Skip to content
Available skills

check-product-segments skill

The read-only skill an agent uses to explain your Product Segmentation segments, what each holds, the budget on Drainers and why a product changed segment.

check-product-segments explains your six product segments with real figures: what each holds, how much revenue and budget it carries, and why a product moved. Product Metrics machine learning segments every product on profitability, stock, product returns, potential and competitor prices, and writes the segment to a custom label for Shopping and Performance Max. The skill only reads.

Use it for

Questions about Stars, Question Marks, Cash Cows, Dogs, Drainers and Ghosts.

Works through

The MCP tools summarize_groups, get_highlights and get_segment_history, or the CLI.

You get

Revenue, ROAS and products per segment, and options to increase or lower priority.

The skill

Full file: on GitHub or at /.well-known/agent-skills/check-product-segments/SKILL.md.

pnpm dlx skills add product-metrics/skills \
  --skill check-product-segments

When an agent uses it

  • “How much of my budget goes to Drainers?”
  • “Which of my products are Stars right now?”
  • “Why did this jacket drop out of Stars last week?”

Example conversation

You:    How are my segments doing?
Agent:  summarize_groups (by: "segment")
        Stars: 84 products, revenue €61,200, ROAS 9.8
        Question Marks: 132 products, revenue €22,400, ROAS 6.1
        Drainers: 210 products, revenue €1,900, ROAS 0.9
        get_highlights
        Drainers take €2,100 of the budget. Stars and Question Marks
        earn far more on it.

        Options: lower priority on Drainers, increase priority on
        Question Marks with room to grow. You decide.

Example data.

Commands and tools it uses

StepCLI / MCP
Totals per segmentMCP summarize_groups (by: "segment"), get_highlights
Products in a segmentMCP list_products_to_investigate (segment)
Why a product movedMCP search_products, get_segment_history, get_product_context
What changed this weekMCP get_changes (group_by: "segment"), get_group_timeseries
Settings and segment labelproductmetrics settings get (settings.get), productmetrics feed-source-settings get (feed_source_settings.get)
Segment counts without MCPproductmetrics settings preview with the saved settings (settings.preview)

A segment explains past results; it does not predict future profit. Ad clicks are one signal among many in the model, so the skill explains segments on profitability and potential.

Where next