Break down Google Ads cost per conversion, campaign by campaign

SKILL.md---
name: google-ads-cpa-by-campaign
description: "Each Tuesday: account CPA for last week with an 8-week trend, every campaign's cost per conversion against its target, and which ones moved. Use when the user asks for this, or when it runs on a schedule."
---

# Google Ads CPA by campaign

Each Tuesday: account CPA for last week with an 8-week trend, every campaign's cost per conversion against its target, and which ones moved.

Account CPA is an average that hides the story. A brand campaign at $12 and a competitor campaign at $123 can blend into a respectable $43, right up until the competitor campaign gets more budget. Bidding to a target CPA does not settle it either: targets drift, and a campaign stuck in learning can miss its target for weeks.

The agent breaks CPA out per campaign every Tuesday, sets each against its own target, and keeps the account figure as a trend line. Running on Tuesday gives the weekend's conversions an extra day to arrive.

## Two ways to get the data

Use the API or MCP route when you have an MCP server, a CLI or an API token for the app. Otherwise, if you can drive a browser the user is signed in to (Claude in Chrome, a Playwright or Chrome DevTools MCP server), use the browser route. Say which one you used.

## Route A: API or MCP server

### What you need

- Read access to the Google Ads account (customer ID) and, through a manager account, its ID as `login-customer-id`.
- Google Ads API credentials (a Google Cloud project with API access and OAuth; developer tokens were retired in September 2026), or the official Google Ads MCP server, which is read-only.
- Your account target CPA, for campaigns that bid without one.

If something is missing, say what and stop; do not invent numbers.

### Steps

1. **Pull cost and conversions per campaign** (Google Ads): Run GAQL on `campaign` with `campaign.name`, `campaign.bidding_strategy_type`, `metrics.cost_micros` and `metrics.conversions`, `WHERE segments.date DURING LAST_WEEK_MON_SUN AND campaign.status = ENABLED`.
2. **Read each campaign's target** (Google Ads): Select `campaign.target_cpa.target_cpa_micros` for Target CPA bidding and `campaign.maximize_conversions.target_cpa_micros` for Maximize conversions with a target. Campaigns without one are compared with the account target you give.
3. **Compare with the week before**: Run the same query for the Monday to Sunday before. Flag campaigns more than 20% over target, and mark a CPA that rose 25% or more on fewer than 10 conversions as noise, not a trend.
4. **Post the tile**: Account CPA is total cost over total conversions, never an average of campaign CPAs. It is the headline with 8 weeks of history, and the card opens to the campaign table.

## Route B: the browser

### What you need

- Signed in to Google Ads with at least Read-only access to the account.

If something is missing, say what and stop; do not invent numbers.

### Steps

Start at https://ads.google.com/aw/campaigns. The user is signed in already; never type passwords or codes, and ask if a sign-in page appears.

1. **Set last week with a comparison**: Open `https://ads.google.com/aw/campaigns`, pick the client account if asked, and in the date picker choose `Last week (Mon - Sun)` with `Compare` on and `Previous period` selected.
2. **Show the right columns**: Under Columns, Modify columns, add `Cost`, `Conversions`, `Cost / conv.`, `Bid strategy type` and `Target CPA`. Filter the table to Campaign status: Enabled.
3. **Read the account and each campaign**: Take account CPA from the Total row (cost over conversions, never an average of campaign CPAs). For each campaign note cost, conversions, Cost / conv., its Target CPA, and the change shown in the comparison row; campaigns without a target use your account target.
4. **Flag and sort the noise**: Flag campaigns more than 20% over their target. A CPA that rose 25% or more on fewer than 10 conversions is noise, not a trend.

Note: Conversions keep arriving for a few days after the click, so reading on Tuesday rather than Monday gives steadier numbers. The Cost / conv. column counts primary conversion actions only, the same as the API.

## The report

- **Account CPA** last week with an 8-week trend.
- **By campaign**: cost, conversions, CPA, target and the change on the week before.
- **Over target**: campaigns more than 20% above their target, with the spend at stake.
- **Noisy movers**: big CPA swings on too few conversions to act on yet.

## Post it to Briefkin

When `BRIEFKIN_TOKEN` is set, or `~/.config/briefkin/env` exists (the installer from https://briefkin.dev/docs writes it), also send the result to the user's Briefkin board as one card. Keep the `item_key` the same on every run so the card updates in place, use a new `event_id` per run, and put today's real numbers in place of the example ones. If the request fails, say so once and carry on.

```bash
[ -n "${BRIEFKIN_TOKEN:-}" ] || . ~/.config/briefkin/env
curl -sS -X POST "${BRIEFKIN_URL:-https://briefkin.dev}/v1/events" \
  -H "authorization: Bearer $BRIEFKIN_TOKEN" \
  -H "content-type: application/json" \
  -d @- <<'JSON'
{
  "v": 1,
  "event_id": "google-ads-cpa-by-campaign/<date of this run, e.g. 2026-10-16>",
  "item_key": "google-ads-cpa-by-campaign",
  "project": "Acme",
  "source": "Google Ads CPA by campaign skill",
  "kind": "metric",
  "title": "Account CPA down to $42.80",
  "metric": {
    "label": "Cost per conversion",
    "value": 42.8,
    "prefix": "$",
    "better": "lower",
    "period_start": "<start of the period, ISO 8601>",
    "period_end": "<end of the period, ISO 8601>"
  },
  "glance": {
    "type": "Grid",
    "props": {
      "columns": 3
    },
    "children": [
      {
        "type": "Stat",
        "props": {
          "label": "Cost",
          "value": 7704,
          "prefix": "$"
        }
      },
      {
        "type": "Stat",
        "props": {
          "label": "Conversions",
          "value": 180
        }
      },
      {
        "type": "Stat",
        "props": {
          "label": "Over target",
          "value": "2 campaigns",
          "tone": "warn"
        }
      }
    ]
  },
  "widget": [
    {
      "type": "Table",
      "props": {
        "columns": [
          "Campaign",
          "Cost",
          "Conv.",
          "CPA",
          "Target",
          "Change"
        ],
        "align": [
          "left",
          "right",
          "right",
          "right",
          "right",
          "right"
        ],
        "rows": [
          [
            "Brand Search",
            "$920",
            74,
            "$12.43",
            "$15",
            "-6%"
          ],
          [
            "Non-brand Search",
            "$3,980",
            62,
            "$64.19",
            "$50",
            "+12%"
          ],
          [
            "Performance Max",
            "$2,064",
            38,
            "$54.32",
            "$55",
            "-9%"
          ],
          [
            "Competitors",
            "$740",
            6,
            "$123.33",
            "$90",
            "+31%"
          ],
          [
            "Account",
            "$7,704",
            180,
            "$42.80",
            "$45",
            "-3%"
          ]
        ]
      }
    },
    {
      "type": "Callout",
      "props": {
        "text": "Non-brand Search is 28% over its $50 target on 62 conversions, a real trend. Competitors is further over, but on 6 conversions it is still noise.",
        "tone": "warn"
      }
    }
  ],
  "size": "md"
}
JSON
```

Widgets come from the Briefkin catalog (https://briefkin.dev/v1/catalog). Without a token, just give the user the report.

Set `period_start` and `period_end` to the window the number covers: the tile names it from them ("Last 7 days", "Today", "This month"). For a count right now (open items, a balance), send `"period_label": ""` so no period shows; for another name, send it as `period_label` (at most 24 characters).

## Run it on a schedule

Suggested: Tuesdays at 8:30 (`30 8 * * 2`). From cron, with the agent signed in on that machine:

```bash
30 8 * * 2 claude -p "Run the google-ads-cpa-by-campaign skill"
# Codex: codex exec, Gemini CLI: gemini -p, Copilot CLI: copilot -p, OpenCode: opencode run
```

Or use the agent's own scheduler (Claude Code routines, Codex automations, Claude or ChatGPT scheduled tasks): https://briefkin.dev/skills/google-ads-cpa-by-campaign#schedule

Source: https://briefkin.dev/skills/google-ads-cpa-by-campaign