How Does a Brand Scan Generate Lookalike Candidates?
Published 2026-11-14 · Reviewed 2026-11-14 · By the Ingenious AI team
The short answer
Deterministically, from your domain's name alone. The generator builds typo variants (deletion, transposition, duplication, keyboard-neighbour), homoglyph swaps, affix squats in plain and hyphenated spellings, and same-name TLD swaps — then checks every candidate against live DNS and Certificate Transparency, reporting only names confirmed to exist, ranked by priority.
What is how the scan generates candidates?
Candidate generation is the first stage of a TakedownCoop scan: the systematic enumeration of the domains an impersonator could plausibly register against your brand. It is a closed, repeatable set — the same domain always generates the same candidates — which is what makes two scans comparable and a monitoring re-scan meaningful.
What is how the scan generates candidates?
Candidate generation is the first stage of a TakedownCoop scan: the systematic enumeration of the domains an impersonator could plausibly register against your brand. It is a closed, repeatable set — the same domain always generates the same candidates — which is what makes two scans comparable and a monitoring re-scan meaningful.
The scan behind this page works in one pass over your whole street. It generates typosquat, homoglyph, affix and TLD-swap candidates from your domain, checks every one against real DNS resolution, searches Certificate Transparency logs for lookalike certificates, and tries a public, no-login check for fake profiles using your exact brand handle. For how the scan generates candidates, that means the first artefact you work from is evidence — what resolved, what holds a certificate, what could not be checked — rather than a screenshot someone sent you.
How the scan ranks and records it
Prioritisation is built into the result list. A hit that resolves and holds a certificate scores 90; resolving without a certificate, 80; registered but not resolving, 50 with a certificate and 40 without; a name seen only in Certificate Transparency logs scores 35, or 30 when it neither resolves nor is registered. You therefore deal with how the scan generates candidates in the order an impersonator would exploit it — live doors first — while the scan states plainly which checks it could not perform.
Seven families, one list, no guesses
The families stack: four typosquat kinds walk the letters of your name; the homoglyph map swaps lookalike characters; the affix list decorates the whole name in four shapes each; and the TLD list re-endings the untouched name. A set size guard keeps names shorter than three characters out, and the first-neighbour rule keeps keyboard substitutions bounded, so a typical brand yields dozens of candidates — the landing page's own description — rather than thousands of novelties.
Generation is only half the design. Each candidate then meets reality twice: a DNS check that records whether it is registered and whether it resolves, and the CT search that can both corroborate a candidate and contribute discoveries of its own. A candidate with neither a DNS record nor a certificate is simply not a hit, and the scan never presents it as one — the discipline that lets the packet say 'detected by generation, confirmed by DNS and CT' and be believed.
A worked example, end to end
A fictional studio, Fernweh Films at fernwehfilms.com, runs a scan and watches the arithmetic work: the generator produces its full candidate set across the seven families; DNS confirms four exist; CT adds a fifth the generator never made. The results page lists five hits — priorities 90, 80, 50, 40 and 35 — out of dozens checked, and the studio files on the two live hits that week while monitoring holds the rest. The other candidates are not reported as threats, because they are not — they are arithmetic.
Turning evidence into a filing
A complaint says 'this is fake'; a packet proves it. The Pro packet generator turns a confirmed hit into a channel-correct evidence packet — one for the registrar, one for the host, and separate ones for Meta, Google Ads, marketplaces and app stores where they apply — each with a filing checklist written for that channel. The honesty rule is printed in every packet: TakedownCoop prepares the evidence and never claims to submit to a channel it cannot actually submit to. Filing stays with you or your agent, which is also why the evidence has to stand on its own.
- Enter the genuine domain exactly — generation starts from its second-level name
- Review hits by family: the kinds present tell you how deliberate the squatting is
- File on confirmed, resolving hits first — generation alone is never the evidence
- Use CT discoveries to catch names outside the generated families
- Re-scan on a rhythm; the deterministic set makes changes between scans visible
After you file: tracking the case
What happens next is tracked, not hoped for. Each target becomes a case — domain, fake store, social profile or scam ad — moving through evidence collected, packet built, submitted, provider response, resolved or reappeared. TakedownCoop distinguishes what you attest (that you submitted, that a provider replied) from what is verified by re-scan (that the item is actually gone), and it schedules a recheck seven days after a case is submitted so silence from a provider becomes a prompt, not a dead end.
Where efforts like this usually go wrong
The commonest failure with how the scan generates candidates is treating it as a message instead of a case. A report typed from memory, sent from the wrong account, missing its ticket number within the week — each is a small thing, and together they are why impersonators outlast brands that were right about everything. The artefacts exist to remove memory from the loop: the evidence block is generated, not recalled; the checklist is written for the channel, not improvised; the case holds the IDs, not an inbox thread nobody can find in month two.
There is also a sequencing mistake: acting on the loudest artefact instead of the highest-priority one. A furious customer message about a dormant name feels urgent; a quiet scan result showing a live, certified clone is urgent. The priority score exists to arbitrate exactly that conflict — 90 before 40, live before loud — and the teams that follow it spend their filing hours where the money is actually leaving.
Keeping it from coming back
Closing the loop matters because clones come back under new names and new accounts. Re-appearance monitoring re-scans and regenerates packets; the verified-genuine page lets a cautious customer verify your real door before paying anyone; and the case timeline — every event, in order — exports as JSON or CSV when a processor, insurer or marketplace asks how you handled it. That is the co-op model applied to how the scan generates candidates: walk the street, knock with evidence, keep watch.
Frequently asked questions
How many candidates does one scan check?
Dozens for a typical brand — the product's own landing page describes 'dozens of typosquat, homoglyph, affix and TLD-swap candidates'.
Is the candidate set the same every scan?
Yes — generation is deterministic from your domain name, which is what makes monitoring re-scans comparable.
Why are some plausible names not generated?
Bounds keep the set actionable: minimum name length, first-neighbour keyboard substitutions, and a fixed affix and TLD list. CT discoveries catch important names outside the set.
Does a generated candidate count as a finding?
No. Only DNS- or CT-confirmed candidates become hits, with a priority score. The rest are not reported as threats.
Keep exploring
TakedownCoop home — run the free brand scan · All TakedownCoop guides · Pricing — Free, Pro and Agency · Get a license key · Takedown cases · My street (dashboard) · Terms · Privacy · Ingenious AI — all platforms