Registrar spread: how far to take it

Two. For nearly every estate the workable answer is two registrars, and not for symmetry: one carries the routine bulk, the second carries the small set of names whose loss would stop the business. Past two, administrative cost climbs faster than risk falls, and the risk you add by spreading is the one that actually takes sites offline.

What spread genuinely buys

• Containment of an account-level failure. A suspension, a card that silently stops working, a stolen login, a billing dispute that drags on for weeks: the cause varies, the shape does not. With everything in one account, one of those events reaches every name you own at once.
• Less of the estate visible from a single vantage point. Anyone who sees inside one account sees every domain in it, including the ones you never intended to associate publicly. Registrant privacy services address public lookups; they do nothing about the account listing itself.
• Freedom from one company's policy quirks. Registrars differ on which TLDs they carry, how quickly they release authorisation codes, how they handle disputes, and how aggressively they lock accounts after a suspicious login.

What spread costs, and why that cost is underrated

Domains are rarely lost to seizure, compromise or a registrar going dark. They are lost to expiry. Somebody left the company, the card on file aged out, the renewal notice landed in a mailbox nobody reads, and the name sat in the account everyone had forgotten. Every registrar you add multiplies logins, notification addresses, payment methods and renewal calendars, and each multiplication makes the forgotten-account failure more likely rather than less.

The secondary costs are smaller but real: no single consolidated renewal view, no bulk operations across the whole estate, and slower reaction when you need to change nameservers everywhere in one sitting. Weigh those against a failure mode that most operators will never experience, and heavy spread looks less clever than it first sounds.

A rough scale

The middle position, step by step

1. Choose a primary. Routine registrations and renewals all live there, so there is one obvious place to look.
2. Name the critical set explicitly, usually one to five: the main brand, the mail domain, anything that would break authentication or invoicing if it went dark.
3. Hold that set at a second registrar, with a different payment card and a different notification mailbox.
4. Break the circular dependency. Renewal notices for a domain must never be delivered to mail that the same domain serves.
5. Register the critical set for multiple years, enable auto-renew in both accounts, and keep one register of record listing every name, its registrar, its expiry and where its authorisation code can be retrieved.

Where the boundary should fall

Split by function, never alphabetically or by whim. Keep a live site's domain, its redirect variants and its defensive misspellings together in one account, because those move as a group during any change. The line worth drawing is between what earns and what merely exists.

Before you split anything

• Fix single-account exposure first: unique credentials, app or hardware 2FA, registrar lock on, and recovery that does not route through mail the domain itself hosts.
• Verify contact records at both registrars, since inaccurate registrant data is grounds for suspension under ICANN policy.
• Plan the move as a project, not an afternoon. A transfer takes days, and a fresh 60-day lock follows it.
• Decide who checks the register of record, and on which date each quarter.

Spread hedges a rare failure; concentration hedges a common one. Two accounts and one written rule about what lives in each gives you most of the protection and almost none of the confusion.