators are missing (product_review_count, return_policy, popularity_score, video_link, structured shipping data). This guide covers the four campaign types, exact campaign structure for catalogs at every size band (under 100 SKUs through 2,000+), the AI Max for Search migration that replaces Dynamic Search Ads in September 2026, and the LLM-readiness checklist. I’m running all four campaign types on the iComchain account right now and will share the structural decisions (without specific numbers) we’ve learned about each. 30 FAQs at the end cover the questions we get most often from operators.

Quick facts
Campaign types worth running for e-com in 2026 Standard Shopping, Performance Max, AI Max for Search (new), Demand Gen
Biggest 2026 change Dynamic Search Ads, automatically created assets, and campaign-level broad match auto-upgrade to AI Max in September 2026
LLM shopping engines pulling from your GMC feed ChatGPT (83% via base64-decoded Shopping API), Perplexity Shop, Gemini AI Mode, Google AI Overviews, Apple Intelligence shopping
ChatGPT feed refresh frequency Every 15 minutes
AI-critical feed fields most brands miss product_review_count, average_rating, return_policy, structured shipping, popularity_score, video_link, model_3d_link
Recommended structure under 100 SKUs 1 PMax (all catalog) + 1 Brand Search + AI Max enabled
Recommended structure 100-500 SKUs 2-3 PMax (segmented by margin) + Brand Search + AI Max + DG retargeting
Recommended structure 500-2,000 SKUs Hybrid: PMax for top 80% revenue + Standard Shopping for control SKUs + AI Max + DG
Recommended structure 2,000+ SKUs PMax asset groups by category + Standard Shopping for hero SKUs + AI Max + DG + custom audience layering
Cost framework Maximum visibility = run all 4 types layered. Minimum cost = aggressive negative-keyword and brand-exclusion management on PMax + AI Max
Bottom line The accounts winning in 2026 treat their GMC feed as the most important asset (it powers ALL paid + ALL LLM visibility) and use PMax + AI Max + DG as orchestration layers on top.

The 4 Google Ads campaign types in 2026 — what each is actually for

Standard Shopping

The classic feed-driven Shopping campaign. You upload products to Merchant Center, Google serves them on Shopping ads inventory (Search results page, Shopping tab, Images, YouTube). You control: bidding, negative keywords, product groups, geographic targeting, schedule. Best for: transparency-required hero SKUs, brand-defending Shopping ads, controlled testing of feed changes, categories where PMax has historically wasted budget. Standard Shopping is no longer Google’s recommended default for most accounts but remains essential as a control layer alongside PMax.

Performance Max (PMax)

Google’s flagship automated campaign. Runs across Search, Shopping, YouTube, Discover, Gmail, Display, and Maps inventory simultaneously. You provide assets (text, images, video, signals), Google provides everything else. In 2024-2026 Google has materially increased advertiser controls: negative keywords (added 2024), brand exclusions (improved 2025), search themes (suggested keyword direction), asset group structure (segment by category/margin), audience signals, and account-level negative keywords. Best for: the engine of most modern e-commerce accounts. The question is no longer “should I use PMax” but “how should I structure it.”

AI Max for Search (the 2026 upgrade)

This is the big change for 2026. AI Max is not a separate campaign type — it’s a campaign-level setting that supercharges your existing Search campaigns. It includes three connected features: Search Term Matching (uses broad match + “keywordless” AI to find queries beyond your keywords), Asset Optimization (generates ad copy/headlines from your landing page and existing assets), and Final URL Expansion (sends users to the most relevant landing page automatically). Critical timeline: Dynamic Search Ads, automatically created assets, and campaign-level broad match auto-upgrade to AI Max in September 2026. If you’re still running DSA, plan the migration now.

Demand Gen (DG)

The 2023+ replacement for Discovery campaigns. Runs on YouTube (Shorts, in-stream, in-feed), Discover feed, and Gmail. Creative-driven, audience-targeted, optimized for upper-funnel demand creation. Best for: brand awareness, top-of-funnel customer acquisition, retargeting (lookalike from existing customers, affinity audiences, custom segments), and lifestyle/visual product categories. Not a primary conversion engine for most accounts but the missing piece in catalogs that have plateaued on PMax.

The LLM shopping shift — why your GMC feed now powers 5 engines, not 1

In 2024 your Google Merchant Center feed powered one ad surface: Google Shopping. In 2026 it powers at minimum five:

  1. Google Shopping ads (the classic surface)
  2. Google AI Mode and AI Overviews shopping carousels (when users ask Google’s AI for product recommendations)
  3. ChatGPT Shopping (researchers confirmed base64-encoded Google Shopping API parameters in ChatGPT’s source code; an estimated 83% of ChatGPT’s product carousel results pull from Google Shopping data)
  4. Perplexity Shop (Perplexity’s commerce surface — pulls from a combination of Google Shopping, direct merchant feeds, and partner integrations)
  5. Gemini AI Mode shopping (Google’s own LLM shopping experience, leveraging the same feed data that powers Shopping ads)

This is the single most consequential shift in e-commerce paid media in a decade. The feed you upload to Merchant Center for Google Shopping now determines whether your products get recommended when a user asks ChatGPT “what’s the best mid-range espresso machine under $400?” or asks Gemini “what hiking boots do reviewers recommend for wide feet?”

The feed you maintained casually in 2024 needs to be production-grade in 2026 because it has five times the surface area of leverage — and at zero additional cost beyond doing the work properly the first time.

The AI-critical feed fields most brands are missing

The basic feed fields (title, description, price, image, GTIN) are necessary but no longer sufficient. AI shopping engines preferentially surface products with the following enriched fields:

  • product_review_count and average_rating — AI engines explicitly weight social proof; products with reviews get cited 3-5x more often
  • return_policy — structured return policy data (window in days, who pays return shipping, restocking fees if any)
  • shipping — structured shipping cost and delivery-time data per region (not just “calculated at checkout”)
  • popularity_score — emerging field that signals demand; used by some AI engines to break ties between similar products
  • video_link — product video URL; AI shopping engines increasingly surface video alongside product cards
  • model_3d_link — 3D model URL for AR shopping; used by Apple Intelligence and Google’s AR shopping experiments
  • highlight — short bullet-style attribute callouts (5-10 per product)
  • product_detail — structured spec data (material, dimensions, color codes, etc.)
  • identifier_exists — explicitly set to true for branded products with GTINs
  • canonical_link — the canonical product URL (matters more in 2026 because LLM citations link to canonical URLs)

Most brands have 30-50% of these fields populated. Brands with 90-100% completion get materially better visibility on every AI shopping engine for the same ad spend.

What changed in the ChatGPT Shopping pipeline (2025-2026)

ChatGPT introduced its own product feed specification in late 2025 (separate from but compatible with Google Shopping format). The spec accepts JSONL (gzip-compressed), CSV (gzip), TSV (gzip), or Parquet with zstd. ChatGPT refreshes every 15 minutes — the fastest refresh cycle of any AI shopping engine. Brands that submit directly to ChatGPT (in addition to maintaining Google Shopping feed) can get 12-24 hours faster recommendation visibility for new products and price changes.

Perplexity has its own merchant API. Apple Intelligence Shopping uses a combination of Apple’s own product graph + partner feeds. Gemini AI Mode pulls from Google Shopping directly (no separate feed required).

The strategic point: maintaining a single perfect Google Shopping feed gets you 80% of the LLM shopping visibility for free. Maintaining additional ChatGPT and Perplexity direct feeds gets you the remaining 20%, which is increasingly important as those engines grow share.

Campaign structure by catalog size — the actual decision tree

This is the playbook. Pick the bracket that matches your SKU count and follow the structure. These structures assume you’re running e-commerce (not lead gen) and have at least one converting product.

Under 100 SKUs (micro catalog)

The structure:

  • 1× Performance Max campaign covering the entire catalog (single asset group with all products)
  • 1× Brand Search campaign (exact + phrase match on your brand name) to defend brand traffic and reduce PMax cannibalization
  • AI Max enabled on the Brand Search campaign (broaden brand-adjacent queries)
  • Skip Standard Shopping unless you have a specific reason
  • Skip Demand Gen until you have $10K+/month spend and converting customer data

Why this works: Under 100 SKUs, segmentation hurts more than it helps. PMax needs volume signals to optimize; splitting into multiple PMax campaigns starves each of data. A single PMax with strong audience signals + a defending Brand Search campaign covers 95% of your conversion potential.

The exception: if 80%+ of your revenue comes from 5-10 hero SKUs, run those as a Standard Shopping campaign with manual bidding control alongside the catalog PMax. This gives you transparency and control on the products that actually matter.

Common mistake: running 3-4 small PMax campaigns hoping for “better optimization.” With under 100 SKUs you don’t have enough conversion data to feed multiple campaigns; consolidate.

100-500 SKUs (small catalog)

The structure:

  • 2-3× Performance Max campaigns segmented by margin tier or category:
    • “PMax — High margin” (products at 40%+ margin, target ROAS aggressive)
    • “PMax — Standard” (products at 20-40% margin, target ROAS moderate)
    • “PMax — Low margin / clearance” (products under 20% margin, max conversions or volume-focused)
  • 1× Brand Search with AI Max enabled
  • 1× Demand Gen retargeting campaign (existing customer + cart abandoner audiences) once you have 100+ converting customers
  • Standard Shopping for hero SKUs (top 10% of revenue) if you need control or testing transparency

Why this works: at 100-500 SKUs, margin segmentation matters more than category segmentation for most accounts. Different margin tiers tolerate different ROAS targets, and PMax doesn’t optimize well when high-margin and low-margin products share the same campaign — Google’s algorithm will preferentially serve the cheapest products to maximize conversion volume, regardless of your margin reality.

The exception: if your catalog has clearly different audiences per category (e.g., men’s vs women’s apparel, indoor vs outdoor furniture), segment by category instead of margin. Audience signals work better when each PMax serves one coherent customer profile.

Common mistake: running one giant PMax with the entire catalog. You sacrifice ROAS optimization for simplicity, and the algorithm spends disproportionately on low-margin products because they convert at higher volume.

500-2,000 SKUs (mid catalog)

The structure:

  • 3-5× Performance Max campaigns segmented by margin × category combination
  • 1× Standard Shopping campaign for your top 50-100 hero SKUs with manual bid control
  • 1× AI Max-enabled Search campaign for non-brand high-intent queries (replacing your old DSA setup if you had one — required by September 2026 anyway)
  • 1× Brand Search with AI Max
  • 2-3× Demand Gen campaigns: customer retargeting + lookalikes, cart abandoners, optional top-funnel awareness

Why this works: at 500-2,000 SKUs you have enough conversion data to support 3-5 PMax campaigns each with their own optimization profile. The hybrid Standard Shopping layer for hero SKUs gives you control where control matters; PMax handles the long tail. AI Max for non-brand search expands query reach without you having to maintain massive keyword lists.

Common mistake: trying to manage 8-12 PMax campaigns “for granular control.” Beyond 5 PMax campaigns you fragment your conversion data and Google’s algorithm performance degrades. Consolidate.

2,000+ SKUs (large catalog)

The structure:

  • Asset group structure within fewer PMax campaigns rather than many separate PMax campaigns
  • 4-6× Performance Max campaigns each containing 5-15 asset groups segmented by sub-category or product line
  • Standard Shopping for hero SKUs and any products needing manual control (typically top 100-300 SKUs)
  • 1-2× AI Max-enabled Search campaigns covering non-brand intent
  • 1× Brand Search with AI Max
  • 3-5× Demand Gen campaigns layered for retargeting, cart abandonment, customer lookalikes, and top-funnel
  • Custom audience signals layered into PMax via Customer Match (your existing customer list) + similar audiences + first-party signal sources
  • Account-level brand exclusions and negative keywords applied account-wide (not per-campaign)

Why this works: at 2,000+ SKUs the binding constraint is asset group structure within PMax, not the number of PMax campaigns. Asset groups let you give Google specific creative + signals + listing groups for each sub-category while still pooling conversion data at the campaign level. Standard Shopping handles the controlled top SKUs. AI Max handles non-brand search reach. DG handles upper funnel and retargeting.

Common mistake: running 20+ PMax campaigns for “category-level granularity.” This fragments conversion data, makes performance review impossible, and underperforms a 4-6 campaign asset-group-driven structure.

Standard Shopping in 2026 — when to keep it, when to retire it

Standard Shopping is not deprecated, but its role has narrowed. Use it when you need transparency on specific SKUs (for hero products where you need to see exactly what queries trigger ads), manual bid control (bid more aggressively on products with high LTV or strategic value), testing feed changes (Standard Shopping shows the impact more cleanly than PMax), categories where PMax historically wasted budget, or brand-defending Shopping ads (running your own brand-name Shopping ads at low cost to defend against competitor takeover).

For most accounts, Standard Shopping ends up running 5-20% of the SKU catalog (the strategically important ones) while PMax runs 80-95%.

Performance Max best practices in 2026

PMax has matured significantly since its 2021 launch. The 2025-2026 controls that actually matter:

  • Negative keywords — apply to PMax now (added 2024). Build a negative keyword list for irrelevant queries (job listings, free, used, DIY, etc.) and apply at account level or campaign level.
  • Brand exclusions — exclude branded queries from PMax so Brand Search captures them at lower CPC. Critical for brands with material brand search volume.
  • Search themes — give Google directional guidance about which keyword categories you want PMax to target. Use 3-5 themes per asset group, treat them as guidance not certainty.
  • Asset groups — segment within a PMax campaign rather than splitting into multiple campaigns. Each asset group gets its own creative, audience signals, search themes, and listing group filter.
  • Audience signals — feed Google your Customer Match list, lookalike audiences, in-market segments, and demographic signals. Signals don’t restrict targeting but accelerate algorithm learning.
  • Listing group filters — exclude products from individual asset groups using labels you define in your Merchant Center feed (custom_label_0 through custom_label_4 are the standard fields for this).
  • Conversion value rules — assign different conversion values to different customer types (new vs returning, geographic location, device) to steer the algorithm without bid manipulation.

The PMax campaigns that win in 2026 use 5-7 of these controls. The ones that lose are still treating PMax as a “set it and forget it” black box.

AI Max for Search — the 2026 migration nobody can skip

If you’ve ever run Dynamic Search Ads (DSA), automatically created assets (ACA), or campaign-level broad match settings, your campaigns will auto-upgrade to AI Max in September 2026. Don’t wait for the forced migration; do it manually now to control the transition.

What AI Max actually does: Search Term Matching uses broad match plus “keywordless” AI to find queries beyond your existing keywords. Asset Optimization generates customized headlines and descriptions from your landing page content. Final URL Expansion sends users to the most relevant landing page on your site.

Performance: Google publishes 7% average lift in conversions/conversion value at similar CPA/ROAS when the full feature suite is enabled vs Search Term Matching alone.

Best practices: Roll out one feature at a time over 4-8 weeks; pair with Smart Bidding (Target CPA or Target ROAS); use brand controls and location controls aggressively; monitor search term reports daily for the first 14 days, weekly thereafter; add brand exclusions to prevent AI Max from cannibalizing your dedicated Brand Search campaign.

Common mistake: enabling AI Max on a campaign with a tiny budget. AI Max needs query volume and conversion data to optimize — if your Search campaign was getting 50 clicks/day, it’ll struggle with the full AI Max suite. Use Search Term Matching alone in low-volume contexts.

Demand Gen — the upper-funnel piece most accounts skip

Demand Gen replaced Discovery Ads in 2023. Runs on YouTube (in-stream, Shorts, in-feed), Discover feed, and Gmail Promotions/Social tabs. It’s audience- and creative-driven rather than keyword-driven.

When DG is worth running: $10K+/month total Google Ads spend, 100+ converting customers (need data for lookalike audiences), visual or lifestyle category, plateaued on PMax/Search and need top-funnel reach, video creative or willingness to produce it.

Common mistake: treating DG as a primary conversion channel. It’s an upper- and mid-funnel channel that pairs with PMax/Search. Direct-response ROAS on DG is typically lower than PMax — the value is in feeding the funnel.

What I’m running on the iComchain account — honest structural learnings

I run all four campaign types (Standard Shopping, Performance Max, AI Max for Search, Demand Gen) on the iComchain account. The structural decisions I’ve made — and what I’ve learned from running them in parallel:

  • All four types are worth running at the right account stage, but not all four are worth running at launch. The right sequence is PMax + Brand Search first, then layer in AI Max once Search has data, then add DG for retargeting once you have 100+ converters, then add Standard Shopping for hero-SKU control once you know which SKUs deserve manual control.
  • Feed quality is the bottleneck for everything. Every campaign type — Shopping, PMax, AI Max Shopping inventory, DG product feed extensions — pulls from the same Merchant Center feed. The hours I’ve spent improving feed completeness (not the campaign settings) have moved the needle far more than any campaign-level change.
  • PMax structure decisions matter more than I expected. Splitting one big PMax into 2-3 margin-segmented PMax campaigns produced cleaner per-campaign ROAS reporting and better algorithm behavior than one consolidated campaign.
  • AI Max needs guardrails. Without brand exclusions and tight location controls, AI Max will expand into queries that look adjacent but convert poorly. The first 30 days require daily search-term monitoring; after that the noise drops.
  • Standard Shopping isn’t dead. Even with PMax doing most of the work, having a Standard Shopping campaign for the top hero SKUs gives you visibility and control that PMax actively hides. I treat Standard Shopping as the “instrumentation layer” for the catalog.
  • DG is hard to evaluate honestly. Demand Gen attribution is messy — most of the value shows up in branded search lift and assisted conversions, not last-click. Set it up with realistic expectations or you’ll kill it after 30 days for the wrong reason.

The honest meta-learning: most operators (including me, early on) over-engineer their campaign structure and under-invest in feed quality. The accounts that win are the ones with production-grade feeds and disciplined-but-simple campaign structures, not the ones with elaborate 20-campaign architectures.

The “maximum visibility, minimum cost” framework

Maximum visibility means appearing across Google Shopping ads (PMax + Standard Shopping), Google Search ads (Brand Search + AI Max), Google Display/YouTube/Discover/Gmail (PMax + DG), Google AI Overviews / AI Mode shopping (driven by GMC feed), and ChatGPT Shopping / Perplexity Shop / Gemini AI Mode (driven by GMC feed + product schema). This requires running all four campaign types AND maintaining a production-grade feed.

Minimum cost means ruthlessly excluding wasteful spend: account-wide negative keyword list (50-200 negatives covering job/free/used/DIY/wholesale), brand exclusions on PMax and AI Max, listing group filters excluding low-performing or out-of-stock products, conversion value rules steering spend toward higher-LTV customer profiles, geographic exclusions for regions with poor unit economics, time-of-day bid adjustments where data supports them, asset group cleanup quarterly.

The accounts that achieve both — maximum visibility + minimum cost — typically run 5-10 campaigns total, have feeds with 90%+ field completion, and spend 4-6 hours per week on disciplined optimization. That’s the actual standard.

When iComchain helps — and when DIY is enough

  1. You’re spending $10K-$100K/month and your account has accumulated structural debt (12+ campaigns that should be 5, untracked negatives, missing AI Max migration, weak feed). We do a 90-minute account audit ($500-$1,000) and produce a written restructure plan you can execute yourself or with us.
  2. You’re entering Google Ads and want the right structure from day one. We set up the catalog-size-appropriate structure, build the negative keyword and exclusion lists, audit the feed for AI-critical fields. Typical engagement: $3,000-$8,000 setup project + optional ongoing optimization retainer.
  3. You want feed-only optimization for AI shopping visibility. We audit the GMC feed against the AI-critical fields list, build the missing schema, set up multi-platform feed distribution (Google Shopping + ChatGPT + Perplexity direct feeds where applicable). Typical engagement: $1,500-$4,000 fixed-fee project, 2-4 weeks.

If your situation matches one of these, message us on WhatsApp at +1 323 647 2657 or email hello@icomchain.com with: monthly Google Ads spend, catalog SKU count, and your three biggest current campaign frustrations. The first 30-minute call is free.

30 Questions Real Operators Ask About Google Ads Campaign Structure in 2026

Choosing campaign types

1. What’s the best Google Ads campaign type for e-commerce in 2026?
There’s no single best — the answer is a mix of all four (Standard Shopping, Performance Max, AI Max for Search, Demand Gen) layered based on your catalog size, spend level, and stage. PMax is the engine for most accounts; Brand Search defends brand traffic; AI Max replaces DSA; Demand Gen handles upper funnel and retargeting.

2. Should I run Performance Max or Standard Shopping?
Both, in most cases. PMax handles the bulk of conversion-driving spend for the long tail of your catalog; Standard Shopping handles your top hero SKUs (typically top 10-20% of revenue) where you need manual control, transparency, or aggressive bidding.

3. Do I need AI Max if I already have Performance Max?
Yes. PMax handles Shopping + automated multi-channel; AI Max enhances your Search campaigns specifically (text ads, dynamic landing pages, AI-generated copy). They serve different inventory and aren’t substitutes. AI Max replaces Dynamic Search Ads and is becoming mandatory in September 2026.

4. Is Demand Gen worth the budget if I’m under $10K/month spend?
Usually no. DG works best as an upper-funnel layer when you’ve already saturated PMax/Search at lower funnel. Below $10K/month spend, concentrate budget in PMax + Brand Search; add DG once you have 100+ converters.

5. Are Dynamic Search Ads still worth running in 2026?
No. DSA is being retired and auto-upgraded to AI Max in September 2026. Migrate proactively now to control the transition; don’t wait for the forced upgrade.

6. Can I run all four campaign types from day one?
Technically yes; strategically no. The right launch sequence is PMax + Brand Search first (covers 90% of conversion potential), add AI Max once Search has 30+ days of data, add DG once you have 100+ converters, add Standard Shopping for hero SKUs once you know which deserve manual control.

Catalog size structure

7. How many PMax campaigns should I run with under 100 SKUs?
One. Splitting a small catalog across multiple PMax campaigns starves each of conversion data. Single PMax + Brand Search + AI Max enabled covers 95% of conversion potential at this catalog size.

8. How should I segment PMax for a 100-500 SKU catalog?
Segment by margin tier (high margin / standard / clearance) rather than by category. Margin segmentation lets each campaign run its appropriate ROAS target. Use 2-3 PMax campaigns total at this size.

9. What about a 500-2,000 SKU catalog?
3-5 PMax campaigns segmented by margin × category, plus 1 Standard Shopping for top 50-100 hero SKUs, plus AI Max-enabled Search, plus Brand Search, plus 2-3 Demand Gen campaigns. The hybrid approach gives you control where control matters and lets PMax handle the long tail.

10. What changes for catalogs over 2,000 SKUs?
Asset group structure within fewer PMax campaigns rather than many separate PMax campaigns. Use 4-6 PMax campaigns with 5-15 asset groups each, segmented by sub-category. Add Standard Shopping for hero SKUs (top 100-300), AI Max for non-brand search reach, multi-layer DG.

11. Is it ever right to run 10+ PMax campaigns?
Almost never. Beyond 5-6 PMax campaigns you fragment conversion data, dilute algorithm performance, and create management overhead that doesn’t produce proportional returns. Use asset group structure inside PMax campaigns for granularity instead.

LLM shopping optimization

12. Does my Google Shopping feed actually power ChatGPT and Perplexity?
Yes. Researchers found base64-encoded Google Shopping API parameters in ChatGPT’s source code; an estimated 83% of ChatGPT’s product recommendations pull from Google Shopping data. Perplexity Shop, Gemini AI Mode, and Google AI Overviews shopping all draw from the same or similar feed pipelines.

13. What feed fields matter most for AI shopping engines?
Beyond the basics, the AI-critical fields are: product_review_count, average_rating, return_policy, structured shipping data, popularity_score, video_link, model_3d_link, highlight (5-10 attribute callouts), product_detail (structured specs), identifier_exists. Most brands have 30-50% of these populated; getting to 90%+ is the visibility unlock.

14. Do I need a separate feed for ChatGPT Shopping?
Optional but recommended. ChatGPT accepts its own feed format (JSONL, CSV, TSV, or Parquet, gzip-compressed) and refreshes every 15 minutes — much faster than the Google Shopping pipeline. Brands that submit directly to ChatGPT get 12-24 hours faster recommendation visibility.

15. How does Perplexity Shop differ from ChatGPT Shopping?
Perplexity has its own merchant API and direct merchant relationships in addition to pulling from Google Shopping. The optimization overlaps significantly with ChatGPT optimization, but Perplexity Shop has additional emphasis on editorial/review citation patterns and tends to favor products with strong off-site review presence.

16. Will Google’s AI Mode replace Shopping ads?
Not entirely. AI Mode adds a new shopping surface (carousel within AI-generated answers); Shopping ads continue to run on the classic search results page and Shopping tab. The feed and bidding signals overlap, so optimizing for one largely improves the other.

17. What’s the ChatGPT Shopping feed refresh schedule?
Every 15 minutes. This is the fastest of any AI shopping engine. Submitting directly gets you the fastest possible recommendation update for inventory changes, price drops, and new product launches.

18. How do I check if my products are appearing in ChatGPT Shopping?
Manually run 5-10 product-discovery queries in ChatGPT (e.g., “best [your category] under $X,” “top-rated [product type] for [use case]”) and check whether your products appear in the carousel. Track weekly. There’s no official analytics dashboard yet; manual sampling is the current best practice.

Performance Max specifics

19. Should I use search themes in Performance Max?
Yes, with 3-5 themes per asset group. Search themes give Google directional guidance about which keyword categories you want to target. Treat them as guidance (Google may serve outside them), not certainty. Refresh quarterly based on search term report patterns.

20. How do brand exclusions in PMax actually work?
You designate specific brand keywords (your own brand, competitor brands, or generic terms you want excluded) and PMax stops serving on those queries. Critical for accounts with meaningful brand search volume — without brand exclusions, PMax cannibalizes Brand Search at higher CPC.

21. Should I use Customer Match audiences in PMax?
Yes. Upload your customer email list as a Customer Match audience and add it as an audience signal on your PMax campaigns. Signals don’t restrict targeting but accelerate algorithm learning. Customer Match is one of the highest-quality signals available.

22. What conversion value rules should I set up?
At minimum: assign higher conversion values to new customers vs returning customers, and adjust by geographic region if your unit economics differ materially. Advanced setups also vary by device, time of day, or campaign source.

AI Max for Search

23. Should I enable AI Max on all my Search campaigns?
Yes for non-brand Search; selectively for Brand Search. On non-brand Search, AI Max expands query reach and improves performance ~7% on average. On Brand Search, AI Max can over-expand into competitor queries — use only the Search Term Matching feature with tight brand controls.

24. How long does AI Max take to show meaningful performance change?
4-8 weeks. Roll out features one at a time (Search Term Matching first, then Asset Optimization, then Final URL Expansion), measure each over 2-4 weeks before adding the next.

25. What goes wrong with AI Max if I just turn everything on at once?
Three failure modes: query expansion into irrelevant terms (without brand exclusions and tight location controls), generated ad copy that misrepresents your products (without text guidelines), and Final URL Expansion sending traffic to wrong landing pages (without monitoring). Daily search term and URL inspection monitoring for the first 14 days catches all three.

Bidding & budget

26. What bidding strategy works best for PMax in 2026?
Maximize Conversion Value with Target ROAS for established accounts with stable conversion data. Maximize Conversion Value (no target) for new accounts during the learning period. Target CPA for conversion-count-driven accounts where revenue per conversion is consistent. Manual bidding is rarely the right answer for PMax in 2026.

27. How should I split budget across the four campaign types?
Rough starting allocation for a $10K/month account: 60% PMax, 15% Brand Search, 15% AI Max non-brand Search, 10% Demand Gen. Adjust quarterly based on actual ROAS by channel. Larger accounts ($50K+/month) typically run 50% PMax / 15% Brand / 20% AI Max / 15% DG.

28. Should I cap my PMax budget or let it scale freely?
Cap during learning periods (first 30 days of any new PMax campaign) and during seasonal/promotional spikes where you don’t want runaway spend. Otherwise let Smart Bidding scale freely up to your daily budget. Smart Bidding is generally better at pacing than manual caps.

Common mistakes & strategy

29. What’s the most common mistake operators make with Google Ads structure in 2026?
Over-engineering campaign architecture (12+ campaigns) while under-investing in feed quality. The accounts that win run 5-10 campaigns total with feeds at 90%+ field completion. Feed quality is the highest-leverage activity in modern Google Ads.

30. How should I think about Google Ads vs LLM shopping in 2026?
They’re not competitors — they’re the same pipeline with different surfaces. Your GMC feed powers Google Shopping ads, Google AI Mode, ChatGPT Shopping, Perplexity Shop, and Gemini AI Mode simultaneously. Optimize the feed, run the four campaign types appropriate to your catalog size, and you’ll capture visibility on all of them at once.

Who wrote this — and how iComchain helps Google Ads operators

This guide was written by Ankit Agarwal, founder of iComchain LLC — a Google Ads and Merchant Center specialist agency. We work with e-commerce accounts on campaign structure, feed optimization, and AI shopping visibility across the four campaign types covered here. I run all four types on the iComchain account; the structural decisions in this guide are the ones I’ve made on real accounts at real spend.

Message us on WhatsApp at +1 323 647 2657 or email hello@icomchain.com with your monthly Google Ads spend, catalog SKU count, and biggest three current frustrations. The first 30-minute call is free.

Related reading on iComchain

Sources & further reading

© 2026 iComchain LLC. This article is educational and is not legal or financial advice. Google Ads features, policies, and best practices change frequently — verify any specific recommendation against Google’s official documentation before implementing. Performance results vary by account, category, and execution; numbers and percentages cited reflect typical patterns but cannot be guaranteed for any specific account.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top