Do aged domains improve cold email reply rates?

What does the aged-domain data actually say?

We read cumulative sender-mailbox counters from three ReplyLead-operated sending programmes—replylead, CC1 and CC2—and checked the current registration date of each sending domain against registry RDAP. Of 186 domains, 185 had a registration event; 8 met the 10-year cutoff, but only 4 had recorded sends. This is a snapshot of the mailboxes still present in those workspaces, not a random sample of B2B outbound campaigns.

The older group recorded 9,106 sent messages, 5,402 contacted-lead counts and 238 mailbox-level unique replies. The newer group recorded 283,692, 167,704 and 5,342, respectively. One domain lacks a registry age and is excluded from the two-group comparison. “Sent” is a platform counter, not proof of inbox delivery.

Outlook and secure email gateways: ReplyLead's operating team also observed better deliverability to Outlook-addressed and SEG-protected recipients with older domains. The public counts below do not include recipient-level inbox-placement tests, so they cannot quantify that observation.

The denominators behind the headline

Our numerator is the source system’s mailbox unique_replied_count; the denominator is total_leads_contacted_count. These are cumulative counters for a current mailbox. Summing them can count a prospect more than once across mailboxes or campaigns. The alternative replies-per-sent figures are 2.61% for older and 1.88% for newer domains; they answer a different denominator question, not a delivery question.

Programme composition changes the comparison

The pooled gap is larger than the within-programme gaps shown here. CC2 has no older-domain sends in this snapshot, so it cannot supply an age comparison. Provider mix also differs: the older Google Workspace cell contains just 248 contacts, while 5,154 older-domain contacts use Microsoft 365. This imbalance blocks a credible Google-versus-Microsoft or domain-age conclusion.

Redirects and sending setup

A current HTTPS check found 63 domains returning an external 301 redirect and 123 returning 200 without a redirect; none returned a 302 in this census. The check happened after the recorded sends and cannot show whether the same redirect existed when a prospect received an email. All 727 current mailboxes in this extract use Google Workspace or Microsoft OAuth, so it contains no independent non-OAuth infrastructure arm. A redirect and a mailbox platform are separate variables; neither can be isolated by the pooled age comparison.

Why this is a preliminary benchmark audit

Current mailbox counters also fail to reconcile with campaign totals: the three workspaces report 293,450 sends on current mailboxes versus 459,464 on campaigns. Missing historical senders or different counter scopes may explain part of this gap, but the API evidence does not isolate the cause. We use one internally consistent mailbox grain for the table and expose the mismatch rather than silently merging sources.

How we would isolate what moves performance

A proper follow-up is a preregistered, prospective allocation test. Randomly assign eligible prospect accounts within the same campaign and audience block to matched sending mailboxes. Keep the offer, copy variants, cadence, list source, send-time distribution, daily volume, authentication and reply window aligned. Analyse at the assigned-account level and account for clustering by sending domain. The comparison estimates the effect of assignment to the available older-versus-newer domain bundles; domain age itself cannot be randomized or separated perfectly from accumulated reputation.

Predeclare one primary outcome—unique human reply from an assigned account within 21 days of its first send—and report auto, negative and positive replies separately. Include all assigned accounts in the primary analysis; publish exclusions, missing joins, crossovers, counts and intervals. Use domain-clustered uncertainty estimates and report by campaign, provider, SMB/enterprise and SEG environment, even if the result is inconclusive. Do not launch a sending test or change live domain redirects just to fill the matrix.

Assignment, balance and feasibility

The planned unit of assignment is the prospect account. Freeze account eligibility and one recipient per account before assigning each account within campaign, audience, company-size, recipient-security and send-week blocks. Balance old/new domain bundles on provider, offer, sender identity, authentication, warmup, daily volume and prior complaint history where possible; randomly allocate accounts to those bundles. A prospective assignment can reduce audience bias, but it cannot randomize registry age or erase unknown prior domain reputation.

Redirect treatment would be assigned to domains and recorded at send time; provider and infrastructure comparisons require genuinely matched, independently operated arms. A single full factorial with two age bands, three redirect states, two providers, two infrastructure routes, two company-size bands and three security classes has 144 cells before campaign strata. Four older domains with sends cannot support that design. Run staged, adequately replicated contrasts instead, and publish a power calculation from a pilot baseline before sending. Do not report an effect for an empty or underpowered cell.

Event-level release specification

A future de-identified release needs an assignment table, a message-event table and an outcome table joined by stable pseudonymous account, domain, mailbox and campaign IDs. Capture assignment time, intended arm, actual arm, send time, SMTP response, bounce category, recipient MX and verified SEG class, provider/tenant/route, authentication result, redirect chain at send time, registry age at send, reply time and automatic/human/positive labels. Link booked and held meetings and sales-accepted opportunities with their dates and a declared attribution window. Publish raw counts, crossovers, exclusions, missing joins, and the code that constructs every rate.

The primary denominator is all assigned eligible accounts; the primary numerator is accounts with a unique human reply within 21 days of first send. Predeclare secondary positive-reply, held-meeting and accepted-opportunity outcomes. Analyse assignment groups as assigned, show per-arm counts and 95% intervals that account for domain clustering, then show provider, size and security strata. Keep undetermined delivery or SEG status as an explicit missing category. Public event files must exclude prospect identities, email bodies and commercially sensitive account details.

Download the observations and check the math

The downloadable CSV contains one row per current mailbox: pseudonymous mailbox and domain IDs, programme, provider, registration-age cohort/year, current HTTPS redirect class, account status, and the four unrounded source-system counters. It contains no prospect names, addresses, reply bodies or sending-domain names. Every rate on this page is reproducible by summing its numerator and denominator rows before dividing—never by averaging displayed percentages.

Full raw delivery and conversion event files are not published here because those linked events do not exist in this extraction. Calling sent counters “delivered” or interested reply messages “won deals” would make the benchmark irreproducible. The controlled-study protocol above defines the event-level dataset needed before making those claims. The private audit retains exact source snapshots and source hashes for internal verification.

Source: read-only sender-mailbox inventory of replylead, CC1 and CC2, extracted 20 September 2026 UTC; registry RDAP registration events; HTTPS redirect census on the same date. The registration lookup followed the IANA RDAP bootstrap registry where available and the .co registry endpoint for .co domains. One .io domain had no IANA RDAP bootstrap entry and is marked unknown. For interpretation of sender authentication and recipient filtering, see Google's sender guidance and Microsoft's guidance on complex mail routing; those documents do not validate the performance of these campaigns.