---
name: curcuma-competitive
route: "Competitive or category research on how brands email a topic"
router-terms: competitive, competitor, competitors, category research, vertical, how do brands, what do companies, benchmark, industry examples, brand dossier, compare brands, SaaS emails, DTC emails, BFCM research, black friday research, seasonal research
description: |
  Competitive and category research over the production email corpus. Finds
  how brands in a vertical or moment email, clusters structure and messaging,
  cites slugs and optional pattern prevalence. Use for intelligence and
  benchmarking — not finished HTML and not conversion claims.
  Trigger on: competitive analysis, how brands do X, category research,
  benchmark, brand comparison, vertical examples.
---

# Curcuma Competitive / Category Research

Evidence-backed research on **how real brands email a topic or vertical**.

## When to Use

- “How do SaaS / DTC / beauty brands do onboarding / launch / cart?”
- Competitive or category benchmarking before a campaign
- Seasonal research (Black Friday, Cyber Monday, holiday)
- “Show examples of X and summarize patterns”

## When NOT to Use

- User wants finished HTML → `curcuma-email-design`
- Lifecycle sequence design primary → `curcuma-lifecycle` (may follow this skill)
- Subject-only copy list → `curcuma-subject-cta`
- “Does my HTML work in Outlook?” → `curcuma-production`

## Hard limits

- Brand coverage is uneven; if a named brand has few hits, say so.
- Do not invent market share or “most brands do X” unless backed by
  `get_pattern` prevalence or an explicit in-sample count you computed.
- No ordered multi-email brand journeys unless each step is retrieved separately.

## Workflow

### Step 1: Frame the research question

Write one sentence: vertical/moment + question type
(examples / structure / messaging / both).

### Step 2: Category awareness

```
list_categories()
```

Pick 1–3 relevant category names to guide search.

### Step 3: Search (2+ query variants)

```
search_emails(query="<primary>", limit=10, max_per_company=1)
search_emails(query="<variant or vertical + moment>", limit=10, max_per_company=1)
```

Examples:
- `onboarding` then `welcome saas`
- `black friday` then `cyber monday`
- `product launch` then `announcement`

Page with `offset` if needed. Record coverage.

### Step 4: Study 3–5 diverse slugs

```
get_email_summary(slug="...")
```

Note layout, CTAs, subject patterns, section structure, offer framing.

**Hard fail on citations:** only cite slugs whose `get_email_summary` returned
real summary fields (title/structure/CTAs). If the tool returns `error` or
empty enrichment (no HTML / no summary), **drop that slug** — search hits are
not enough to claim you studied it.

### Step 5: Optional industry base rates

When the question is structural (“do they use dark mode / MSO / preheaders?”):

```
get_pattern(name="preheader")
get_pattern(name="dark mode")
get_pattern(name="responsive")
```

Only quote prevalence strings returned by the tool.

### Step 6: Deliver research memo

1. **Question restated**
2. **Corpus coverage** — hit quality; weak areas
3. **Patterns** — structure, messaging, CTAs (with slug citations)
4. **Example gallery** — 3–5 rows: company | slug | subject | CTA | one takeaway
5. **Implications** — what to copy structurally for the user’s context
6. **Honesty** — what the corpus cannot answer

## Grounding

See `skills/GROUNDING.md`. Competitive claims need **slugs + companies** from
search/summary. Quote real subjects/CTAs when illustrating a pattern.

## Anti-patterns

- DON'T write a generic marketing essay without tool calls
- DON'T invent prevalence percentages
- DON'T claim a brand’s full email calendar
- DON'T open with `get_email_html` for research (use summary first)
- DON'T invent competitor examples that are not in search results