> ## Documentation Index
> Fetch the complete documentation index at: https://996f455-productmetrics.manicule.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.
> For every page in one file, fetch https://996f455-productmetrics.manicule.dev/docs/llms-full.txt (large).
> To search and read these docs as tools, connect the MCP server at https://996f455-productmetrics.manicule.dev/docs/api/mcp.
> Every agent surface: https://996f455-productmetrics.manicule.dev/docs/.well-known/llms-manifest.json

# analyse-google-ads skill (/docs/agent-skills/analyse-google-ads)

<!-- agent-signals: reading_time_min: 2 · est_tokens: 828 · updated: 2026-11-16 -->
Related: [Product Metrics agent skills](/docs/agent-skills/overview.md), [Install Product Metrics skills](/docs/agent-skills/install.md), [set-up-tracking skill](/docs/agent-skills/set-up-tracking.md), [connect-google-ads skill](/docs/agent-skills/connect-google-ads.md), [check-tracking skill](/docs/agent-skills/check-tracking.md), [bid-on-profit skill](/docs/agent-skills/bid-on-profit.md)

> The read-only skill an agent uses to answer questions about Google Ads performance, profit and wasted spend with real figures.

**`analyse-google-ads` answers questions about your Google Ads account with real figures from Product Metrics, then explains what drove them.** It picks the right tool for the question, reads results the way Product Metrics means them (ROAS, POAS, conversion delay, segments), and never changes Google Ads or your store.

## When an agent uses it

* "How are my Google Ads doing this month?"
* "Why did ROAS drop last week?"
* "Which products lose money after ads?"
* "Where is spend wasted?" "Who changed the account?"

## Example conversation

```text
You:    Why did revenue from Google Ads drop since Monday?
Agent:  list_properties → Example Shop (EUR)
        explain_google_ads_change (metric: "conversion_value", change_date: Monday)

        Conversion value fell from €8,420 to €5,960 (last 7 days vs the 7 before).
        Most of it is PMax Shoes (−€1,900): its budget was lowered on Monday
        (change history: budget €300 → €180 per day). The store's own revenue fell
        only 4%, so this is less traffic from that campaign, not a tracking problem.
        Option: increase priority on PMax Shoes again;
        its POAS is 1.4, above break-even.
```

## Commands and tools it uses

**MCP tools (preferred):** `list_properties`, `get_google_ads_report`, `explain_google_ads_change`, `get_google_ads_changes`, `get_poas_report`, `get_store_overview`, `compare_product_segments`, `get_product_targeting`, `get_product_context`, `get_inventory_state`, `run_google_ads_query`. See [MCP tools](/docs/mcp/tools).

It uses `propose_change` only when you ask it to file a proposal.

**CLI, without MCP:** the CLI has no Google Ads report command, so the agent answers from store and spend data and says when a question needs MCP.

| Question                   | CLI                                                                               |
| -------------------------- | --------------------------------------------------------------------------------- |
| Stores and accounts        | `productmetrics properties list`                                                  |
| Revenue, margin, ad spend  | `productmetrics store get-health-slice`, `productmetrics store get-health-series` |
| Campaigns below break-even | `productmetrics insights get --spot campaign_break_even`                          |
| Since the plan was applied | `productmetrics google-ads get-plan-results`                                      |
| Spend per day              | `productmetrics usage get-ad-spend`                                               |
| Products running out       | `productmetrics inventory get-state`                                              |
| Anything else              | `productmetrics data query` (slow, rate-limited)                                  |

## The skill

Full file: [`/.well-known/agent-skills/analyse-google-ads/SKILL.md`](https://www.productmetrics.io/docs/.well-known/agent-skills/analyse-google-ads/SKILL.md).

```bash
npx skills add https://www.productmetrics.io/docs \
  --skill analyse-google-ads
```

## Where next

* [Connect the MCP server](/docs/mcp/connect)
* [MCP tools](/docs/mcp/tools)
* [bid-on-profit skill](/docs/agent-skills/bid-on-profit)
