Drops

Drops

Top menu → Research → Drops (/drops). A workbench for expired domains: import, vet, watch, buy only what's worth buying, then bring the bought domain back to life.

The funnel

Import → DNS pre-filter → registry check → free → enrich → vet → buy. It's shown as a stage strip with live counts:

Stage	Meaning
Imported	In the catalogue, not checked yet
No DNS	Not delegated; a candidate for the registry check
Delegated	Has DNS, so it's registered; retired for free, no registry call
Checking	Registry lookup in progress
Free / Confirmed free	Available. Confirmed needs two agreeing sources
Taken	Registered (can be watched)
no registry	The zone has no public registry (e.g. .gr). It's a stage, not a weekly retry
Rejected / Bought	Your decisions

Importing

Paste a raw list or upload a CSV. Rows are reduced to their registrable apex, and rejected junk is reported per reason. Ahrefs exports are read by column: the file's own DR and referring-domain numbers land on the rows at import, so those rows never need paid enrichment.

Availability checks

• RDAP + WHOIS. A "free" verdict needs two agreeing sources.
• Proxy pool (optional) for the registry stage: SOCKS5 carries WHOIS (port 43 is raw TCP), HTTP proxies carry RDAP. Politeness is enforced per (zone, proxy) pair. IPv6 proxies don't help here.
• Dynadot (DYNADOT_API_KEY or the Settings card) as a second opinion.
• .gr has no public registry. It answers through the easy.gr registrar API (EASY_GR_USERNAME / EASY_GR_PASSWORD, IP-allowlisted, never proxied). A taken domain's expiry date becomes a future drop date for the watch loop.
• Manual verdicts for everything else: export a zone, check it in any registrar panel, paste the answers back.

Enrichment before anyone spends

Signal	Cost
DR with its monthly series (a fall of 5+ points flags a penalty, not lost links)	free
Wayback snapshots	free
Toxicity classifier over anchors and archived history	free
Majestic TF / CF in one batched call	~$0.000002 per domain
Referring domains	paid, behind a confirm
AI history pass: reads the archived life of the domain, returns clean / topic shift / spam period	your AI key (cheap Domain history AI task)

Hard vetoes: what you can actually buy

Four conditions mean "don't buy, whatever the score":

1. a spam period in the archived history;
2. dead for more than two years;
3. a fall of ≥ 5 points across the monthly DR series;
4. Trust Flow 20+ points below a DR of 20+ (a bought profile).

A veto caps the score at 0 and shows as a red chip naming its evidence. The no-veto filter is your "what can I buy" view.

Working the list

Range filters (DR, referring domains, TF), collapsible groups ("buy in October", "defer"), bulk actions over everything matching the filter (not just the visible page, minus rows you untick), CSV export or a plain domain list.

Watch: taken domains are re-checked by the scheduler (daily; every 15 minutes once the registry says pendingDelete, hourly in redemptionPeriod). A corroborated "free" sends one notification and ends that watch.

Activation: reviving a bought domain

The Activation tab picks up where the funnel ends: from a freshly bought domain to one Google crawls and indexes again. The rule built into it: the indexer network links to donor pages, never to the asset.

1. Harvest the domain's legacy URL space from Wayback CDX (and GSC once the property exists).
2. Bundle: sitemap.xml, robots.txt, the IndexNow key file and an nginx snippet. Order matters: those three files must sit before the catch-all 301, or IndexNow silently rejects the key.
3. Sitemap + Search Console is the main channel. Add the property in GSC yourself (verification can't be automated), then submit the sitemap from the tab.
4. IndexNow for Bing/Yandex, in 10,000-URL batches.
5. Donor accelerator: doorways from the indexer network that have proven Google crawl (≥ 1,000 hits in 30 days, recomputed live) enqueue donor URLs only.
6. Measure: paste the asset host's nginx access log and the parser counts Googlebot hits (total, last 7 days, last seen).

Glue (hreflang/canonical "gluing" of a drop to a money site): builds annotation blocks in cluster mode (mutual, each page canonical to itself) or funnel mode (canonicals point at the drop), and checks a live cluster for reciprocity, self-reference, x-default, dead/noindex targets and cloaked annotations.

MCP

drops_list, drops_ingest, drops_prefilter, drops_check, drops_enrich_dr / _tf / _refdomains / _wayback, drops_history_ai, drops_toxicity, drops_dr_history, drops_groups, drops_watch, glue_plan, glue_check. Enrichments that spend units or AI credits need confirm: true.