Cold email in English-speaking markets tends to be a one-ISP problem. Gmail dominates, Outlook is a secondary concern, and everything else is noise. If Gmail is delivering, you ship.
Russia is not like that. Four providers matter — Mail.ru, Yandex, Rambler and Gmail (in that order of share for consumer mail) — and their filters disagree more than any Western ISP cluster does. The same message, sent to a seed list covering all four, will often land in Inbox on one and Spam on another. This article is the data from a sample of our own test runs plus the concrete implications for a cold-email sequence.
GlockApps, the most-used placement tool in English-speaking markets, does not seed Mail.ru, Yandex or Rambler. Nor do Mailgenius or Litmus. If your vendor cannot show you a per-provider Mail.ru number, it is not measuring what it claims to measure.
Methodology
We ran the same English-language cold-email sequence (three-touch, conversational tone, no images, single tracked link) to a dedicated seed pool of:
- 8 × Gmail personal + Google Workspace seeds
- 8 × Mail.ru + bk.ru + inbox.ru seeds
- 8 × Yandex.ru + ya.ru seeds
- 4 × Rambler.ru + lenta.ru seeds
The sending domain was warmed for 6 weeks with standard engagement patterns. SPF, DKIM (2048-bit) and DMARC (p=quarantine) all pass, alignment is relaxed on both. Sample size is illustrative, not statistical — the headline finding (spread between providers) matches what we see across hundreds of customer campaigns.
Aggregate Inbox rates across the three-touch sequence:
- Gmail: 68% Inbox, 24% Promotions, 8% Spam
- Yandex: 51% Inbox, 0% Promotions, 49% Spam
- Mail.ru: 38% Inbox, 0% Promotions, 62% Spam
- Rambler: 44% Inbox, 0% Promotions, 56% Spam
In Western markets the spread between best and worst ISP on a warm setup is typically 15–20 points. Here it was 30 points. That is the gap that kills Russia-targeted cold sequences run from a Western playbook.
The single biggest delta-driver
Across individual debug runs, one variable moved the spread more than any other: DKIM selector strictness. Specifically two things. First, DKIM key length — moving from 1024 to 2048 bits lifted Mail.ru inbox by 14 points in side-by-side tests on otherwise identical sending setups. Second, DKIM-From alignment — a subdomain-signed message (d=mail.example.com) against a root-From (from: hi@example.com) aligns under relaxed DMARC but gets a lower ML score at Mail.ru than a strictly- aligned signature.
Gmail does not care about this distinction. Yandex cares moderately. Mail.ru cares a lot. If you use a typical cold-email tool that signs with the tool's own subdomain (e.g. lemlist.co, instantly.ai), you are running with weak alignment at exactly the provider that punishes it hardest.
1. Upgrade DKIM to 2048 bits. 2. Use strict DMARC alignment (adkim=s) with same-domain DKIM signing. 3. Send from a warmed root domain, not a newly-registered mail subdomain. Expect 15–25 points of Mail.ru inbox gain from these three changes alone.
Why Mail.ru spam-buckets sequences harder
Mail.ru's ML classifier is trained against a cold-email attack landscape that is heavier than anything Gmail sees in English. The pattern of "template, personalised with name + company, three follow-ups on a 3-day cadence, tracked link to a pricing page" is exactly what Russian-language cold-email spam looks like, and the classifier has learned it.
What triggers the pattern-spam classifier
- Three or more messages from the same sender to the same recipient inside 10 days, all with no reply.
- Subjects that follow the "Re: <name>" or "quick question <name>" mould.
- Tracked links where the tracked domain is unfamiliar to Mail.ru (less than 30 days on the network).
- A high ratio of sends-per-reply across recent history (engagement signal is central to Mail.ru's scoring).
The counter-play is engagement. Mail.ru lifts senders with a reply-per-send ratio above ~5% meaningfully. Below 1% the classifier treats you as a pattern-spammer regardless of content. This is why a Russia-targeted cold sequence needs tight list targeting — the reply rate is the deliverability tool.
Cyrillic encoding gotchas
Even when the body is English, the From-name or subject line in a Russia-targeted campaign often contains Cyrillic. Three bugs we see constantly in debug runs:
- From-name in Windows-1251 instead of UTF-8. Older sending tools still default to 1251 for Cyrillic. Mail.ru treats this as suspicious. Force UTF-8 with base64 encoding:
=?UTF-8?B?...?=. - Mixed-script From-name. "Ivan Ivanov — Альфа Sales" is parsed as a potential homograph attack. Keep From-names in a single script.
- Subject with a single Cyrillic word in an otherwise ASCII line. Split the subject encoding correctly. Some tools mangle this, producing a visually-correct subject with corrupt MIME headers that Mail.ru's parser flags.
Warm-up strategy for CIS inboxes specifically
Western warm-up tools (Lemwarm, Warmup Inbox, Mailwarm) use pools of Gmail and Outlook inboxes. For Russia-targeted sending that is half-useless — the reputation you build is with Gmail, and it does not transfer to Mail.ru or Yandex. The practical workaround:
- First 2 weeks: warm against a Western pool (any standard tool). Builds baseline domain reputation.
- Weeks 2–4: manually send real correspondence to Mail.ru, Yandex and Rambler contacts — ideally ones who will reply. A team of five humans exchanging real mail with Russian-side contacts beats any automated tool.
- Week 4+: start real cold-email volume, but gated by engagement. If reply-per-send on a segment falls below 2%, pause the segment for a week. Mail.ru will tighten fast if you push through low engagement.
Run a free seed test that covers Gmail, Mail.ru, Yandex, Rambler and List.ru in one shot. Per-provider Inbox / Spam / Missing verdicts, full headers, Rspamd scores, and SPF/DKIM/DMARC parsing. No signup.
What this means for a Russia-targeted sequence
The short version of the playbook if you are cold-emailing Russian companies and want Mail.ru/Yandex inbox placement:
- Domain: root-domain sending, 2048-bit DKIM, strict DMARC alignment. Six weeks of mixed warm-up including real Russian-side correspondence.
- Content: plain-text or text-dominant HTML. No images in the first touch. Single tracked link maximum. Subject in sentence case, no Cyrillic unless the body is also Cyrillic.
- Cadence: maximum three touches, spaced 5+ days apart. Kill the sequence on any reply, including negative ones.
- Targeting: tighter than you think. Reply rate is the deliverability mechanism. Below 3% reply-per-send on any segment, Mail.ru will bucket you into pattern-spam territory.
- Measurement: seed test every campaign, because Gmail data is not a proxy. Track per-provider Inbox rates as separate KPIs.