What we checked.
What a rating means.

Ratings measure testing priority

5 / First benchmark

High relevance, strong public evidence and useful integration leverage for its role.

4 / Strong specialist

A credible candidate worth including in a role-specific comparison.

3 / Situational

Useful when the channel, workflow or vertical fits.

2 / Investigate

Thin evidence, substantial dependencies, or an adjacent job.

1 / Archive or hold

Retired, ambiguous, missing usable evidence, or unsuitable for a new dependency today.

Ratings consider task fit, current evidence, portability, integration leverage and continuity burden. They are editorial judgement, not measured accuracy, reliability, security, ROI or a weighted benchmark.

Read the public status literally

Documented: current product/technical documentation was readable. This does not always mean an unrestricted API. Public offering: a product is advertised, but execution or a detailed API contract was not established.

Changed: a rebrand, redirect or product surface differs from the historical reference. Unconfirmed: evidence is missing, blocked, thin or ambiguous. DNS, TLS or HTTP errors alone do not establish closure.

Retired / retiring: an official lifecycle notice. Placeholder / unrelated destination: observed current page behavior. Code available: source was accessible, not executed.

Confidence applies to this public-status assessment, not overall product quality.

Execution remains untested

No paid or authenticated vendor data jobs, export orders, mailbox purchases or outbound campaigns were run for these ratings. One public Jina Reader content fetch returned a cached page; this narrow check does not establish broad extraction reliability.

Vendor accuracy, scale, safety and compliance statements are claims rather than independently verified outcomes. No normalized price winner is declared. “Comparison pick” marks a candidate added for the modern workflow comparison; it does not mean recently founded.

Run a fair benchmark

1. Freeze a representative account/contact seed set and record timestamps.
2. Compare tools within the same job using identical inputs.
3. Verify correct-person matches, fresh evidence, unique accepted outputs, errors and exportability.
4. Include subscription minimums, credits, retries, overlaps, cleanup and maintenance in cost.
5. Test activation separately using owned accounts before any prospect outreach.

Cost per usable unique record = total workflow cost ÷ correctly matched, qualified, usable unique records

Collections and privacy

The directory has no account requirement, advertising trackers or affiliate links in this edition. Saved collections are stored in this browser. Shared collection links contain public tool identifiers. Clearing browser storage clears the saved collection.

The public dataset contains vendor information and editorial assessments. Private bookmark provenance, account details and research notes are excluded.