Verify extracted website emails: five addresses, filters and full reports

01Choose an input that matches the question

Single-page mode reads one supplied page. Same-origin website mode follows discovered links and available sitemaps within that site. Import mode accepts saved HTML, text, page data or pasted content. None guarantees discovery of every address anywhere on a website. One successful request is not a completed whole-site survey.

URLs and imported content are sent to the server for parsing; organization and export happen in the browser. Remote page scripts are not executed. Login, CAPTCHA, dynamically loaded content, attachments, unlinked pages and access restrictions can leave gaps. Do not paste a private address book or credentials as though it were a public-page example.

02Complete exercise: one HTML input, five addresses, seven occurrences

Download contacts.html, matching the current built-in sample. Open website email collection, choose import/pasted source and set Source URL to https://harbor-labs.example/contact. Import the HTML or paste the entire block, then select Extract emails. Do not paste source-instruction text or an extra URL line as additional page content.

This is a fictional page. Parsing sends its content to the service but does not fetch the example domain or execute its scripts. Expect one scanned page and five distinct addresses. hello appears twice in body text and once in page data: three occurrences. Four others appear once each, giving seven matches. Multiple methods or matches do not create additional contacts.

<!doctype html>
<html lang="en"><head><title>Harbor Labs · Public contacts</title></head><body>
<h1>Harbor Labs public contacts</h1>
<p>General enquiries: hello@harbor-labs.example</p>
<p>Our published contact is hello@harbor-labs.example. This duplicate should become one email record.</p>
<p><a href="mailto:partners@harbor-labs.example?subject=Partnership">Partnerships</a></p>
<p>Press: press [at] harbor-labs [dot] example</p>
<p><span class="__cf_email__" data-cfemail="42313732322d3036022a2330202d306f2e2320316c273a232f322e27">[email protected]</span></p>
<div data-email="events@harbor-labs.example">Public events</div>
<script type="application/ld+json">{"@type":"Organization","email":"hello@harbor-labs.example"}</script>
<a href="/about#team">Our team</a><a href="/contact?utm_source=sample">Contacts</a>
<a href="https://outside.example/contact">External partner (outside website scope)</a>
<p>Fictional example addresses demonstrate extraction. They do not indicate deliverability.</p>
</body></html>

For example, retain partners@harbor-labs.example without ?subject=Partnership. RFC 6068 separates addresses from parameters such as subject; parsing does not establish that an address exists.

03Filter press, then compare the filtered and complete files

Enter press in result search. Only press@harbor-labs.example should appear. Download filtered TXT and CSV and inspect the single address. CSV includes headers and provenance; its header is not a second email record.

Keep the filter active and download the JSON and XLSX reports. JSON records must still contain five entries, with hello occurrences equal to three. Excel Emails, Sources and Scanlog represent addresses, provenance and page processing respectively. The full report ignores the result filter.

04Describe what a site scan actually covered

Begin with a reviewable page limit such as10 and decide whether to try sitemap discovery. The interface allows at most500 pages, depth0–6 and bounded storage including5,000 distinct addresses. Reaching a limit or exhausting the discovered queue does not demonstrate that every page on the website was visited.

Review successful, failed, skipped and pending pages, plus truncation warnings. A starting page that redirects outside the site prevents same-origin traversal. Partial results may still be useful when some pages fail, but retain those failures. Our editorial suggestion is to record the seed URL, date, settings and known gaps instead of claiming all valid addresses.

Work does not continue on the server after the browser closes. Pause and download a backup; import it later and explicitly resume. Backups contain addresses, source details and pending URLs, so store them according to the purpose of the material.

05Reconcile actual downloads and keep their source/version scope

Inspect each XLSX sheet: Emails has five address records; Sources has five address-to-page mappings; Scanlog includes task metadata above its page-log table, which must contain one successful page record. Count Emails/Sources headers separately; metadata rows are not pages. hello occurs three times on one page, so occurrences cannot turn into seven source pages.

Check complete email values, occurrences, sources.url and methods in JSON. Reload the correct HTML if source content is incomplete, or correct the Source URL and re-extract if it is wrong. Do not conceal the error by editing only exported provenance or renaming a report JSON as a backup.

For several outputs, use column-aligned merging while retaining provenance, and CSV record/value checks for handoff. Merging cannot recover unread pages, verify ownership or establish permission to contact anyone.