Research method
Expandi-oriented work is broken into explicit rows so an SDR manager can inspect inputs, actions, outputs, and limits without navigating a feature catalogue.
Write the ideal customer profile, geography, account exclusions, relevant functions, seniority range, and a defensible outreach purpose. A broad title search is not a campaign specification. Preserve unknowns when a profile or employer cannot be resolved.
Evidence: saved criteria and an exclusion sample. Limit: profile availability and freshness vary.
Review identity, current employer, role responsibility, and source date before a person enters a sequence. Similar names, holding companies, agencies, and stale roles should enter an exception queue instead of being forced into a confident match.
Evidence: source URL, checked date, match rationale. Limit: a job title does not prove buying authority.
Specify connection, LinkedIn message, email, and manual steps with delays, branch rules, limits, and stop conditions. Document what happens after acceptance, reply, bounce, opt-out, or ambiguous sentiment. Do not let a follow-up cross a suppression boundary.
Evidence: sequence diagram and test cohort. Limit: connected channels have separate policies.
Use only verified, relevant facts. Separate the observation from the inference it supports, avoid sensitive personalization, and ensure a reviewer can reject or edit every proposed message. Variation should improve relevance, not disguise repeated volume.
Evidence: approved examples and rejection reasons. Limit: automation cannot guarantee replies or deliverability.
Measure acceptance, replies, opt-outs, bounces, review time, and correction work on a defined cohort. Pause when account health, complaint signals, or match quality move outside the team's threshold. Store deletion and revocation procedures beside the campaign record.
Evidence: dated run log. Limit: outcomes depend on audience, offer, account state, and execution.
Install the skill, choose ten profiles whose correct disposition is already known, and compare the output with your acceptance policy.
npx -y @okki-global/okki-go-taroball