Performance Max is opaque by design. Google controls placement, format, and audience expansion. Most AI tools can read surface-level PMax data. PaidSync goes further: 18 tools that cover asset group inspection, asset-level performance scores, audience signal management, search themes, and store-visit conversion data, with full write access to push changes back. The gap between "AI that reads your PMax" and "AI that fixes your PMax" is what this piece covers.
Performance Max replaced Smart Shopping and Local campaigns in 2022. It now accounts for a large share of Google Ads spend at most e-commerce and local advertisers. The core problem is visibility. PMax consolidates placements across Search, Shopping, YouTube, Display, Discover, and Maps, and Google's reporting rolls most of that up. Without API-level access, you are working with campaign totals and almost no asset-level signal.
The second problem is write access. Most AI-adjacent tools for PMax are read-only: they pull performance data and surface insights. They cannot touch the asset groups, cannot swap a low-scoring headline, cannot add a search theme, cannot modify the audience signals attached to a specific asset group. PaidSync has write access to all of it.
The third problem is signal quality. PMax bidding relies on your conversion data. If the conversion goal is misconfigured, if offline visit data is missing, or if the wrong conversion actions are weighted, the machine-learning model is optimizing toward the wrong outcome. PaidSync's store-visit and conversion diagnostic tools surface these mismatches before they compound.
These are the PMax-specific tools available inside Claude or ChatGPT through the PaidSync Google Ads connection. They run against the Google Ads API v17+ and return structured data for the AI to reason over.
Connect Google Ads to Claude or ChatGPT and run these tools today.
Try PaidSync free See Google Ads toolsThis is the workflow for a full PMax audit and optimization pass. Each step maps to a specific task. Run it at the start of a new month or after a major budget change.
Start with the campaign list so you know what you are working with.
"List all Performance Max campaigns in account [ACCOUNT_ID] with their budgets, target ROAS, and conversion goals. Flag any with ROAS below 3 or cost above $1,000 last 30 days with no conversions."
Runs: list_performance_max_campaigns + get_pmax_performance_reportOnce you have the campaign list, go one level deeper into the asset groups that matter.
"For campaign [CAMPAIGN_ID], list all asset groups with their current status. Then pull the full asset details for the top-spending asset group."
Runs: list_asset_groups + get_asset_group_detailsThis is where PMax visibility usually breaks down. The performance breakdown shows exactly which creative elements Google has flagged.
"Run the asset performance breakdown for asset group [ASSET_GROUP_ID]. List every asset rated Low, show me the exact text or image URL, and suggest a replacement for each based on the top-rated assets in the same group."
Runs: get_asset_performance_breakdownAudience signals guide Google's initial targeting before the model expands. A missing customer list or stale remarketing segment here means the model starts cold.
"Check the audience signals on asset group [ASSET_GROUP_ID]. List what's attached. Then pull available customer match lists and recommend which ones to add as signals based on size and recency."
Runs: get_audience_signals + list_customer_match_lists + update_audience_signalsSearch themes are the strongest signal in PMax for advertisers who also run Search campaigns. They tell Google which intent areas to prioritize before it auto-expands.
"Show me the current search themes on asset group [ASSET_GROUP_ID]. Flag any that are disapproved. Based on the campaign's top converting search terms from the last 90 days, suggest 10 new search themes to add."
Runs: get_search_themes + get_pmax_performance_report + update_search_themesFor local and retail accounts, store-visit conversions often represent the majority of actual value. Missing or misconfigured visit data leads the bidding model toward wrong signals.
"Pull store-visit conversion data for account [ACCOUNT_ID] over the last 60 days. Show totals by campaign, and flag any PMax campaign where visits are zero but physical locations exist in the account."
Runs: get_store_visit_conversions + list_performance_max_campaignsAfter reviewing, push the approved changes and re-pull the asset scores to confirm the update took.
"Update asset group [ASSET_GROUP_ID]: replace the three Low-rated headlines with the ones we discussed. Add the two new search themes. Then re-pull the asset group details to confirm all changes are live."
Runs: update_asset_group + update_search_themes + get_asset_group_detailsThe most common question agencies bring to this topic. The answer is usually not either/or, it is both running in parallel with clear scope boundaries.
| Signal | Use PMax | Use Search |
|---|---|---|
| Conversion volume | 50+ conversions/month per campaign | Under 30/month |
| Asset availability | Rich creative (video, multiple images) | Text-only campaigns |
| Keyword control need | Low, willing to trust signals | High, need exact match control |
| Negative keywords | Limited (account-level only) | Full campaign-level negatives |
| Cross-channel inventory | Shopping + YouTube + Display + Search | Search only |
| Brand protection | Use alongside, not instead of Search | Brand campaigns in Search |
| Reporting transparency | Limited placement data | Keyword + placement + device breakdown |
The agency-standard setup: PMax handles non-brand, top-of-funnel inventory across channels. Exact-match and phrase-match Search campaigns cover brand and high-intent terms. The PMax campaign's final URL expansion is set to send only to specific URLs to prevent overlap with dedicated landing page campaigns.
The single signal to watch: If Search impression share for your top keywords starts dropping after PMax launches, PMax is cannibalizing traffic. Use the Google Ads MCP tools to pull impression share by campaign type and compare before/after periods.
Several AI tools and MCP servers can query PMax data. The gap is write access. Reading a low-scoring asset is the easy part. Replacing it, adding a search theme, swapping an audience signal, and confirming the change in the same conversation is what PaidSync's full-stack PMax tooling enables.
The specific capability that no read-only tool replicates is the audit-then-fix loop. PaidSync lets Claude identify the three Low-rated headlines, propose replacements based on the Best-rated patterns in the same group, push the update, and re-read the asset group to confirm the change landed, all without leaving the conversation.
Yes. PaidSync exposes 18+ PMax-specific tools that let Claude or ChatGPT read asset group performance, audit audience signals, manage search themes, and push changes back to Google Ads. The AI reads low-scoring assets and suggests replacements based on performance data.
Search themes are keyword-intent signals you add to a PMax asset group to guide Google's targeting. They are not match-type keywords; they tell Google which search intents are relevant. PaidSync's get_search_themes and update_search_themes tools let you read and edit them inside Claude or ChatGPT.
Ask Claude to run get_asset_performance_breakdown for your campaign ID. PaidSync returns each asset with Google's Low, Good, or Best rating, so you can identify exactly which headlines, descriptions, images, or videos are dragging performance.
Yes. Unlike read-only AI tools, PaidSync has full write access. The update_asset_group tool lets Claude push new headlines, descriptions, or swap images and videos. Every change is a visible task you confirm before it goes live.
PMax wins when you have strong conversion volume (50+ per month per campaign), diverse assets, and cross-channel inventory to fill (Shopping, YouTube, Display, Search). Standard Search wins when you need tight keyword control, clear negatives, and transparent match-type reporting. Many accounts run both in parallel with PMax covering top-of-funnel and Search owning brand and high-intent terms.
Audience signals are suggestions, not hard targets. You attach a customer list or custom segment as a signal, and Google uses it as a starting point but expands beyond it to find more conversions. Standard audience targeting in other campaign types restricts who sees the ad. PaidSync's get_audience_signals tool shows what signals are attached and how they are performing.
Yes. PaidSync's get_store_visit_conversions tool pulls offline store-visit data from the Google Ads API, lets you see which campaigns are driving physical visits, and confirms whether the conversion goal is correctly weighted in the campaign's bidding setup.
Connect Google Ads to Claude or ChatGPT. 440+ tools across 13 platforms. Free to start.