| name | cx-intelligence-digest |
| description | Turn the last 7 days of Gorgias tickets into an executive-ready weekly CX digest — analyzed across five business functions (Product, Marketing, Ecommerce/Revenue, Product Marketing, Leadership) with a TLDR, signal-status tagging, a PDP-gap table, and owner-tagged actions. Read-only; optionally publishes to Notion. |
CX Intelligence Digest Generator
Your support queue is the highest-signal feedback loop in the business — but the insight is scattered across hundreds of tickets and never reaches the teams who could act on it. This recipe compresses a week of tickets into one structured, executive-ready digest, read through five functional lenses, with owner-tagged actions so it doesn't die in someone's inbox.
It's the heavyweight sibling of weekly-voc-digest: that one is a quick VoC pulse; this is the full cross-functional readout you'd send to leadership. Read-only on Gorgias — it can optionally publish straight to Notion.
Heads up: the volume, automation, and resolution figures rely on Gorgias MCP analytics, which are in beta and directional. The prompt's built-in rules (no invented data, (heuristic) labels, small-sample caveats) keep it honest — cross-check exact numbers in your Statistics dashboard before reporting them up.
When to use it
- A weekly leadership/cross-functional CX readout
- Feeding Product, Marketing, and Ecommerce real customer evidence on a cadence
- Replacing the manual "what did support hear this week" doc someone writes by hand
Customize before you run
| Variable | Example |
|---|
{{BRAND_NAME}} | "Acme" |
{{START_DATE}} | "2026-05-22" |
{{END_DATE}} | "2026-05-28" |
{{PARENT_PAGE}} | a Notion page/database to publish under (optional — omit to get markdown back) |
The five lenses (Product / Marketing / Ecommerce / Product Marketing / Leadership) are tuned for a DTC e-commerce brand — rename or swap them to match your org.
The workflow
Fill in the placeholders and paste this into your connected AI client:
You are a CX Intelligence Analyst. Generate a weekly CX Intelligence Digest for
{{BRAND_NAME}} using support data from our connected Gorgias helpdesk. [If a Notion
connector is available: publish the result as a new Notion page under {{PARENT_PAGE}};
otherwise return it as formatted markdown.]
Data scope & rules
- Window: the last 7 days ({{START_DATE}} → {{END_DATE}}). State the window explicitly.
- Source of truth: Gorgias only. Pull non-spam tickets, intents/tags, channel, AI Agent
handover vs. resolution status, sentiment tags, and CSAT responses.
- No invented data. Every number must trace to Gorgias. Where a figure comes from keyword
or tag matching rather than a structured field, label it (heuristic).
- Redact PII in all customer quotes (names, emails, order numbers, addresses).
- Note any metric with too small a sample to be reliable (e.g. CSAT under ~30 responses)
instead of reporting it as fact.
- Classify every signal with a status emoji so trends are visible at a glance:
🟢 emerging (new or accelerating this week), 🔵 existing (recurring / tracked),
🟡 fading (declining vs. recent weeks). If only a single week of data is available,
infer status directionally from this week's volume and language, and say so.
Structure to produce (keep this exact skeleton)
1. Top callout — "TLDR: top 5 learnings this week (all functions)": a highlighted callout
with exactly 5 bullets synthesizing the single most important learning across every
function. Lead each bullet with a bolded takeaway, then the supporting number.
Prioritize risk, revenue, and the AI automation gap.
2. "Week at a glance" — short bulleted overview: Volume (total non-spam + top 5 intents
with counts); Channels (volume per channel + % AI-handled); AI Agent (tickets touched,
# and % handed to human, # and % fully resolved); Sentiment flags (negative/urgent tag
counts; CSAT note); #1 risk signal of the week.
3. Five department sections, each a collapsible toggle:
🛠️ 1. Product — Recurring customer pain report
📣 2. Marketing — Voice of customer insights
🛒 3. Ecommerce / Revenue — Shopper friction report
📚 4. Product Marketing — Feature confusion & enablement gaps
🧭 5. Leadership — Weekly CX intelligence summary
Inside each section, use these blocks in order:
- What we learned — 4–6 tight bullets with key signals and figures (labeled (heuristic)
where relevant). Prefix each bullet with its status emoji (🟢 / 🔵 / 🟡) and add a
one-line status key under the heading. End with an italic one-line volume/impact summary.
- What we should do — action items as to-do checkboxes (- [ ]), each a bolded action +
one line of rationale.
- What customers said & source tickets (redacted) — 3–5 representative verbatim quotes,
each under ~25 words, with the source Gorgias ticket linked inline at the end of each
quote. If two quotes share a ticket, link both; if a relevant ticket has no quote, list
it as an italic "Also relevant:" line.
4. Two section-specific additions:
- In Ecommerce / Revenue, after "What we should do," add a table titled "Products needing
better PDP descriptions" with columns: Product | Gap in current PDP | What to add |
Priority. Populate only from products actually referenced in this week's conversations.
- In Leadership, lead with a signal status legend (🟢 emerging · 🔵 existing · 🟡 fading),
then split signals into three labeled lists — "Emerging signals" 🟢, "Existing signals" 🔵,
and "Fading signals" 🟡 — each bullet prefixed with its emoji.
5. Footer — "✅ Top cross-functional actions for this week": a collapsible toggle with the 5
highest-priority actions across all functions, each with a named Owner (e.g. CX Leadership,
Product/Engineering, Ecommerce, Product Marketing, AI Agent Ops). Close with an italic
methodology note explaining (heuristic) figures, the CSAT caveat, spam exclusion, and the
date window.
Tone & formatting
- Crisp and executive. Bullets over paragraphs. No filler.
- Bold the takeaway in each bullet; keep supporting detail to one clause.
- Spell out terms in full (e.g. "success rate," not abbreviations).
- Make department sections collapsible toggles so the page scans top-to-bottom from the callout.
What it writes
- Gorgias: nothing — read-only.
- Notion (optional): if you provide
{{PARENT_PAGE}} and have the Notion connector enabled, it publishes the digest as a new page. Otherwise it returns formatted markdown you can paste anywhere.
Tips
- Run it weekly. Save it as a scheduled task and only update the date window each run.
- The trust rules are load-bearing. "No invented data" and the
(heuristic) labels are what make leadership believe the numbers — keep them in any variant.
- Single-week caveat: the 🟢/🟡 status calls are directional until you feed it multi-week data. For true trends, run it against several weeks and ask it to compare.
- Pair with
intent-gap-guidance: the recurring pains this surfaces are exactly what your AI Agent should learn to handle next.