Xreacher Logo

Original data · H2 2026 edition · a production-database study, not a survey

Telegram Cold DM Benchmarks

128,562

delivered Telegram cold DMs analyzed — replying at 13.5%.

992 sending accounts45 campaigns · 21 users180 days · Jan-Jul 202694 message templatesaggregated & anonymized

Aggregated, anonymized benchmarks from 128,562 delivered Telegram cold DMs sent through Xreacher's Telegram product over its first six months: what reply rates campaigns actually get, and — because every message's text is stored — exactly which copy earns replies. Message length, a question, and the right amount of spintax each move the number by a third or more; personalization by name does not. Recomputed each edition from the full campaign dataset with the methodology published below.

Data window: the 180 days ending July 5, 2026 · Published 2026-07-05 · Free to cite with attribution and a link

The dataset

The dataset

Every figure is computed across all Telegram campaigns sent through Xreacher in the stated window, aggregated and anonymized. One record = one initial send; a send counts as delivered when Telegram accepts it, and delivered messages are the denominator for every reply rate.

Delivered cold DMs analyzed

128,562

This edition covers 128,562 delivered Telegram cold DMs sent in the 180 days ending July 5, 2026 — the base for every reply-rate figure on this page.

Sending accounts in dataset

992

The data spans 992 distinct Telegram sending accounts across 45 campaigns and 21 users.

Distinct message templates

94

Senders used 94 distinct message templates in the window; every delivered DM is tracked to the template it came from.

Volume concentration

98.3%

A handful of scaled operations dominate raw volume: the five users running 11+ accounts sent 98.3% of all delivered DMs.

Methodology note: Because a few large senders drive most volume, this page leans on campaign-level medians and cohort splits rather than one pooled average.

Finding 01 · Reply rates

What reply rates actually look like

Telegram cold DMs reply far higher than most channels. The pooled average and the median campaign sit close together (13.5% vs 12.6%); only 18 campaigns cleared 200 delivered DMs, so the distribution is thin at the tails.

Campaign reply-rate distribution (18 campaigns with ≥200 delivered DMs)

p10bottom campaigns
1.96%
p25
5.84%
Median campaign
12.63%
Pooled averageall delivered DMs
13.52%
p75
15.91%
Top decile (p90)
25.74%

Reply rate by number of sending accounts used (per user)

2-3 accounts4 users · 215 threads
34.42%
1 account (solo)9 users · 707 threads
29.28%
11+ accounts (scale)5 users · 126k threads
13.41%
4-10 accounts4 users · 1,299 threads
12.55%

Share of total DM volume by cohort

11+ account operations — 98.3% of volume

Small senders (blue, 1.7% of volume) replied at 29-34%; the 11+ account operations that drive 98.3% of volume replied at 13.4%.

Overall reply rate

13.5%

Across all 128,562 delivered Telegram cold DMs, the overall reply rate was 13.5% (13.52%) — replies divided by delivered messages.

Median campaign reply rate

12.6%

The median Telegram campaign (across the 18 campaigns with at least 200 delivered DMs) earned a 12.6% reply rate; the top 10% of campaigns exceeded 25.7%.

Methodology note: Campaign-level median across the 18 campaigns with ≥200 delivered DMs. Only a handful clear the volume bar, so the tails (p10/p90) rest on a few campaigns each.

Solo and small senders (1-3 accounts)

29-34%

Users sending from one to three accounts saw 29-34% reply rates — far above the scaled operations that drive most volume.

Methodology note: Directional: the small-sender cohorts are only a few hundred threads each. The direction (small senders reply better) is consistent; the absolute numbers need more volume.

Scaled operations (11+ accounts)

13.4%

Operations sending from 11 or more accounts generated 98% of all volume but replied at 13.4% — roughly half the rate of small, targeted senders.

Methodology note: Cohorts are assigned by accounts actually used in the window, not by plan.

Finding 02 · Message copy

The copy that gets replies

Because every DM's rendered text is stored, the copy itself is measurable. Analyzed on a 60,000-message sample of delivered threads. Length and links move the number most; personalization and emoji barely register.

Reply rate by message length (characters)

50-100 charsthe sweet spot
15.80%
< 50 chars
13.15%
100-200 chars
12.77%
200-400 chars
7.90%

Reply rate by content feature

Asks a questionvs 12.1% without
14.04%
Name available, unusedvs 13.9% with name used
14.33%
No link
13.52%
Contains a linkdirectional · ~200 msgs
2.97%

Best message length

50-100 characters

Messages of 50-100 characters replied best at 15.8%, versus 7.9% for long 200-400 character messages — short beats long by roughly 2x.

Methodology note: Length is partly entangled with template (the long band is mostly a few hard-sell templates); the 50-100 sweet spot holds across many templates.

Asking a question

14.0%

DMs containing a question replied at 14.0% versus 12.1% without — a reliable, if modest, +16% relative lift.

First-name personalization

no measurable lift

Inserting the lead's first name showed no measurable lift: 13.9% when the name was used versus 14.3% when a name was available but left out. Emoji were a wash too (13.7% vs 13.2%).

Methodology note: Correlational, on the sample where a first name was available. A common tactic that did not, on this data, earn more replies.

Finding 03 · Spintax

Spintax has a sweet spot

Spintax (rotating message variations) is standard practice to avoid sending identical copy. Some variation dramatically beats none — but piling on hundreds of renders erodes the gain.

Reply rate by number of spintax variations in the template

11-100 variationsbest
15.59%
2-10 variations
14.36%
100+ variations
11.44%
1 · no spintaxworst
3.01%

Best spintax range

11-100 variations

Templates with 11-100 variations replied best at 15.6%; 2-10 variations reached 14.4%, but past 100 variations the rate fell back to 11.4%.

Methodology note: Variation count correlates with specific templates and senders, so read this as a strong association, not a clean dose-response.

No spintax at all

3.0%

Templates sent with no spintax variation at all replied at only 3.0% — about a fifth of the rate of well-varied copy, likely because identical repeated messages get throttled or filtered.

Reply-rate spread across openers

6% to 17%

Wording alone swung reply rate from 6% to 17% across templates — the strongest openers were short; the laggards were long, hard-sell copy.

Finding 04 · Sending accounts

Older sending accounts reply better

Joining each send to its sending account, account tenure is the clean signal: longer-lived accounts reply at a meaningfully higher rate.

Reply rate by sending-account tenure in the platform

90+ days60,544 delivered
13.49%
30-90 days18,151 delivered
11.99%
< 30 days498 delivered
10.44%

Reply rate by account tenure

13.5%

Accounts active 90+ days replied at 13.5%, versus 12.0% at 30-90 days and 10.4% under 30 days — warmed-up accounts do better.

Methodology note: 'Tenure' is time since the account was added to the platform, not the Telegram account's true age. Correlational.

Finding 05 · Read receipts

Read receipts: seen, then replied

Telegram exposes read state on outgoing messages — a signal most channels can't measure. A DM that gets opened replies at six times the rate of one that doesn't, and most opens happen fast.

Delivered → seen → replied

Delivered
128,562
Confirmed seen
24,425
Replied
17,380

Reply rate by whether the DM was opened

Seen (opened)
41.6%
Not marked seen
6.9%

Time from send to open (seen) · 24,425 confirmed opens

< 1h21%
1-6h27%
6-24h26%
1-3d16%
> 3d11%

About half of opens happen within six hours and ~74% within a day. Read detection is polled ~2-hourly, so buckets are slightly coarse.

Reply rate once the DM is seen

41.6%

Once a lead opens a cold DM they reply 41.6% of the time, versus 6.9% when the message is not marked seen — a 6x gap.

Methodology note: Partly mechanical (reading precedes replying). 'Seen' is only known for read-checked messages, so coverage is partial.

Confirmed read (seen) rate

19%

At least 19% of delivered DMs were confirmed opened; read receipts are polled, so this is a floor on the true open rate.

How fast DMs are opened

74%

Of confirmed opens, about half happen within six hours and 74% within a day.

Finding 06 · Send timing

Send timing

How send time relates to reply rate, in UTC. Unlike some channels there's no dramatic window — reply rate stays in a narrow 11-16% band all day.

Reply rate by send hour (UTC) · delivered DMs

00
03
06
09
12
15
18
21
reply rate ≥ 15% (soft peak) other hours

Best send window (UTC)

16:00-18:00 UTC

Reply rate is fairly flat by hour; the gentle high points sit around 16:00-18:00 UTC and 10:00 UTC (~16%), the lows around 07:00-08:00 and 19:00 (~11%).

Methodology note: A weak correlation across a global sender base, not a causal test — treat send-hour as a minor lever at most.

Reply rate by send hour — full table

UTCNew York (ET)Berlin (CET)DMs sentReply rate
00:0020:0002:005,30413.86%
01:0021:0003:003,81413.90%
02:0022:0004:003,52815.53%
03:0023:0005:002,90713.07%
04:0000:0006:002,63011.71%
05:0001:0007:001,99413.39%
06:0002:0008:005,52712.96%
07:0003:0009:0010,47411.52%
08:0004:0010:008,63411.13%lowest
09:0005:0011:007,56612.62%
10:0006:0012:007,15415.68%
11:0007:0013:006,74613.21%
12:0008:0014:005,87213.47%
13:0009:0015:004,79414.66%
14:0010:0016:004,35613.09%
15:0011:0017:004,39413.47%
16:0012:0018:003,99916.33%
17:0013:0019:004,71315.15%
18:0014:0020:004,97216.37%highest
19:0015:0021:005,15511.87%
20:0016:0022:005,72913.37%
21:0017:0023:004,17514.85%
22:0018:0000:006,87013.71%
23:0019:0001:007,26013.55%

Delivered DMs, 180 days ending July 5, 2026. Reply rate stays within a narrow 11-16% band across the day, so send-hour is a minor lever. Hours are UTC; a global sender base spans many local timezones.

Finding 07 · Reply latency

How fast replies come in

Time from send to reply on single-message threads. Telegram replies land fast.

Time from send to reply · 17,356 replied single-message threads

< 1h15.9%
1-6h34.6%
6-24h23.5%
1-3d10.2%
3-7d6.3%
> 7d9.5%

~74% of replies arrive within 24 hours and ~84% within three days; 9.5% arrive only after a full week.

Replies within 24 hours

74%

About 74% of replies to a Telegram cold DM arrive within 24 hours and ~84% within three days — but 9.5% arrive only after a full week, so keep threads under watch.

Methodology note: Reply detection is polled roughly every two hours, so times are upper bounds and the sub-hour bucket is noisy.

Finding 08 · Targeting

Who replies

Reply rate by the lead's stored profile. The attributes captured at scrape time are limited, but the ones available show clear patterns.

Reply rate by lead's public @username

Has a public @username
14.03%
No public username
9.84%

Leads with a public @username

14.0%

Leads with a public @username replied at 14.0% versus 9.8% for those without — a 1.4x gap.

Methodology note: Directional, sampled (~124k threads). Username presence is the only lead-profile attribute stored, so richer 'who replies' cuts aren't available this edition.

Effect of the lead's source

~4x

Which group a lead was scraped from swung reply rate by roughly ~4x between the best and worst sources — source selection matters as much as the copy.

Methodology note: Correlational. Specific source breakdowns are kept internal.

Finding 09 · Historical trend

Reply rate over time

How the monthly reply rate has moved across the six months of data.

Reply rate by month · delivered threads in parentheses

2026-019,839 threads
14.34%
2026-0227,849 threads
12.08%
2026-0336,221 threads
14.72%
2026-0419,124 threads
14.73%
2026-0517,841 threads
13.27%
2026-0615,958 threads
11.76%

Monthly reply-rate band

11.8-14.7%

The monthly reply rate has held in a 11.8-14.7% band since launch, with no sustained decline — volume peaked in March and has since eased.

Summary

Practices associated with higher reply rates

Each guideline restates a finding above, with its qualifier. Correlational findings describe associations in observational data, not causal effects.

01

Keep the message short. 50-100 character DMs replied best (15.8%); long 200-400 character messages fell to 7.9%.

15.8% vs 7.9%
02

Ask a question. DMs with a question replied at 14.0% versus 12.1% without — a small but reliable lift.

+16%
03

Don't paste a link in the first message. Messages with a link replied at 3.0% versus 13.5% without. Directional, but the drop is steep — links likely hurt delivery.

3.0% vs 13.5%
04

Use spintax, but don't overdo it. No variation replied at 3.0%; 11-100 variations peaked at 15.6%; past 100 the gain faded to 11.4%.

15.6%
05

Warm accounts up. Accounts active 90+ days replied at 13.5% versus 10.4% for accounts under 30 days.

13.5% vs 10.4%
06

Don't over-invest in name personalization. Inserting the lead's first name showed no measurable lift (13.9% vs 14.3%).

no lift
07

Benchmark against the campaign median (12.6%), not a single pooled number. Top-decile campaigns exceeded 25.7%.

>25.7%
08

Keep threads under watch for a week: 9.5% of replies arrived more than seven days after the initial DM.

9.5%

Methodology

How these figures are computed

Source

Full production dataset of Xreacher's Telegram product: 128,562 delivered cold DMs, 992 sending accounts, 45 campaigns, 21 users, 180 days ending July 5, 2026. Read-only aggregation over the campaign message store, joined to templates, accounts, and leads. Aggregated and anonymized; no individual customer, account, campaign, template, or scrape source is identifiable.

What counts as delivered / replied

Delivered = a send Telegram accepted. Replied = the tracker matched an inbound message on the thread. Reply rate = replied ÷ delivered; failed sends are excluded from the denominator.

Content & sampling

Message-content and personalization cuts run on a random 60,000-message sample of delivered threads; lead joins on a ~200,000-thread sample. Cell sizes are disclosed inline. The link arm is low-volume and marked directional.

Entanglement

Message length, spintax variation count, and template are correlated (long templates tend to be the hard-sell ones), and volume is dominated by a few large operators. Content and template findings are strong associations, not controlled tests, and are labeled directional where samples are thin.

Timing precision

Reply and read detection are polled roughly every two hours, so time-to-reply and time-to-open are reported in buckets as upper bounds; the sub-hour buckets are noisy.

What is not published

Absolute delivery/failure rates (most failure records are uncategorized), specific template text or scrape-source handles, and any per-customer detail. Follow-ups are excluded this edition — the feature launched recently and appears on under 0.2% of threads. Every figure is recomputed from scratch each edition.

FAQs

Where does this data come from?

From aggregated, anonymized campaign data across all Telegram outreach sent through Xreacher's Telegram product in the stated window — 128,562 delivered cold DMs from 992 sending accounts. Figures are computed with database-level aggregations over the full dataset (large random samples for message-text and lead joins), and no individual customer, account, campaign, template, or scrape source is identifiable.

What counts as a reply?

A thread counts as replied when the contacted lead sends any response after the initial DM. Reply rate is replies divided by delivered messages (sends Telegram accepted); failed sends are excluded from the denominator. Reply detection is polled roughly every two hours, so reply timing is reported in buckets, not to the minute.

Does personalizing with the lead's name help?

Not measurably, in this data. DMs that inserted the lead's first name replied at 13.9% versus 14.3% for messages that had a name available but didn't use it. What moved reply rate far more was message length, whether the message asked a question, whether it contained a link, and how much spintax variation the template used.

Why report a median campaign rate instead of one overall average?

Because volume is concentrated: five users running 11+ accounts sent 98% of all DMs. A single pooled number is dominated by those operations, so we publish the campaign-level median (12.6%) and cohort splits alongside the pooled rate (13.5%).

Can I cite these benchmarks?

Yes. Every stat has an anchor link you can reference directly, and you may quote any figure with attribution to 'Xreacher Telegram Cold DM Benchmarks' and a link to this page. The edition label states the data window so citations stay accurate as the page is refreshed.

How often is this updated?

Each edition is recomputed from the full dataset and the edition label and publication date are updated accordingly. This is an early dataset — the Telegram product is roughly six months old — so smaller cohorts are labeled directional and will firm up as volume grows.

Last updated 2026-07-05 · Cite as “Xreacher Telegram Cold DM Benchmarks (H2 2026)” with a link to this page; every stat above has its own anchor.

Related Xreacher pages