Sponsored Newsletter Finder

Sponsored Newsletter Finder

A discovery + evaluation composite. Discovery is keyless (the host agent's web search + directory pages); the deterministic scripts turn chosen URLs into structured data and estimate cost bands. Scoring, ranking, tiering, budget allocation, and outreach templates are the host agent's job over that data.

When to use

• "Find newsletters I can sponsor to reach [ICP]."
• "What newsletters does my target audience read?"
• "Give me a ranked list of newsletters for [industry/role]."

Feeds newsletter-signal-scanner (which newsletters to subscribe to and monitor).

Workflow

1. Discover (agent web search)

Run multiple query angles to surface candidate newsletters: "[industry] newsletter" sponsorship, site:substack.com [topic], site:beehiiv.com [topic], "best newsletters for [role]", plus directories (newsletter.directory, paved.com, swapstack.co, sparkloop.co). Collect candidate newsletter + About/Advertise URLs into a list.

SERP path (if-set/else): if DATAFORSEO_LOGIN/DATAFORSEO_PASSWORD (or SERPER_API_KEY) are set, run those query angles through the paid SERP for broader, more reliable result coverage. If not, fall back to the host agent's keyless web search (the default) — same queries, fewer guaranteed results. Either way, extraction (step 2) is keyless.

2. Extract candidate pages (deterministic)

python3 ${SKILL_DIR}/scripts/page_extract.py \
  --urls https://news.example.com/advertise https://news.example.com/about \
  --output ${WORKSPACE}/newsletters_raw.json
# or
python3 ${SKILL_DIR}/scripts/page_extract.py --urls-file ${WORKSPACE}/urls.txt \
  --output ${WORKSPACE}/newsletters_raw.json

Returns per page: {url, title, meta_description, headings, text, links, domain, signals:{subscribers, open_rate, cpm, flat_rate}}.

3. Estimate cost bands where pricing is undisclosed

python3 ${SKILL_DIR}/scripts/cpm_estimate.py --subscribers 18000 --open-rate 40 --json

Returns {size_band, cost_min_usd, cost_max_usd, implied_cpm_min/max, estimate:true} from the design's size benchmarks (micro/small/mid/large).

4. Score, tier & write the shortlist (agent)

Read newsletters_raw.json, score each 1-5 on Audience match / Reach / Engagement / Niche specificity / Sponsor accessibility (total /25), apply the CPM estimates, cross-reference competitor-sponsorship findings (search "[competitor]" "sponsored by" newsletter), tier (T1 20-25, T2 15-19, T3 10-14), allocate budget within the user's range, and write the markdown digest + cold-outreach / media-kit-request templates to a workspace file.

Outputs

A markdown shortlist (${WORKSPACE}/newsletter-sponsors-[DATE].md): Tier 1/2/3 with URL, subscribers, open rate, audience, frequency, sponsorship type, estimated cost + CPM, past sponsors, fit rationale, sponsor-page link; a "where competitors advertise" table; a budget table; and outreach templates.

Credentials / env

• Required: none — discovery and extraction are keyless.
• Optional (paid, if-set/else):
  • DATAFORSEO_LOGIN / DATAFORSEO_PASSWORD (or SERPER_API_KEY) — if set → paid SERP for broader newsletter discovery (step 1); else → the host agent's keyless web search (the default). Extraction (step 2) is keyless either way.
  • ANTHROPIC_API_KEY / OPENAI_API_KEY — if set → optional script-side synthesis; else → the host agent scores/tiers/writes the shortlist (the default; without any LLM at all, output degrades to a raw discovered list).
  • SUPABASE_URL / SUPABASE_KEY or AIRTABLE_API_KEY — if set → persist the shortlist as a pipeline; else → workspace markdown file (the default).

Notes & edge cases

• Subscriber counts and open rates are often undisclosed — use cpm_estimate.py and label outputs as estimates.
• A competitor already sponsoring a newsletter = validated fit → bump priority.
• Best for teams with ~$500-5,000/month and a specific (not mass-market) ICP.
• Route discovery through a proxy/geo when running many queries to avoid IP blocks.