Brand Logo

Research operations

Salesloft defines prospecting as an inspectable method

A compact specification separates inputs, transformations, evidence, and approval decisions so CROs, founders, RevOps teams, and GTM engineers can understand why an account entered a list.

Service specification

ICP framingFirmographic and technographic criteria, geography, exclusions, confidence fields, and the minimum public evidence required for inclusion.
Account researchObservable company context, role relevance, retrieval timing, and a distinction between sourced facts and inferred fit.
Contact validationProvider provenance and verification state kept separate from legal permission, sender authentication, and inbox placement.
Outreach preparationDrafting grounded in approved evidence, with human checkpoints for relevance, factual accuracy, suppression, tone, and channel policy.
Evaluation outputAccepted-match rate, reviewer overrides, source completeness, provider cost, review time, and downstream outcomes measured on a controlled sample.

Email verification is evidence about an address at one moment; it is not a promise of inbox placement. SPF, DKIM, DMARC, mailbox health, and applicable regional requirements remain operating controls owned by the sender.

Five-step methodology

  1. 01 · Frame.

    Translate a market thesis into explicit inclusion and exclusion rules. A narrow prompt reduces false positives and makes disagreement inspectable.

  2. 02 · Collect.

    Gather company evidence, role signals, and configured enrichment fields while recording source context and retrieval timing.

  3. 03 · Reconcile.

    Use waterfall enrichment deliberately. Preserve provider disagreement, missing fields, rate-limit effects, and transformations instead of hiding them behind a single verified label.

  4. 04 · Review.

    A person approves targeting, contact choice, outbound language, suppression, opt-out handling, and the legal standard relevant to the recipient and region.

  5. 05 · Measure.

    Track accepted accounts, stale sources, reviewer overrides, bounce outcomes, and positive replies using your own labeled dataset. Performance varies with territory, provider, query specificity, and time.

The method does not claim universal accuracy, guaranteed deliverability, or automatic compliance. It provides a reproducible task definition that can be tested under the buyer’s own runtime, provider, and governance constraints.

Run the method inside your agent

Copy the installer, inspect what it requests, then configure a narrow test workflow.

npx -y @okki-global/okki-go-taroball