The problem with raw job boards
Senior SEO job searching against raw job board output produces predictable noise: junior roles labelled as “specialist,” EU-only listings appearing in Canadian remote searches, titles like “Digital Marketing Coordinator” mixed with “SEO Lead,” and roles requiring skills stacks that don’t match a Local SEO / Technical SEO specialist profile.
Running this noise through manual review costs time and produces frustration, not leads. The solution is a filter-first system.
The scoring methodology
JobHarvest applies a documented scoring methodology (HOW_SCORING_WORKS.md, 688 lines1) before any lead reaches review.
Hard exclusions applied before scoring:
- EU and non-Canadian international locations (even when listed as remote)
- Junior-level roles (Coordinator, Assistant, Entry Level by title or JD signal)
- Roles requiring primary expertise outside the documented skills profile
Match threshold: 75% minimum. Below 75%, the lead is excluded without review.
Scoring inputs:
- Role title alignment to target titles: SEO Specialist / SEO Lead / SEO Manager / Digital Marketing Specialist
- Location: remote Canada or Winnipeg on-site preferred; other on-site excluded
- Skills overlap against documented profile (Local SEO, Technical SEO, GBP, WordPress, Google Ads, Meta Ads2)
- Seniority signal from JD language and required years of experience
Resume sources used for matching: Resume.txt, resume data sheet.txt, old stuff i could beef my resume up with.txt. Not the standard one-page resume file — the one-page format obscures experience depth that the scoring system needs to assess.
Top leads identified
The system produced actionable leads at or above the 75% threshold:
- Webrunner Media — Senior SEO Lead — strong title match, Canadian remote, agency context aligned to service profile3
- CareForMix — Digital Marketing Specialist — healthcare sector, remote Canada, skills overlap confirmed3
- GCOM Support — SEO Specialist — technical SEO emphasis, confirmed Canadian, role scope aligned3
What the system does not do
JobHarvest does not auto-apply. It does not send emails. It does not generate cover letters. It produces a filtered, scored lead list and stops there. Human judgment handles everything past the scoring output.
No conversion-to-interview rate is documented. No offer data exists. The system’s value is in the front-of-funnel: eliminating noise and surfacing leads worth manual time.
Sources
HOW_SCORING_WORKS.md. 688 lines. Documents the full scoring methodology, threshold rules, and exclusion logic. ↩︎
Resume data sheet.txt and Resume.txt. Skills profile source files used for match scoring. ↩︎
Indeed.ca; LinkedIn.ca. Job postings verified at time of lead identification, 2026. Lead details on file in job-scraper project outputs. ↩︎ ↩︎ ↩︎