Skip to content

gooseworks-ai/goose-skills / skills/marketing/research-competitors

research-competitors

Verified

Competitor research: what every main rival is running, launching, charging and saying now, what works for them, and what that means for you. Use when the marketer asks for competitor research or a teardown, "what ads are my competitors running", "how do they price", "find gaps and angles", "what changed this week" (a weekly watch), or "X just launched / shut down / cut prices" (react now). Also builds a sales battlecard, checks a competitor page's tools/pixels, or studies who engages with their LinkedIn posts. Delivers a readable report saved to the brand, so the next run reports what changed. Never copies their words or disparages them. Not for what people say about you or your market — use research-market; not for who your buyers are — use research-audience.

references/report-format.md

View on GitHub
# Report format

Load at steps 5–6, with the `write-report` skill (`../../write-report/SKILL.md`) (read it fully — its shape, length budget (body ≤ 3,000 words) and §3 check govern). A busy founder reads the top section and knows what changed and what to do. Examples are for a fictional B2B scheduling tool — never copy them.

## The report

Brand copy `research/competitors/<YYYY-MM-DD>-report.md` (title `Competitor research — <Brand> — <YYYY-MM-DD>`, tags `research-competitors`, `competitor-report`, keep-history); local copy `gooseworks-reports/<brand-slug>/<YYYY-MM-DD>-competitor-research.md`. A repeat run opens "In short" with what changed since the last report.

```markdown
# <Brand> — competitor research, <D Month YYYY>

## In short
**What we looked at:** <the rivals, the sources (ads, prices, reviews, news), the date, and the main limit — e.g. "buyer quotes mostly from one review site", "ad libraries read from one country", "LinkedIn ads couldn't be opened">.
1. **<The biggest thing a rival did>** — <so what for us> (<confirmed / likely / first look>).
2. **<Where we're ahead or behind>** — <so what> (<how sure>).
3. <Optional: a timely opening or a threat, with its end date.>
**So, do this first (because of <1 and 2>):** <one move, who, by when — never something this report calls table stakes; never resting on an asset we haven't confirmed (named customers, budget, a live offer) — then the move is "confirm X">.
(≤ 150 words. A repeat run adds one line on what changed since the last report; a sibling report with a different first move or lead buyer gets one line, settled.)

## Decisions for you
| Decision | My recommendation | Matters by |
|---|---|---|

## The frame we compared in
One line: the category we compete in, the other frames we could claim, and why we picked this one (it decides which rivals matter).

## Timely openings (only when there are any)
### <What happened>, <date> — the window closes <date, and why that date>
What happened (linked), who it affects, the move, the claims that need your OK — and for a counter-position, why it's worth answering (we're the challenger, we do it better, we have dated proof, our answer reaches enough of their buyers).

## Who we compared, and why
The 3–6 rivals buyers compare us with now and how we know; the alternatives buyers use instead (doing it themselves, an agency, the platforms' own tools).

## The rivals, one by one
### <Rival>
<One sentence: what they are and what matters most about them now.>
- **What's new:** last 90 days (dated, linked).
- **What they're running:** ads per platform (or "couldn't open (reason)" / "none found, <date>" only where the library loaded); founder and team posts.
- **How they sell:** headline word for word, price and what you pay per (vendor page or third-hand), trial or demo.
- **What buyers say:** a praise and a complaint, quoted and linked.
- **Customers:** named ones (confirmed / likely), and logos that left their site this year — likely lost, a switch target.
- **Where we win, where we lose.**
**For you:** <only if it matters>

## How they compare
- **Pricing:** list price and unit; **what a buyer like ours actually pays a month** (one stated scenario, e.g. "a 10-person team making 40 ads a month"); the cheapest tier that has the features buyers care about; hidden costs; changed recently. Us in the same table.
- **Messaging:** headline word for word, who it's for, lead claims.
- **Channel scorecard, judged against each one's stage** (a seed-stage rival with no analyst coverage isn't weak): self-serve and free trial · search and content · reviews · founder voice · paid ads · events · partners and marketplaces · community · video — each "strong", "fine for their stage", "weak" or "couldn't check", with the evidence. The biggest gap nobody fills is a candidate move.
- **Topics:** a short table — topics each rival publishes on that we don't, and topics buyers ask about that nobody covers.

## You vs them
| | What | For you |
|---|---|---|
| Where we're ahead — say it louder | | |
| Where we're behind | | |
| Where everyone sounds the same | <at least 2 phrases shared by 2+ rivals> | <what to say instead> |

## Formats and angles worth testing
The top 5 formats (linked; why we think they work; how we'd make our own) and the top 5 angles (the reason to buy, who it's for, where it came from, proof status). The rest live in the angle bank for the ad makers.

## Where this disagrees with your other reports (only if it does)

## What this means for positioning
Three lines: who we're for, what we're up against, why we win — as this research sees it.

## What to do
1. <Move> — who does it · what it needs · how we'll know it worked (a number or something you can see, on a realistic timeline) · by when.
(≤ 6, ranked.)

## What we're not doing, and why
- <e.g. "Not answering Higgsfield's launch: they're 100× our reach and we can't prove we do video better.">

## What we're watching
What the next run checks, and what makes us tell you sooner.

## Baselines for next time
| Measure | Us | Each main rival | Source, date |
|---|---|---|---|
| Rank for "<rival> alternative" searches | | | |
| AI answers that name us (of the questions asked) | | | |
| Review count and rating | | | |
| Ads in the Meta library (where read) — <what this library counts, as its page says> | | | |
| Ads in the LinkedIn library (where read) — <what it counts> | | | |
| Ads in the Google library (where read) — <what it counts> | | | |
| Ads in the TikTok library (where read) — <what it counts> | | | |
Each library counts something different (`reading-ads-honestly.md`): never add them up or compare one library's count with another's, and an ad count is never spend.

## What we couldn't check
- <source — why — what it would change>

## Sources
- <title> — <link> — <date>
Source skew: <e.g. "Buyer quotes are mostly Trustpilot, which over-represents complaints.">
```

**"For you" only where a finding matters,** as a plain sentence, never shorthand ("nothing to do here" is fine): does it fit us and why (from our facts); what to make or do; the claim it rests on — approved, "needs your OK", or "you'd need proof of X first"; and who does it in plain words (static ads, video ads, organic posts, creators, landing page, email, the founder, the campaign plan; a video idea is offered as the next job with the goose-video skill). When every rival does the same thing, that is a reason to weigh something else; what buyers say outweighs what rivals do. Facts-only asks leave it out. With an empty brand brain: "I can't tell what fits you yet — tell me what you sell, to whom, and why people pick you, and I'll mark what fits."

**Never in the report** (write-report §1): angle or quote ids, file paths, tool or playbook names, "Fact / Inferred" tags (say "confirmed", "likely", "a guess, because…"), symbols instead of words, "jobs served" lines, links to local working files, process narration. When a dimension couldn't be read for every rival, say it once in plain words in "In short" and under "What we couldn't check" — never a column of question marks.

## Before you deliver — the check

Run write-report §3 (the `write-report` skill, `../../write-report/SKILL.md`): a miss blocks delivery. This playbook adds:
- **Every fact has its link and date inline**, next to the fact — not only in the sources list; the named source matches the link's site; a computed number shows its inputs.
- **"What to do":** each move names **who does it** as a role (the founder, whoever runs the website, the ad maker), a success check you can count or see, and a realistic date.
- Also banned: "worker", "maker", "free pass".

## The closing message (in chat, after the link)

```text
Do this first (because <the findings it rests on, in a few words>): <the move, who, by when>.
Report: [<title>](<local path>) · saved to your brand: [open](<link>)
Decide: 1. <one line, with my recommendation> 2. … (at most five, one line each)
<Repeat: one line with the reason and cadence — only if it would help>
```
Nothing else — no re-telling the findings.

## A rival record — `research/competitors/<rival-slug>.md` (supersede)

Line 1: `COMPETITOR RESEARCH — rivals' claims and prices, not ours. Superseded by newer dated versions.` Line 2: `Do this first, because… <the one move this rival's file implies>`. Then the rival's profile as returned by the `research-competitors-rival-pass` skill (`../../research-competitors-rival-pass/SKILL.md`), ending with its **Snapshot** block (ad ids and start dates — or the library links checked and "not read" — pricing text, headline text, open roles by team, review themes, last-checked date). Title `Competitor file: <Rival> — <Brand> — <date>`, tags `research-competitors`, `competitor-file`, `rival:<slug>`. The weekly watch diffs against this snapshot; keep the previous one as `## Previous snapshot`.

## Landscape — `research/competitors/landscape.md` (supersede)

Line 1: the same marker. Line 2: "Do this first, because…". Then the latest report's "You vs them", "How they compare" and "Formats and angles worth testing" sections, so planners have one stable file. Title `Competitor landscape — <Brand> — <date>`, tags `research-competitors`, `competitor-landscape`.

## Angle bank — `research/competitors/angle-bank.md` (supersede)

Line 1: the marker. Line 2: "Do this first, because… <brief A-NN now>". One row per angle in the shape of `angles-and-gaps.md`, ids stable across runs (`A-07`); makers tag ads `angle:a-07`. The ids live here for the makers, never in the report. Title `Angle bank — <Brand> — <date>`, tags `research-competitors`, `angle-bank`.

## What-changed reports — `research/competitors/watch/<YYYY-MM-DD>.md` (keep-history)

The shape, including the fixed line 2 that analyze-performance reads (`Week of <date> · <n> changes that matter · rivals watched: <…> · not re-checked this time: <…>`), is in `watch-and-react.md` — one place, so the two never drift. Title `Competitor watch — <Brand> — <date>`, tags `research-competitors`, `competitor-watch`.
add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall