gooseworks-ai/goose-skills / skills/marketing/research-competitors
research-competitors
VerifiedCompetitor 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.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 11.7 KB |
| references/angles-and-gaps.md | markdown | 7.4 KB |
| references/battlecard.md | markdown | 3.8 KB |
| references/focused-outputs.md | markdown | 2.7 KB |
| references/linkedin-engagers.md | markdown | 6.3 KB |
| references/page-technology.md | markdown | 2.7 KB |
| references/reading-ads-honestly.md | markdown | 8.6 KB |
| references/report-format.md | markdown | 10.0 KB |
| references/research-spec.md | markdown | 34.5 KB |
| references/rival-craft.md | markdown | 16.0 KB |
| references/run-memory.md | markdown | 6.6 KB |
| references/watch-and-react.md | markdown | 12.7 KB |
| skill.meta.json | config | 5.5 KB |
| tests/examples/2026-10-22/battlecard.md | markdown | 5.5 KB |
| tests/examples/2026-10-22/engagement-report.md | markdown | 3.7 KB |
| tests/examples/2026-10-22/engagers.csv | other | 1.7 KB |
| tests/examples/2026-10-22/excluded.csv | other | 1.0 KB |
| tests/examples/2026-10-22/technology.md | markdown | 3.2 KB |
| tests/fixtures/engagers.csv | other | 1.9 KB |
| tests/fixtures/focused-evidence.md | markdown | 3.8 KB |
| tests/fixtures/page-source.html | other | 614 B |
| tests/test-by-hand.md | markdown | 7.5 KB |
references/research-spec.md
View on GitHub# Competitor Research: What Great Looks Like
## 1. Summary
- **Competitor research tells you what to do next.** It is not a list of ads. Each run ends with at most six moves the founder can make, each dated and sourced, and one of them first.
- **Cover every dimension for every main rival before you report anything:** news, launches, ads on every platform, organic and founder posts, pricing, positioning, hiring, reviews and community talk. Stopping after one source is how a real test run missed its biggest story: a rival's shutdown, one news search away.
- **Look for news first.** A rival's shutdown, price rise, outage, launch or acquisition is a short window to act. These go at the top of the report, above any ad analysis.
- **Copy formats, never words.** Use a rival's demo structure, hook pattern or offer type. Never copy their copy, footage or claims. Every comparison must be true, provable and fair.
- **Deliver a readable document, not a chat reply.** It opens with In short in the write-report shape (`../../write-report/SKILL.md` §1), then timely openings, rival profiles, a positioning map, a swipe file with links, and an angle list. It is saved to the brand's memory so the next run only adds what changed.
---
## 2. The jobs: why founders and marketers research competitors
People rarely want "competitor research" for its own sake. They want help with one of these seven jobs. Work out which ones apply before you start, because they decide which dimensions get the most depth.
| Job | Questions it answers | Decision it feeds |
|---|---|---|
| **Positioning** | What does each rival claim to be, and for whom? Which seat at the table is empty? Where do buyers see us as the same? | Category, target customer, one-line pitch, homepage hero |
| **Messaging** | Which words, pains and promises does everyone use? Which do buyers use that nobody uses? Which claims are tired? | Headlines, ad copy, sales talk track, objection handling |
| **Creative** | Which formats, hooks and styles are rivals betting on? Which ones keep running, which suggests they work? What are creators making about this category? | Next ad and content briefs, which formats to test |
| **Pricing and packaging** | How do they charge (per seat, per use, per outcome)? What is free? Where does the price jump? What has changed lately? | Our price, plan names, free tier, discount and switch offers |
| **Product** | What did they ship recently? What is on their roadmap, judging by hiring, changelogs and talks? Where do buyers say they fall short? | Roadmap calls, where to win on features, what to say "we already do" about |
| **Timing and openings** | Has anything just happened (shutdown, price rise, outage, bad press, acquisition, leadership exit, end-of-life for a product)? | Switch campaigns, rapid-response content, sales outreach lists |
| **Threats** | Who just raised money, hired a sales team, partnered with a platform we rely on, or started bidding on our name? | Defensive moves, budget, where to double down |
**A good report says which jobs it served.** A founder who asks "what are they running and what's working?" mostly wants **creative + messaging + timing**. But any timely opening found along the way always goes at the top, whatever was asked.
---
## 3. The dimensions to research
For each dimension: **what to look for**, **why it matters**, **best sources (free first)**, and **how deep to go**.
"Main rivals" means the 3 to 6 the founder names or that buyers compare us to most. Also scan 2 to 4 **"adjacent" rivals**: the spreadsheet, agency, incumbent suite or DIY option buyers use instead. Note any **new entrant** you find along the way.
### 3.1 News, funding, M&A, shutdowns, leadership changes (check first)
- **Look for:** funding rounds, acquisitions (both ways), shutdowns or "sunsetting" notices, layoffs, new CEO, CMO or CRO, rebrands, lawsuits, outages and security incidents, price-rise emails, big customer wins and losses, analyst mentions.
- **Why:** this is where time-boxed openings come from. A shutdown, price rise or outage means a switch campaign. A funding round means more ad spend soon. A new CMO means new messaging within about 90 days.
- **Name the event exactly.** An acquisition is several different events. A company buying a shut-down rival's assets or customers (e.g. one recruiting tool buying another's assets) is running a **migration offer to the customers it inherited** and will try to own "<rival> alternative" searches — for us that's a race for the customers it doesn't keep and for those searches, not a "switch campaign" it is running. A rival buying a company → a bundle or new feature soon. A rival being bought → roadmap and pricing doubts among its buyers. A shutdown → a home for their customers' work.
- **Sources (free):** web and news search on `"<rival>"` plus each keyword (shut down, acquired, raises, layoffs, pricing, outage, sunset, CEO), limited to the last 90 days. Also the rival's own blog and press page, their LinkedIn company page, Crunchbase or PitchBook free pages, TechCrunch, trade press, status pages (status.<rival>.com), and the founder's own posts, where news often breaks first. Then run the **same search on our own brand**. In one real case the biggest story — buying a shut-down rival's assets — was the founder's own news.
- **Depth:** every main rival, last 90 days, every result read. Any hit from the last 30 days becomes a candidate "timely opening".
### 3.2 Paid ads, across every platform
- **Look for:** every active ad, the start date of each, how many versions of the same idea exist, format (static, video, carousel, document, conversation ad, search text), hook (first line or first 3 seconds), offer and CTA, landing page, the audience it seems to target, and whether it runs in several countries.
- **Longevity as a sign of a winner:** an ad running for **60+ days**, especially with **several versions of the same idea**, probably performs. Advertisers kill losers. Also tell the founder where this sign is weak:
- Low-spend B2B accounts often leave ads running because nobody checks.
- Brand-awareness ads run long whatever their results.
- New ads (under 14 days) are tests. They show what the rival is trying, not what works.
- Say "**likely winner (inferred from run length)**", never "this works".
- **Why:** shows which formats and angles rivals have bet money on, which offers they push, and where they send traffic.
- **Sources (all free, check every one for every rival):**
- **Meta Ad Library** (Facebook, Instagram, Messenger, Audience Network). Shows all active ads and start dates. EU searches show reach and targeting.
- **Google Ads Transparency Center** (Search, YouTube, Display, Shopping). Shows format, region and dates last shown. Note it does not show the keywords bought. To find who bids on our brand, search our brand name in a clean browser in several regions.
- **LinkedIn Ad Library** (linkedin.com/ad-library, no login). **For B2B this is the most important library.** Shows ads from the last year, with impression ranges for EU ads.
- **TikTok:** Commercial Content Library, plus Creative Center "Top Ads" for what works across the category.
- **X Ads Repository** (EU), **Reddit** (look at promoted posts in relevant subreddits), and **Microsoft/Bing**, which matters for older or enterprise buyers.
- Paid tools if available (Semrush/SpyFu for search ads and keywords, plus ad-intel tools for creative). Treat these as bonus data, not the baseline.
- **Depth:** every library, every main rival. Pick the **top 5 to 10 ads per rival** by run length and number of versions. Watch every video ad. Open every landing page. Record the **"nothing found" results too**. "No LinkedIn ads" tells you something.
### 3.3 Organic content, platform by platform
- **Look for:** the content mix (education, product, culture, memes, customer stories), how often they post, the formats that do better than their usual (2 to 3 times their typical engagement), recurring series, and which topics they own.
- **Why:** organic posts that do well are free creative tests. They show which ideas a rival might turn into ads next, and which topics are still unclaimed.
- **Sources:** each rival's company accounts on LinkedIn, X, YouTube, TikTok, Instagram and Threads (weight by company type, see section 5), plus their blog, newsletter and podcast.
- **Depth:** last 30 to 60 days per platform. Find each rival's **top 3 posts by engagement relative to its own average**, not by raw numbers. Skip platforms where a rival is clearly inactive, and say so.
### 3.4 Founder and executive voice
- **Look for:** what the CEO, founders, CMO and head of product post, and the stories they tell over and over. Opinions on the category ("search is dead", "agents replace the old workflow"). Teasers of what's coming. Fights they pick. Hires and customers they announce.
- **Why:** in B2B and founder-led consumer brands, **the founder's feed is often the real marketing**. Roadmap hints, new positioning and launch dates often appear there weeks before the website changes. It also shows which story the market is starting to believe.
- **Sources:** LinkedIn and X personal profiles, podcast guest spots (search "<name> podcast"), conference talks on YouTube, and Substacks.
- **Depth:** the 1 to 3 most visible people per main rival, last 30 to 60 days, top posts by engagement. Note any **claim or prediction** they repeat. Those become things we can agree with or argue against.
### 3.5 Product launches and releases
- **Look for:** new features, models, integrations and tiers. The launch video or demo. The launch post. The Product Hunt page and comments. The changelog. **What others posted about it**: creator demos, reaction videos, "I tried X" threads, YouTube walkthroughs.
- **Why:** launches give us (a) **formats** that already work, because the market shared them, (b) a **window to counter-position** if we already do the thing, or do it better, and (c) a sense of where the category is heading.
- **Sources:** changelog and release notes pages, the product blog, Product Hunt, the rival's YouTube channel, LinkedIn and X launch posts, then searches on YouTube, TikTok, X and LinkedIn for "<rival> <feature>", "<rival> demo" and "<rival> review".
- **Depth:** every launch in the last 90 days. For the 1 to 2 biggest, **watch the demo** and **the 3 best third-party posts about it**. Write down their structure: hook, setup, reveal, proof, CTA. Note what people praised and complained about in the comments.
### 3.6 Pricing and packaging
- **Look for:** plan names, prices, what counts as a unit (seat, credit, contact, search), free trial or free tier, annual discount, what is behind "contact sales", the feature that pushes buyers up a tier, and **recent changes**.
- **Why:** sets our price, shows switch-offer openings (a price rise means a "lock in your old price with us" offer), and shows what buyers can compare side by side.
- **Sources:** pricing pages. Compare against the **Wayback Machine** for changes over 6 to 12 months. Also G2 and Capterra pricing sections, Reddit threads mentioning "<rival> pricing", and sales-call notes from the founder if he has them.
- **Depth:** every main rival. Put it in **one comparison table**. Say "price not public" where it isn't, and don't guess.
### 3.7 Positioning and messaging
- **Look for:** homepage headline and subhead, category name, who they say it's for, top 3 claimed benefits, proof shown (logos, numbers, awards), how they describe rivals, and changes over time (Wayback Machine).
- **Why:** the basis for the positioning map and the "empty seat" analysis. If everyone claims "AI-powered" and "10x faster", those words are worthless and we should find different ones.
- **Sources:** homepage, product pages, "vs" and "alternatives" pages (theirs about us is gold), meta titles, the About page, and G2 category placement.
- **Depth:** every main and adjacent rival. Pull **exact headline wording** into a table so the founder sees the sameness.
### 3.8 Funnel and offers
- **Look for:** what happens after the click. Landing page type, free trial vs demo request vs freemium, lead magnets (reports, templates, calculators), webinars, email sequences after signup, retargeting, and discount or switch offers.
- **Why:** shows how hard rivals work to convert, and which offers we can beat. A rival with only "Book a demo" leaves an opening for "Try it free in 2 minutes".
- **Sources:** click through their ads and homepage CTAs. Sign up for the newsletter or a free resource **with a dedicated research inbox, only if the founder approves and only with test or company details, never personal data**. Sign up for free trials only where it's free and allowed. Note what the onboarding emails say.
- **Depth:** the main CTA path for each main rival. Read the full email sequence over time in the ongoing watch, not in the first teardown.
### 3.9 Customers and case studies
- **Look for:** logo walls, case studies (industry, company size, result claimed), testimonials, and which segments they lean into.
- **Why:** shows who each rival really wins. That points us toward the segments they're weak in, and gives the sales team a "who else is looking" list.
- **Sources:** customer pages, case-study PDFs, G2 reviewer company sizes and industries, "powered by" footers, and LinkedIn posts that tag customers.
- **Depth:** list segments and 5 to 10 named customers per main rival. Note where a rival's story conflicts with what reviews say.
### 3.10 Reviews and complaints (the voice of the buyer)
- **Look for:** the top 5 praises and top 5 complaints per rival, in the buyers' own words. Reasons for switching ("we left X because…"). Reviews that mention us. How recent the complaints are.
- **Why:** **this is the most useful dimension for messaging.** Complaints become our angles, objection handling and comparison pages. Praise shows what we must at least match. Buyer wording beats any marketer's wording.
- **Sources:**
- B2B: G2, Capterra, TrustRadius, Gartner Peer Insights.
- Consumer: Trustpilot, App Store, Google Play, Amazon, Sitejabber.
- Everyone: Reddit, YouTube comments, X replies, Product Hunt comments.
- **Depth:** for each main rival, read **at least 30 recent reviews** (last 12 months), weighted toward 2 to 3 stars, where the real trade-offs show. Quote short, exact phrases, each with a link.
### 3.11 Community chatter
- **Look for:** threads asking "X vs Y?", "best tool for…", "alternatives to X", people asking for help, and anger after changes. Also note what the category talks about this month.
- **Why:** shows how buyers frame the choice before they meet a salesperson, and gives early warning of trouble at a rival.
- **Sources:** Reddit (relevant subreddits, e.g. the two or three subreddits where the category's buyers ask for advice), Hacker News, niche Slack and Discord groups (only ones the founder is in), LinkedIn comment sections, Quora, and Facebook groups for DTC or local.
- **Depth:** last 90 days, searching "<rival>", "<rival> alternative" and "<category> recommendation". Summarise themes, not single loud posts.
### 3.12 Events, webinars and podcasts
- **Look for:** conferences they sponsor or speak at, their own webinars and events, podcast appearances, and roadshows.
- **Why:** shows where their buyers gather, where to show up, and when a rival will announce something (big launches often line up with a flagship event).
- **Sources:** event pages, "events" and "webinars" pages, LinkedIn Events, conference speaker lists, and podcast search (Spotify, Apple, YouTube).
- **Depth:** the next 90 days of known events for each main rival, plus the last 90 days of talks (skim the 1 to 2 most watched).
### 3.13 Partnerships and integrations
- **Look for:** integration directories, marketplace listings (Salesforce AppExchange, HubSpot, Shopify App Store, the category's own marketplaces), co-marketing, resellers, and new platform partnerships.
- **Why:** shows distribution advantages and lock-in. "Native in Greenhouse" is a buying reason. A partnership with a platform we rely on is a threat.
- **Sources:** the rival's integrations page, partner marketplaces, and partner press releases.
- **Depth:** list them all once and track changes.
### 3.14 Hiring signals
- **Look for:** open roles by team, new markets (country-specific roles), new segments ("Enterprise AE", "Head of Partnerships"), new tech (an "ML engineer, voice" role), and any hiring freeze or sudden halt.
- **Why:** job posts are the most honest roadmap a company publishes. Ten sales roles in EMEA means an EMEA push. Senior marketing hires mean a messaging shift is coming.
- **Sources:** the rival's careers page (Greenhouse, Lever and Ashby boards), LinkedIn Jobs, Wellfound, and LinkedIn headcount growth on the company page.
- **Depth:** count roles by team and note any that stand out, for every main rival.
### 3.15 Search and AI-answer presence
- **Look for:** which rivals rank for the category's money searches ("best <category>", "<rival> alternative", "<rival> vs <rival>"), who bids on whose brand name, which "alternatives" and "vs" pages exist, and **what ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews say** when asked "best <category> tool" and "alternatives to <rival>".
- **Why:** these searches are bottom-of-funnel demand. AI answers increasingly shape shortlists. If AI answers leave us out, or describe us wrongly, that's a job for content and PR.
- **Sources:** manual Google searches in a clean browser, the AI assistants themselves (same prompts each time so results compare), and Semrush, Ahrefs or Similarweb if available, for traffic and keywords.
- **Depth:** 5 to 10 money queries, run against every AI assistant available. Record who is named and how.
### 3.16 Email and newsletters
- **Look for:** how often they send, subject lines, offers, launch announcements, and nurture sequences.
- **Why:** shows their offer calendar and how they talk to existing users, who are our switch targets.
- **Sources:** subscribe with a research inbox (with founder approval), plus public newsletter archives (Substack, Beehiiv) and Milled for DTC.
- **Depth:** set it up in the first teardown, then read it in the weekly watch.
---
## 4. Turning research into action
Research is only finished once it suggests specific moves. Each move in the report must point to the evidence behind it.
### The moves
| Move | When to use it | What to make |
|---|---|---|
| **Copy the format, not the words** | A rival's or creator's ad or demo format is clearly working (long-running, many versions, high engagement) | A brief: "use the same structure (hook type, length, beat order, setting), with our product, our proof and our words" |
| **Counter-position a launch** | A rival launches something we already do, or do better | Same-week content: side-by-side demo, "we've had this since…" post, founder take. Ads aimed at people searching the launch name, where policy allows |
| **Switch campaign** | Rival shutdown, acquisition, price rise, outage, removed feature or end-of-life | Landing page plus offer (migration help, price match, credit). Ads on the rival's brand terms where allowed. Outreach to their known customers. Founder post. **Time-boxed, with an end date** |
| **Comparison and alternatives pages** | Buyers search "<rival> alternative" or "<us> vs <rival>" | A fair, sourced, dated comparison page per main rival, updated when anything changes |
| **Objection handling** | Reviews show what buyers fear or hit | Sales talk track, FAQ entries, ad lines that answer the fear before it's asked |
| **Gap angles** | Everyone claims the same thing, or buyers want something nobody claims | New headline and hook angles in the white space |
| **Steal the channel** | A rival is quiet on a channel where buyers are active | Test there first. Less competition, cheaper attention |
| **Defend** | A rival bids on our brand, launches a "vs us" page, or raises and hires a lot of sales staff | Our own brand-term ads, a reply comparison page, refreshed proof, a heads-up to customer success for at-risk accounts |
### Guardrails (always on)
- **No copying copy or creatives.** Never reuse a rival's or creator's words, footage, images, music, characters or exact script. Formats, structures and *kinds* of offer are fair game. Specific creative work is not. "Same beat order, different everything else" is the test.
- **No using their marks to confuse.** Bidding on a rival's brand name is generally allowed in the US. Ad copy and landing pages must never suggest we are them or are linked to them. Using their trademarks in ad text is limited on some platforms and in some countries. Check each platform's policy before launch.
- **Comparisons must be true, provable and fair.** Every "faster, cheaper, better than X" claim needs dated evidence we can show (a head-to-head test, a public price page snapshot, a documented feature). Compare like with like and say what we compared. When a rival's facts change, update or pull the claim.
- **Run comparisons past the founder.** Some countries (the EU, UK and others) have stricter comparative advertising rules than the US. The agent flags any named comparison for founder or legal sign-off. It never auto-publishes one.
- **No disparaging.** Attack the problem, not the people. Never mock a shutdown or a rival's staff. For a shutdown, the tone is "here's a home for your work". Never make fun of their customers' bad luck.
- **Never invent proof.** No made-up customer quotes, metrics or "switched from X" stories. Use only what the founder has approved.
- **Respect sites and privacy.** Use public pages. No logging into a rival's product with fake details. No scraping that breaks a site's terms. No collecting personal data about rival employees beyond their public professional posts.
---
## 5. How it differs by company type
The dimensions stay the same. What changes is **where to spend depth**. The agent should name the company type at the start and weight accordingly.
| Company type | Weight most | Weight less | Notes |
|---|---|---|---|
| **B2B SaaS** | LinkedIn ads + organic, **founder and exec posts**, G2/Capterra reviews, pricing and packaging, launches and changelogs, hiring, integrations, comparison and alternatives search, AI-answer presence, events | Meta and TikTok ads (unless self-serve or SMB), Instagram | Long sales cycles, so messaging and proof matter more than hooks. Ad run length is a weaker sign here because spend is low. |
| **DTC / ecommerce** | Meta and TikTok ads (volume, run length, versions), TikTok and Instagram organic, creator and UGC partnerships, offers and promos, email (Milled), Amazon and Trustpilot reviews, pricing and bundles | Hiring, events, LinkedIn | Creative goes stale in weeks, so the weekly watch matters more than the deep teardown. Ad count and versions per idea are strong signs. |
| **Consumer app** | App Store and Google Play (ratings, reviews, screenshots, ranking, update notes), TikTok and Instagram ads and organic, creator content, onboarding and paywall, pricing and trial length | LinkedIn, events | Paywall and trial design are the funnel. Review complaints right after an update are switching openings. |
| **Marketplace seller** (Amazon, Etsy, Shopify marketplaces) | Listing titles, images, A+ content, price and coupon history, review themes, bestseller rank, sponsored-product placements, bundles | Founder posts, hiring, events | The listing *is* the ad. Price-tracking tools (e.g. Keepa for Amazon) show history. |
| **Local business** | Google Business Profile (reviews, photos, posts, categories), local search results and map pack, Yelp and Facebook reviews, local Meta ads, offers, hours and service area | Hiring (except as a growth sign), AI-answer, LinkedIn | Rivals are found by "near me" searches. Review volume and how fast owners reply matter most. |
---
## 6. Depth and process for the agent
### Ground rules
1. **Do the whole job before reporting.** Research every dimension in section 3 for every main rival before presenting anything. **Never stop after one source to ask "go deeper?".** The founder asked for competitor research, and depth is part of the request. The only reasons to pause midway are a missing fact only the founder knows (section 6.4) or an action that needs permission (signing up, paid tools). Both go in the closing message with the report.
2. **News first, then everything else.** Run section 3.1 (news) for every rival and for our own brand in the first minutes. If it finds a timely opening, keep going, but mark it so it leads the report.
3. **Run rivals in parallel.** One worker per rival, each with the same checklist and output shape. One more worker covers the category-wide checks (AI answers, money searches, community threads, creator demos). Merge at the end.
4. **Date and source every fact.** Each claim gets a link and an "as of" date (date seen, plus date published where known). Ads get their start date. Prices get the date captured.
5. **Label fact vs inference.** Use plain tags:
- **Fact:** seen directly, with a link.
- **Inferred:** a reasoned guess from evidence, with the reason given (e.g. "likely winner, running 94 days, 6 versions").
- **Unknown:** we couldn't find out.
Never say "this ad works" or "their best channel is X" as fact. We can't see their results.
6. **Keep a "what we could not read" list.** Every source that failed or was empty: blocked pages, login walls, libraries with no data for a region, video we couldn't transcribe, paid tools we didn't have. Say what it might have changed. Silence is not the same as "nothing there".
### What "done" looks like (checklist)
- [ ] Company type named and dimensions weighted (section 5).
- [ ] Main rivals (3 to 6) and adjacent options (2 to 4) listed, with a reason for each.
- [ ] News scan (90 days) done for every rival **and our own brand**.
- [ ] Every ad library checked for every main rival, including the empty ones.
- [ ] Top 5 to 10 ads per rival recorded with start date, format, hook, offer, landing page and link. Video ads watched.
- [ ] Organic top posts per active platform per rival.
- [ ] Founder and exec posts for 1 to 3 people per rival.
- [ ] Launches in the last 90 days. Biggest launch demo plus top third-party posts watched and broken into their structure.
- [ ] Pricing table, with changes from the Wayback Machine.
- [ ] Positioning table with exact headline wording.
- [ ] Funnel path and main offer per rival.
- [ ] Customers and segments per rival.
- [ ] At least 30 reviews per rival, condensed into top praises and complaints with quotes.
- [ ] Community themes (90 days).
- [ ] Hiring counts by team.
- [ ] Partnerships and integrations list.
- [ ] Money-search and AI-answer results.
- [ ] Events in the next 90 days.
- [ ] Every fact dated and linked. Inferences labelled.
- [ ] "What we could not read" list.
- [ ] 3 to 5 "do this first" moves, each tied to evidence and with guardrails checked.
### Ask the founder very little, find the rest
These questions go in the closing message with the report, never before research (`../../marketing-guide/research-pattern.md` (in the `marketing-guide` skill)).
**Ask, only if not already in brand memory:**
- Who are your main rivals? Who do buyers mention on sales calls? (The agent should propose a list and let him correct it, not start from blank.)
- Which claims can we make? Do we have proof for "better than X" on anything? Any approved customer stories?
- Anything off-limits? (A rival we partner with, a legal dispute, a "never name them" rule.)
- Which jobs matter most right now, if the question doesn't make it clear?
**Find, never ask:** what rivals are running, their prices, news, launches, reviews, hiring, positioning, our own recent news, and who ranks for what. If the agent can look it up, it must not ask.
### Cadence
| Mode | When | Scope | Output |
|---|---|---|---|
| **Deep teardown** | First time for a brand, then every quarter, or when a big rival moves | Everything in section 3, every main rival | The full document (section 7). Brand memory created or refreshed |
| **Weekly watch** | Every week | What changed since last time: new and stopped ads, new posts that do well, pricing-page diffs, changelog, news, hiring changes, new reviews, email | A one-page "what changed + what to do" note added to the document, with memory updated |
| **React-now alert** | Within hours of a trigger: shutdown, acquisition, price change, big launch, outage, rival bidding on our name, viral post about the category | The trigger only, plus a quick check of how the market is reacting | A short alert: what happened (sourced), the window, the suggested move with a draft brief. Sent to the founder immediately |
The weekly watch compares against **saved snapshots** in brand memory: ad IDs and start dates, pricing-page text, headline text and job counts. This is the only reliable way to see changes, especially on LinkedIn, where the library gives little structured data.
---
## 7. The output
### What the founder gets
A **readable document** (a shareable doc or artifact page), not a chat reply. The chat message is three lines: the single biggest finding, the top move, and the link.
### Document structure
1. **In short** — the write-report shape (`../../write-report/SKILL.md` §1; this playbook's version: `report-format.md`): what we looked at, the findings with how sure, then the one first move resting on them, and a line on what changed since the last report. The ranked moves (at most six, each with why now, expected upside, effort and a link to evidence) sit in "What to do", not in the summary.
2. **Timely openings** (only when there are any, and always above everything else)
- E.g. *"Rival X shut down on <date>. Its users need a new home now. Window: roughly 4 to 8 weeks, until their data is deleted."* Then the switch-campaign plan: audience, offer, channels, assets, end date, and claims that need sign-off.
3. **The landscape**
- **Positioning map:** a 2-axis chart on axes buyers actually care about (e.g. "self-serve to enterprise" by "point tool to full suite"), with each rival plotted and the empty seats marked.
- **Messaging table:** each rival's exact headline, who it's for, top claims and proof. Phrases everyone uses are highlighted as "worn out".
- **Pricing table.**
- **Channel heatmap:** rivals by channel (Meta, Google, LinkedIn, TikTok, organic platforms, founder voice, events), coloured by activity level.
4. **Rival profiles** (one per main rival, the same shape every time, about 1 page each)
- Snapshot: what they are, who they serve, size and funding, last 90 days of news.
- What they're running: top ads with links, start dates and "likely winner" tags. Organic top posts. Founder voice themes.
- Product: recent launches, roadmap hints from hiring and talks.
- Pricing and offer.
- What buyers say: top praises and complaints with short quotes.
- **Where we win, where we lose, and what to say**, with each point labelled fact or inferred.
5. **Creative swipe file**
- The best ads, launch demos and organic posts in the category, each with a link, date, platform, run length or engagement, and a **format breakdown** (hook, beats, length, style, CTA). Each has a "how we'd use this format" note, never a copy of the words.
6. **Angle list**
- 10 to 20 hooks and angles for us, each tied to a gap or complaint found, with the evidence and a claim-safety note (OK as is / needs proof / needs sign-off).
7. **Watch list and triggers**
- What we'll monitor weekly, and what would set off a react-now alert.
8. **Sources, method and "what we could not read"**
- Every source checked per rival (including the empty ones), dates, tools used and gaps.
### Where it lives
- **The document:** published as a shareable doc or page the founder can read, comment on and forward to the team.
- **Brand memory:** a structured competitor record per rival (profile, ad snapshot with IDs and start dates, pricing snapshot, headline text, hiring counts, review themes, last-checked dates) plus the dated report. The next run starts from memory and **reports what changed**, so it never begins from zero.
- **Swipe file items** also go into the brand's creative inspiration library so ad briefs can link straight to them.
---
## 8. Where I agree and disagree with the founder
**Agree, strongly:**
- **Watch what rivals and their founders post.** In B2B, the founder's feed is often ahead of the website. Leaving it out is a real gap.
- **Launch demos are format gold.** When the market makes and shares demos of a rival's launch, the format is already proven. Using it quickly is smart.
- **Depth is part of the request.** "Go deeper?" after one source fails the job. A real test run shows the cost: its most valuable finding, a rival's shutdown, was one news search away.
- **A document, not a chat summary.** Research needs to be read, shared and acted on by a team, and compared over time.
**Push back, with reasons:**
1. **"Take that format/script for our own ads": take the format, not the script.** Copying a script word for word, or reusing someone's footage, raises copyright and passing-off risk. It also makes us sound like them, which weakens positioning. Keep the beat structure and hook pattern, and write our own words with our own proof.
2. **"Advertise against a launch we do better": only when we can prove it, and not always.** Naming a rival in ads gives them free awareness. That usually pays off for a challenger against a leader, and usually doesn't for a leader against a small player. "Better" must be shown side by side and dated, or it becomes a legal and trust risk. Often the stronger move is to **demo our version the same week without naming them**, and save named comparisons for comparison pages and search ads, where the buyer is already comparing.
3. **"What works for them" can't be seen from outside.** We can't see rivals' results. Ad run length, number of versions and engagement are useful signs but not proof, especially for low-spend B2B accounts where ads run forgotten for months. The report should say "likely" and explain why, rather than overclaim.
4. **Depth must be proportional, and readable.** "Proper depth" doesn't mean a 40-page dump. The research should be exhaustive. The document should be short at the top: "do this first" on screen one, with depth below for whoever wants it.
5. **Don't let rival-watching drive the strategy.** If you only study rivals, your messaging ends up sounding like theirs. The most valuable part of this research is **what buyers say** (reviews, Reddit, sales calls), not what rivals run. I'd weight reviews and community talk as heavily as ads, and use rival ads mainly to find the empty space, not to set the agenda.
6. **Check our own news too.** In one real case the biggest story was partly the brand's own (it bought a shut-down rival's assets). The agent should always scan the brand's own recent news alongside rivals', so it doesn't "discover" the founder's own deal or contradict it.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |