Source: lore thread, 2026-09-02 Permalink: https://ryevalley.slack.com/archives/C09QEFHJW7P/p1788362765960019 From: Emma (Lore account lead), in response to Hermes pipeline proposal
Key guidance (verbatim capture for pipeline update)
Pain track and relevance gate
The DataForSEO numbers are not 100% accurate. Logic must be applied: if it is a good, obvious pain point the company was built for and solves, we do want to hit it, and that can override the low-traffic default even on small numbers.
How “obvious pain point” is decided: first crawl the website. What do they do? What problems do they solve? What industries or locations do they cater for? Full deep dive to see who they actually serve, what they do, and who it solves. Everything we do should span off this, including sense-checking keywords: does it actually align with what the whole business does? That crawl is the relevance gate for the pain track. Without it, the track fills up with plausible-sounding phrasings nobody asks.
Client review format
A zero-volume row on its own would get rejected by most clients. Instead of adding a number, tag it “context setting” or similar. Pair the tag with a one-line “why”: the problem it maps to and the part of the business that solves it. Then it reads as intentional rather than unjustified.
Track structure
Do not keep tracks separate. Rank together (one slate, one priority order). Pain-phrased rows get tagged so it is clear why they are there, not a second-class column off to the side.
Citation and link lanes (important distinction)
If competitors are being cited by AI engines, that is relevant to link building, not keyword selection. That is more on the link agent than the keyword selection skill.
Two separate criteria:
- Keyword slate admission gate: relevance (does it align with what the business does and who it serves?)
- Link budget allocation gate: volume still matters here (prioritize actual links for the most valuable keywords for growth)
For natural link acquisition and coverage (product round-ups, competitor vs feature pages), the pain/citation data is useful. But do not drop volume criteria everywhere - the two lanes have different filters.
Emma: “Awesome to run on a client you already have so I can check in.”
Action needed
The Lore keyword pipeline skill / agent prompt needs updating to incorporate:
- Website crawl as mandatory first step before any keyword selection
- “Context setting” tag with one-line rationale for zero-volume pain-track rows
- Single ranked slate (no separate pain column)
- Separate relevance gate (slate admission) from volume gate (link budget)
Emma offered to review the first live run against an existing client.