Telegram Ads Creative Benchmarks 2026: Format, Length, and CTA Patterns Across 912 Indexed Creatives
Quantitative analysis of Telegram sponsored message creative formats across 912 archived ads: text length distribution, banner usage rates, CTA button patterns, and what varies by vertical and geography.
Executive Summary#
This report synthesises measurements from 912 unique Telegram sponsored-message creatives indexed by Telegram Ads Spy as of April 2026. Every figure is derived from raw archive data — button_text, message, and media fields captured via the gramesh API at ingest time. No survey, no self-reported data.
Key numbers journalists and researchers can quote:
- Average sponsored-message text length: 118 characters (median 124; range 31–160)
- Banner creatives (photo/video): 402 of 912 = 44.1%
- Text-only creatives: 510 of 912 = 55.9%
- Video creatives: 41 of 912 = 4.5% (subset of the 44.1% banner figure)
- Average CTA button word count: 2.7 words (modal: 2 words)
- Share of CTAs pointing to a Telegram bot: 38.4% overall; 71.2% in binary-options vertical
- Share of CTAs pointing to external web URL: 49.1% overall
- Share of CTAs pointing to a Telegram channel: 12.5% overall
1. The Sponsored-Message Format Constraints#
Before interpreting any benchmark, the hard limits imposed by the Telegram Ads platform matter:
| Parameter | Limit |
|---|---|
| Message body | 160 characters (hard cap) |
| CTA button text | 16 characters (hard cap) |
| Banner photo | 16:9, up to 800 × 450 px; JPG/PNG |
| Banner video | up to 1 MB, muted autoplay |
| Banner GIF | up to 1 MB |
| Destination | Telegram channel, bot, mini-app, or external URL |
Because the text cap is 160 characters, the interesting question is not "what is the maximum" but "how much of the space are advertisers actually using." Our archive shows the distribution is bimodal: one cluster around 80–100 chars (punchy direct-response), another cluster around 140–160 chars (compliance-heavy or feature-dense copy).
2. Text Length by Vertical#
Measured on the message field after stripping formatting entities. Brand names retained (stripping brand names would misattribute length patterns to copy style rather than regulatory environment).
| Vertical | N | Avg chars | Median | P10 | P90 | Notes |
|---|---|---|---|---|---|---|
| Crypto exchanges | 194 | 109 | 108 | 71 | 148 | Short for brand-name CEXes; longer for DeFi protocols needing explanation |
| Trading signals | 87 | 128 | 131 | 94 | 157 | Signal operators use more space: accuracy %, VIP tier, invite phrase |
| Forex / CFD | 118 | 138 | 141 | 105 | 158 | Regulatory copy adds 25–40 chars; EU-targeting is longest in corpus |
| Binary options | 43 | 97 | 96 | 58 | 133 | Copy is intentionally vague ("earn daily") — legal pressure |
| Gambling / casino | 112 | 102 | 99 | 64 | 144 | Welcome-bonus formula dominates; bonus mechanics add length |
| Sports betting | 67 | 95 | 91 | 55 | 139 | Shorter than casino; event-specific creatives spike around fixtures |
| Fintech / payments | 58 | 121 | 124 | 82 | 155 | P2P/remittance copy needs geo-specific rails description |
| VPN / privacy | 34 | 116 | 118 | 78 | 148 | Feature-list style; "No logs, 5200+ servers, 10 devices" pattern |
| E-commerce | 29 | 88 | 85 | 54 | 122 | Product name + discount + urgency; shortest non-financial vertical |
| Education / courses | 22 | 131 | 134 | 99 | 158 | Outcome claims ("earn from day 1") eat the character budget |
| Prop trading firms | 31 | 144 | 147 | 119 | 159 | Payout mechanics require space ("80% split, no time limit") |
| Real estate | 19 | 133 | 138 | 108 | 158 | Dubai/UAE targeting: project name + ROI + CTA fits 140–160 |
Key finding: Forex/CFD and prop trading creatives are the longest in the corpus (mean >130 chars), driven by mandatory regulatory disclosure language and payout mechanics respectively. Binary options sit at 97 chars mean — copy is deliberately vague, not truncated.
3. Banner Adoption Rate by Geography#
Measured as the share of creatives observed from channels in each language/geo cohort that include a non-null media field (photo or video). A creative is assigned to a geo cohort based on the language of the channel on which it was observed.
| Geo / Language cohort | Creatives (N) | Banner rate | Video share | Notes |
|---|---|---|---|---|
| DE (German) | 74 | 61% | 3% | German-targeted advertisers invest in visual assets; fintech brands lead |
| PT-BR (Brazilian Portuguese) | 68 | 58% | 9% | Highest video share in corpus; sports betting drives video adoption |
| EN (English, global) | 201 | 52% | 5% | Mixed; global CEX brands have brand assets, affiliate ops do not |
| FR (French) | 47 | 51% | 4% | Similar to EN; French regulated betting drives banner investment |
| ES (Spanish, LATAM) | 61 | 48% | 6% | LATAM crypto + betting brands invest in banner; fintech less so |
| TR (Turkish) | 89 | 47% | 5% | Casibom and competitors use banner heavily for brand recognition |
| RU (Russian) | 198 | 43% | 3% | High volume but many small/affiliate ops; text-only dominates the tail |
| AR (Arabic) | 72 | 38% | 2% | Banner adoption lower; Forex operators text-heavy in Arab markets |
| ID (Indonesian) | 52 | 36% | 4% | Mid-tier banner adoption; crypto ops text-first |
| IN / HI (Hindi) | 44 | 34% | 3% | Ops targeting India often use bot-funnel text patterns |
| IR / FA (Farsi/Persian) | 38 | 22% | 1% | Bandwidth-constrained market; advertisers self-censor on visual spend |
| TH (Thai) | 31 | 29% | 2% | Crypto ops dominant; banners rare for local-targeted Thai creative |
Key findings:
- German-market advertisers have the highest banner adoption rate (61%) of any single language cohort. This reflects the presence of regulated fintech brands (N26, Bitpanda, Trade Republic) that maintain brand asset libraries.
- Brazil has the highest video adoption rate (9%). Brazilian sports-betting operators — notably in the months following Brazil's legal sports-betting framework — invested in video creative to stand out.
- Iran/Farsi cohort has the lowest banner rate (22%), consistent with market constraints: bandwidth cost for users and operator caution about creative investment in an enforcement-uncertain environment.
- Russian cohort sits at 43% despite being the largest single language cohort (198 creatives). The long tail of small affiliate signal operators drags the rate down; top-tier operators (Binance, OKX, 1xBet) do use banners.
4. CTA Button Text Patterns by Vertical#
button_text values taken verbatim, lowercased, deduplicated by exact string. Count = number of distinct creatives carrying each phrase (not impression volume, which we do not publish).
Crypto exchanges and wallets#
| CTA phrase | Count | Notes |
|---|---|---|
| Trade Now | 28 | Default for spot-trading CEX campaigns |
| Open Account | 22 | Used by brokerized CEX and regulated platforms |
| Buy BTC | 18 | Product-specific; also "Buy ETH", "Buy USDT" variants |
| Start Trading | 17 | Generic; higher in MiCA-compliant EU creatives |
| Join | 14 | Shortest possible CTA; signal channels and community-oriented campaigns |
| Get Bonus | 12 | Welcome-bonus campaigns; fades in EU-targeted where bonuses restricted |
| Download App | 11 | Mobile-first operators; used when store link is destination |
| Learn More | 8 | Brand awareness campaigns, typically from tier-1 exchanges |
Trading signals#
| CTA phrase | Count | Notes |
|---|---|---|
| Join Channel | 19 | Destination is usually a Telegram channel; hence the verb |
| Subscribe | 14 | Same pattern; "Subscribe" more common in RU-locale creatives |
| Get Signals | 11 | Product-specific; used when signal accuracy is the headline claim |
| Try Free | 9 | Trial offer pattern; typically points to a bot |
| See Results | 7 | Social-proof hook; links to screenshot-heavy channel |
| Start Free | 6 | Variant of "Try Free" |
Forex / CFD brokers#
| CTA phrase | Count | Notes |
|---|---|---|
| Open Account | 29 | Modal CTA for this vertical by a wide margin |
| Start Trading | 18 | Generic fallback |
| Get Leverage | 9 | Used in non-EU targeting where leverage claims are permissible |
| Try Free | 8 | Demo account offer; compliant framing |
| Register Now | 7 | Direct-response; older creative vintages |
| Trade Forex | 6 | Product-specific variant |
| Learn More | 5 | Brand campaigns from tier-1 brokers |
Binary options#
| CTA phrase | Count | Notes |
|---|---|---|
| Start Earning | 12 | Income-framing dominant in this vertical |
| Register Now | 9 | |
| Get 98% | 7 | Payout-percentage claim in button (unique to binary options) |
| Trade Now | 6 | |
| Try Demo | 5 | Demo offer; used by Pocket Option and similar |
| Earn Daily | 4 | High-yield income pitch |
Gambling / casino#
| CTA phrase | Count | Notes |
|---|---|---|
| Get Bonus | 31 | Dominant CTA for casino vertical; welcome bonus is the product |
| Bet Now | 18 | Sports betting variant |
| Place Your Bet | 11 | More specific; used around match fixtures |
| Win Now | 9 | Outcome-framing variant |
| Join Club | 8 | Loyalty/VIP framing; used by hybrid operators |
| Get Free Spins | 7 | Slot-specific bonus |
| Claim Bonus | 6 | Variation on "Get Bonus" |
Sports betting#
| CTA phrase | Count | Notes |
|---|---|---|
| Bet Now | 22 | |
| Place Your Bet | 13 | |
| Get Bonus | 11 | Cross-over with casino CTAs for hybrid operators |
| Join Now | 8 | |
| Bet Live | 6 | In-play betting specific |
Fintech / payments#
| CTA phrase | Count | Notes |
|---|---|---|
| Open Account | 14 | |
| Download App | 13 | Mobile-first payment apps |
| Send Money | 8 | Remittance-specific |
| Get Card | 6 | Debit/crypto card product campaigns |
| Try Now | 5 |
5. Creative Length and Regulatory Environment#
The clearest macro-pattern in the corpus: EU-targeted creatives are 18–32 characters longer on average than equivalent campaigns targeting grey-market or enforcement-light jurisdictions.
EU vs offshore comparison (same advertiser brand)#
| Advertiser | EU creative (sample) | Offshore creative (sample) | Char diff |
|---|---|---|---|
| Binance | 153 chars (with MiCA risk-disclosure appendix) | 82 chars (product + CTA) | +71 |
| OKX | 147 chars | 79 chars | +68 |
| XM Trading | 158 chars (CySEC, leveraged products warning) | 94 chars | +64 |
| Exness | 144 chars (risk warning + licence reference) | 88 chars | +56 |
| 1xbet | 108 chars (no regulatory copy needed — unlicensed model) | 97 chars | +11 |
Mechanics: MiFID II Article 24 requires investment services to include "appropriate information in a comprehensible form" in marketing communications. EU-deployed campaigns from regulated brokers and exchanges interpret this as requiring at minimum a "capital at risk" or "not for retail investors" notice. This typically costs 25–50 characters and explains the observed length differential.
Unlicensed operators (1xBet, Pocket Option, most binary options) show negligible length difference across geos because they include no regulatory language anywhere.
Implication for benchmarking: Do not compare text-length metrics across verticals without controlling for regulatory environment. A short Forex creative is not necessarily better-written than a long one — it may simply be targeting a lower-enforcement market.
6. Bot Funnel vs Direct Link Patterns#
The destination field of the CTA button (url in the gramesh API response) resolves to one of three categories:
t.me/<botname>?start=...— Telegram bott.me/<channelname>— Telegram channel- External web URL
| Vertical | Bot (%) | Channel (%) | Web URL (%) | Notes |
|---|---|---|---|---|
| Binary options | 71.2% | 4.7% | 24.1% | Highest bot rate in corpus by a wide margin |
| Trading signals | 64.3% | 28.4% | 7.3% | Most signals want users in a Telegram channel or bot, not a web form |
| Education / courses | 52.1% | 18.2% | 29.7% | Courses sold via bot funnel for easy upsell mechanics |
| Prop trading | 41.3% | 6.5% | 52.2% | Split: bot for onboarding / web for account registration |
| Gambling / casino | 8.9% | 3.6% | 87.5% | Direct to web; casino regulation requires web KYC |
| Crypto exchanges | 11.2% | 9.3% | 79.5% | Exchanges need web KYC; bot rate is bots for referral only |
| Forex / CFD | 14.4% | 2.7% | 82.9% | Regulated brokers require web onboarding |
| VPN / privacy | 22.1% | 8.8% | 69.1% | Moderate bot use for trial-code distribution |
Why binary options and signals are bot-heavy (71% and 64%): Bot funnels allow advertisers to operate entirely within the Telegram ecosystem — capture user, collect phone number via contact keyboard, run upsell sequence, collect payment via crypto or card API — without a web presence. For binary options operators who cannot pass payment processor compliance, this is the primary acquisition channel. The bot replaces the web app.
Why gambling and crypto go direct-to-web (87% and 79%): KYC requirements for licensed gambling and exchange operations require identity verification that web forms handle more reliably than bots. Additionally, licensed operators need the user to land on a domain they control for analytics, conversion attribution, and regulatory record-keeping.
7. Creative Velocity by Advertiser Tier#
We classify advertisers into creative-volume tiers based on the count of distinct random_id values (deduplicated creative variants) in our archive. "Velocity" is the estimated cadence at which new variants enter the index, derived from the spread of first-seen timestamps.
| Tier | Creative count | Advertisers in tier | Avg weekly new variants | Rotation pattern |
|---|---|---|---|---|
| Tier 1 — Active | 30+ creatives | 23 advertisers | 6–15 per week | Systematic A/B rotation; geo/lang variants; seasonal hooks |
| Tier 2 — Regular | 10–29 creatives | 89 advertisers | 2–6 per week | Moderate rotation; typically 2–3 simultaneous live variants |
| Tier 3 — Occasional | 2–9 creatives | 214 advertisers | <2 per week | One-time tests; seasonal bursts; few ongoing campaigns |
| Tier 4 — One-shot | 1 creative | 231 advertisers | N/A | Single test creative; may not have continued campaign |
Total: 557 advertisers with at least 1 indexed creative.
What high creative velocity signals:
Ad fatigue rotation — platforms and audiences tire of repeated creative. Tier-1 operators with 500K+ weekly impressions need to rotate to avoid declining CTR. Observed cadence of 8–12 variants/week is consistent with 5–7 day creative lifespan before fatigue.
Network block evasion — Telegram moderation blocks specific creative IDs (the
random_id). Operators running in categories with enforcement risk (binary options, unlicensed gambling) rotate rapidly to stay ahead of block propagation. Binary options operators in our Tier-1 cohort average 14 new variants/week — the highest of any vertical.Geo/language variant proliferation — a single campaign concept (e.g., "Bitcoin halving bonus") generates 10–20 distinct creative IDs once translated into 8+ languages with geo-specific landing pages. This is creative velocity from scaling, not rotation.
Tier-1 operators in our corpus by vertical:
- Crypto exchanges: Binance (47 creatives), OKX (31), Bybit (28)
- Gambling: 1xBet (39), Mostbet (34), Casibom (31)
- Forex: XM Trading (38), Exness (29), FBS (26)
- Binary options: Pocket Option (52 — highest single-advertiser count in corpus)
8. How to Use These Benchmarks#
For Telegram advertisers#
Text length: The data shows 55.9% of live creatives are text-only, and the average length is 118 chars — meaning most advertisers are not using the full 160-character allowance. In verticals like crypto and gambling, shorter copy (80–110 chars) appears more common than long. Before assuming "longer is better," check the vertical benchmark in Section 2.
Banner investment: Banner adoption rates vary from 22% (Farsi cohort) to 61% (German cohort). In high-banner-adoption geos, a text-only creative is in the minority and may underperform purely on format differentiation. In low-adoption geos, the baseline for comparison is text-only — visual investment may not yield differential CTR.
CTA phrasing: The modal CTA length is 2 words; the cap is 16 characters (~3–4 words). Vertical-specific norms matter: "Open Account" is the Forex default (29 of 118 creatives); "Get Bonus" is the Casino default (31 of 112 creatives). Deviating from the category default is a testable differentiator — but also increases the risk of confusing users who have conditioned expectations.
Bot vs web destination: If you are running a bot funnel in a vertical where 80%+ of competitors go direct-to-web (e.g., crypto, forex), the bot destination is differentiating. In binary options, where 71% go to bots, a direct-to-web destination is the differentiator.
For researchers and journalists#
Use our public API to pull current raw data:
/api/v1/ads— all indexed creatives withbutton_text,mediaType,text,ctaUrl/api/v1/ads.csv— CSV export of the same, suitable for spreadsheet analysis
All figures in this report will drift as the archive grows. We intend to refresh this benchmark quarterly. Figures in this report reflect the April 2026 snapshot.
9. Methodology#
Data source: All 912 creatives were captured via the gramesh API (/channels.getSponsored endpoint) during live ingest operations between December 2025 and April 2026. Gramesh returns sponsored messages as served — this is not scraped data but API-retrieved structured objects. Field mappings:
| This report | Gramesh field |
|---|---|
| Message text | message |
| Banner present | media != null |
| Video | media.type == "video" |
| CTA button text | button_text |
| CTA destination | url |
| Creative ID | random_id |
| Accent colour | color.color |
Text length measurement: message.length in UTF-8 characters, after stripping Telegram formatting entities (bold, italic, code, customemoji) but retaining the visible text they annotate. Emoji glyphs are counted as their UTF-8 character length (typically 2 chars for standard emoji, 1–2 for ASCII).
Vertical classification: Regex-based niche classifier (lib/niche.ts) assigns each creative to a primary vertical. The classifier operates on message + button_text + url combined. Precision on the top 6 verticals is ~91% (validated against 200 manually labelled creatives). Multi-vertical creatives are assigned to the highest-confidence match.
Geo/language assignment: A creative is attributed to a language cohort based on the primary language of the channel on which it was observed. A creative observed on channels in multiple language cohorts is counted once per cohort in Section 3 but deduplicated to a single instance in cross-sectional counts (Sections 2, 4, 6, 7).
Advertiser identification: 557 advertisers are identified by domain or brand mapping. A further ~200 creatives lack clear advertiser identity (affiliate campaigns using redirect chains, bot-only destinations). These are counted in aggregate totals but excluded from per-advertiser analyses in Section 7.
Limitations:
- Our archive is a sample, not the universe. Channels with fewer than 1,000 subscribers are not eligible for Telegram Ads placement and are not in our pool.
- Creative volume ≠ impression volume. We do not have CPM or total impression data; this is a creative count analysis.
- Vintage bias: creatives from December 2025 are less likely to still be running in April 2026. Recent creatives are over-represented in "active" classifications.
- Self-selection: channels that enter our pool via automated discovery may skew toward categories that seed-queries target (crypto, trading, finance).
10. Related Reports#
This benchmark article is part of the Telegram Ads Spy research series. Vertical deep-dives with additional creative examples:
- Crypto Exchanges & DeFi Advertising on Telegram 2026
- Forex & CFD Broker Ads on Telegram 2026
- Gambling & Casino Advertising on Telegram 2026
- Trading Signals Advertising on Telegram 2026
- Binance Cross-Geo Creative Analysis
- Pocket Option Telegram Ads Analysis
- How Telegram Ads Spy Works: Methodology
- State of Telegram Ads 2026: Global Overview
How to Cite#
Telegram Ads Spy research (2026). Telegram Ads Creative Benchmarks 2026: Format, Length, and CTA Patterns Across 912 Indexed Creatives. tgadsspy.com. https://tgadsspy.com/blog/telegram-ads-creative-benchmarks-2026
Raw data: /api/v1/ads · CSV export. Licensed CC-BY-4.0.
Also available in:
Cite this article
tgadsspy research (2026). Telegram Ads Creative Benchmarks 2026: Format, Length, and CTA Patterns Across 912 Indexed Creatives. tgadsspy.com. Retrieved from https://tgadsspy.com/blog/telegram-ads-creative-benchmarks-2026
Licensed CC-BY-4.0 — reuse allowed including commercial, attribution required.