LeadJourney

View saved version →

---
name: campaign-link-setup
description: Build correctly tagged UTM links and optional short links in LeadJourney for a new campaign, so every click is attributed to the right channel and campaign.
---

Use this skill when the user wants a tracking link for a campaign, asks to "create a UTM", "make a short link", or is about to launch an ad, email, or social post and wants it tracked.

## Inputs to establish

Ask for anything missing before building. Required:

- **Destination URL** — the page the click should land on.
- **Channel / source** — where the link will be posted (LinkedIn, Google Ads, newsletter, etc.).
- **Campaign name** — a short slug the user will recognize in reports.

Optional: medium (cpc, email, social, organic), content (creative variant), term (keyword).

## Steps

1. Call `get_current_workspace` and switch if the user names another brand.
2. Call `list_utm_channels` to see the source/medium vocabulary the workspace already uses. Reuse an existing pairing when one matches the requested channel; consistent naming is what keeps reports clean. Do not introduce a new source spelling ("linkedin" vs "LinkedIn") when one already exists.
3. Call `build_utm_link` with the destination, source, medium, campaign, and any optional fields.
4. If the user wants a short link or the destination is long, call `list_tracking_domains` and then `create_short_link` on one of the workspace's own tracking domains. If no tracking domain exists, tell the user and offer `create_tracking_domain` rather than creating one unasked.
5. If the user is launching several variants (A/B creatives, multiple placements), build one link per variant with a distinct `content` value.

## Output

- The full UTM link, and the short link if created, each on its own line in a code block.
- A one-line breakdown of the parameters used so the user can verify.
- One reminder: which campaign name to look for in `get_campaign_report` once traffic arrives.

## Do not

- Do not create tracking domains, custom traffic channels, or tags without the user explicitly asking.
- Do not change the destination URL's path or query string beyond appending UTM parameters.
- Do not shorten links on domains the workspace does not own.

Explicit user instructions override these defaults.

View saved version →

---
name: channel-attribution-report
description: Report which traffic channels, campaigns, or ad creatives are producing leads and revenue in the user's LeadJourney workspace for a date range, and compare them.
---

Use this skill when the user asks where their leads come from, which channels or campaigns perform best, how marketing performance changed over a period, or wants a comparison between channels, campaigns, or creatives.

## Inputs to establish

1. **Workspace.** Call `get_current_workspace`. If the user names a different client or brand, call `list_workspaces` and `switch_workspace` before anything else.
2. **Date range.** If the user does not give one, default to the last 30 days and say so. If the user says "this month", use the current calendar month to date and state the exact dates used.
3. **Level of detail.** Channel → `get_traffic_channel_report`. Campaign → `get_campaign_report`. Creative → `get_ad_creative_report`. If the question spans two levels ("which Google Ads campaign brought the most leads"), start at the channel report and drill into the campaign report for that channel.

## Steps

1. Run the fixed report that matches the level of detail for the date range.
2. If the fixed reports do not contain the metric or dimension the user asked for (for example revenue per lead by landing page), call `describe_report_builder` first, then `run_report` with ids taken only from that response. Never invent metric or dimension ids; an invented id returns empty rows instead of an error.
3. Rank results by the metric the user cares about. Default to lead count; use revenue when the user mentions ROI, revenue, or value.
4. For comparisons, run the same report for both periods or both channels and present them side by side.

## Output

- A short table with the name (channel, campaign, or creative) and the metrics the report actually returns, such as leads and revenue. Only show columns that have data in the response; do not add empty columns.
- One to three sentences of interpretation: the leader, the biggest mover, and anything that looks off (a channel with traffic but zero leads, for example).
- State the date range and workspace name once at the top.

## Do not

- Do not invent numbers, fill gaps, or extrapolate trends beyond what the report returns.
- Do not present zero rows as "no leads" without first checking that the date range and workspace are the ones the user meant.
- Do not create dashboards, tags, or links from this skill; if the user wants the report saved, hand off to the dashboard-builder skill.

Explicit user instructions override these defaults.

View saved version →

---
name: dashboard-builder
description: Create or update a LeadJourney dashboard with report cards (channel, campaign, creative, or custom reports) so the user has a recurring view of marketing performance.
---

Use this skill when the user wants a dashboard, wants a report "saved" or "pinned", asks for a weekly/monthly overview they can return to, or wants to add a chart to an existing dashboard.

## Inputs to establish

1. **Workspace.** `get_current_workspace`; switch if needed.
2. **New or existing dashboard.** Call `list_dashboards`. If a dashboard with a matching name exists, add to it rather than creating a duplicate. Confirm with the user when the match is only approximate.
3. **Which reports.** Each card is a report. Clarify the date range behaviour (rolling last 30 days vs fixed period) and the metric the user cares about.

## Steps

1. Call `describe_report_builder` once to get the valid metric and dimension ids. A card takes the same arguments as `run_report`, so every id must come from this response.
2. For each requested card, call `run_report` first and check that it returns rows. A card built from a report that returns nothing renders as an empty tile; fix the arguments before adding it.
3. If creating: call `create_dashboard` with a clear name. If adding: use the existing dashboard id.
4. Call `add_dashboard_report` once per card, with the verified `run_report` arguments and a card title the user will understand.
5. Only call `remove_dashboard_report` or `delete_dashboard` when the user explicitly asks to remove or delete, and confirm the target by name before doing so. Deletion cannot be undone.

## Output

- Dashboard name and a list of the cards added, each as `title — metric by dimension, date range`.
- If any requested card could not be built (metric not available, empty results), say which one and why, and suggest the closest alternative from `describe_report_builder`.

## Do not

- Do not guess metric or dimension ids.
- Do not delete or remove anything without an explicit instruction naming the dashboard or card.
- Do not add more than the user asked for; suggest extra cards in one line instead of creating them.

Explicit user instructions override these defaults.

View saved version →

---
name: lead-journey-summary
description: Find a specific lead in LeadJourney and summarize their full journey from first touch to conversion, including which channel and campaign brought them in.
---

Use this skill when the user asks about a single lead: "where did this lead come from", "show me the journey of", "summarize lead X", or pastes a LeadJourney lead URL such as `?slug=lead-overview&lead_id=<uuid>`.

## Inputs to establish

1. **Workspace.** Call `get_current_workspace`; switch with `switch_workspace` if the user names another brand.
2. **Which lead.** Accept an email, phone number, name, or lead id.
   - If given an id or URL containing `lead_id`, go straight to `get_lead_journey`.
   - Otherwise call `get_leads` with the search term. If more than one lead matches, list them (name, email, created date) and ask the user to pick before continuing. Do not guess.

## Steps

1. Call `get_lead_journey` for the chosen lead.
2. Read the timeline in order. Identify:
   - First touch: channel, source/medium, campaign, landing page, timestamp.
   - Any later touches and what changed (returning via a different channel, a specific ad creative, a retargeting click).
   - Conversion event: form, call, purchase, or CRM stage, and when.
   - Revenue attributed, if present.

## Output

Present in this order:

1. One-line identity: name, email, company if available, date created.
2. **Journey**: a numbered list, one line per touchpoint, oldest first — `date · channel · campaign/creative · page`.
3. **Attribution**: which channel gets first-touch credit and which gets last-touch credit, in one sentence each.
4. **Status**: converted or not, revenue if any.

Keep the whole summary under 200 words unless the user asks for full detail.

## Do not

- Do not include raw internal ids, tokens, or debug fields from the tool response.
- Do not infer intent, sentiment, or buying stage that the data does not show.
- Do not modify the lead, add tags, or trigger agents unless the user explicitly asks.

Explicit user instructions override these defaults.