ad-research

Ad research

A teardown of what a brand (or a whole market) advertises, from the public ad libraries: what they say, which ads they keep paying for, where they send people, and what to test in response. For a 4-call snapshot first, use the lurk skill.

Calling LurkAPI

With MCP (the LurkAPI plugin or connector), call the tools by name, like tiktok_profile; your client may prefix them (mcp__plugin_lurkapi_lurkapi__tiktok_profile). With only find_endpoints and call_endpoint, run a tool as call_endpoint with endpoint set to its name and params to its arguments. Results come trimmed: no nulls, one URL per image or video. A long result comes in parts: its _note gives the offset for the next part, which is free.

Without MCP (Codex, Cursor, scripts), every tool is GET https://api.lurkapi.com plus its path, with the same params as a query string (URL-encoded) and the key from LURKAPI_API_KEY in the x-api-key header:

curl -s "https://api.lurkapi.com/v1/tiktok/profile?handle=nike" -H "x-api-key: $LURKAPI_API_KEY"

If LURKAPI_API_KEY isn't set but LURKAPI_KEY is (the name older docs used), use that. Raw answers are big: pipe them through jq to keep the fields you need. No key yet? Ask the user to create one at https://lurkapi.com/dashboard (100 free credits on sign-up, then a balance below 25 tops back up to 25 every day) and to set LURKAPI_API_KEY in their shell. Never ask for the key in the chat, and never print it.

Credits. Each call costs what the table at the end says, not_found included; failures on our side (5xx) are free. Every answer carries credits_remaining. Before a run of more than 20 calls, check the balance with account_credit_balance (free) and tell the user what the run will cost. If a call answers missing_api_key, insufficient_credits or signup_required, stop and pass on what it says.

Pick the libraries

Default to Meta and Google; add LinkedIn for B2B and TikTok when Europe matters or the user asks.

Find the advertiser

• Meta: facebook_search_companies(query=<brand>), then the page_id of the right page in searchResults: the exact name, BLUE_VERIFIED in verification, the most likes. Big brands run regional pages; ask which one if it matters.
• Google: with the website, skip the lookup and pass domain below. Without it, or to catch other sites, google_search_advertisers(query=<brand>) gives advertisers[].advertiser_id with number_of_ads_estimate, and the sites in websites[].domain.
• LinkedIn: search by company directly, below.
• TikTok: tiktok_ad_library_search(query=<brand>) first, to read the advertiser_name exactly as the library spells it (often a legal entity, like a B.V. or a GmbH); then search with advertiser_name.

Pull the live ads

1. facebook_company_ads(pageId=<page_id>, status=active, trim=true): up to 30 ads, most seen first, and searchResultsCount, the estimated number live. More pages (cursor, 10 ads each) only if the first page doesn't show the patterns; two usually do. media_type=video focuses on video.
2. google_company_ads(domain=<website>) (or advertiser_id): up to 100 ads a page with format, firstShown, lastShown and daysShown. A domain lists everyone advertising the site, resellers, affiliates and agencies included: count an ad as the brand's only when its advertiserName contains the brand's name. The list has no copy: for the text and landing page of the few that matter, google_ad(url=<adUrl>), one call each.
3. linkedin_search_ads(company=<brand>): about 24 previews a page, whose text can be cut short. For full copy and the payer, linkedin_ad(url=<adLibraryUrl>) on the ones you quote.
4. tiktok_ad_library_search(advertiser_name=<name>): 12 a page, from the last 30 days. Then tiktok_ad_library_ad(ad_id=<id>) on two or three: its targeting (countries, age, gender, interests) and reach_by_country show who they're after.

For a market rather than a brand, search the ads by keyword instead: facebook_search_ads(query=<keyword>, status=active, country=US) (group the results by page_name to see who competes), tiktok_ad_library_search(query=<keyword>) and linkedin_search_ads(keyword=<keyword>).

Find the winners

Performance isn't public, but persistence is. An advertiser keeps paying for what works.

• Meta: days since start_date (Unix seconds) on an ad that's still active. Over 30 days is a keeper; the longest-running live ads are the swipe file. A high collation_count (many versions of one ad) means they're scaling or testing it.
• Google: daysShown, and a lastShown of today or yesterday.
• Repetition: the same hook or offer across many ads and formats is a deliberate bet.
• What's new: facebook_company_ads(pageId=<page_id>, status=active, sort_by=relevancy_monthly_grouped, trim=true) lists the newest launches first, which is what they're testing now.

Read them

For each library, then across them:

• Angles: the pain, desire, persona or comparison each ad leads with. Cluster them and count.
• Hooks: the opening line of snapshot.body.text (carousels: snapshot.cards[].body), verbatim.
• Offers and calls to action: discounts, trials, bundles, guarantees; snapshot.cta_text.
• Landing pages: snapshot.link_url. Several products or campaigns usually show up as several destinations.
• Formats and placements: snapshot.display_format, publisher_platform, Google's format.
• Gaps: angles a competitor never uses, and markets or platforms they skip.

Skip catalog ads whose text holds template tags like {{product.name}}; they carry no message.

Budget

One brand on Meta and Google is about 4 calls, plus one per ad you open in full. A brand on all four libraries with a few details opened is about 10 to 15. Several competitors multiply that: say the estimate first, and look at all of them shallowly before going deep on the most active one.

Report

# Ad teardown: <brand>, <date>

| Library | Live ads | Oldest live ad | Main formats |
| --- | --- | --- | --- |

## Angles
| Angle | How often | Example (quoted) | Link |

## Hooks worth stealing
- "<verbatim>" (<library>, running <n> days)

## Offers and calls to action
## Landing pages
## What they're testing
Ads with many versions, recent launches, new angles.

## Tests for us
3 to 5, each tied to evidence above.

Link every ad you cite: Meta https://www.facebook.com/ads/library/?id=<ad_archive_id>, Google its adUrl, LinkedIn its adLibraryUrl, TikTok its url.

Pitfalls

• Don't call an ad a winner because it's live; run time is the signal, and even that is a proxy.
• Spend, impressions and targeting are public only for political and issue ads (and in the EU, some reach data). Don't invent them.
• Google ads more than about a year old don't come back.
• TikTok's library is Europe only and LinkedIn's about a year back: an empty answer limits what you can say, nothing more.

Calls and credits

Each tool's link opens its docs as markdown: every param and its values, and a real response.