Author: Performance Lead креативов TG Target · планирование объёма креативов для Telegram Ads · фокус: математика бюджет→подход и последовательность волн (EU/CIS).
How many creatives to test in Telegram Ads — depends on the pilot media budget, the rarity of the main KPI and the wave structure. In TG Target (EU/CIS) for €8-15k media: 3-4 families of approaches in wave 1, then 2-3 hook/CTA variations within the winner - a total of 5-7 unique creatives per cycle. For €25–40k, 4–5 approaches in parallel + deeper wave 2 are possible. 10+ unrelated ads for €10k is statistical noise, not an “aggressive test”. The decision to scale is based on CPA winner, not based on the number of creatives uploaded.
The main thing from the article
- Count approaches from budget: ~€2.5–4k media minimum per approach (EU/CIS rule).
- Wave 1: 3–4 families; wave 2: 2–3 options only for the winner.
- Approach ≠ editing the hook - do not mix in one counter.
- ~3k–5k impressions per approach before disconnecting via CTR (directional).
- All approaches lose → pack/funnel audit, not “+6 creatives”.
What is considered “one creative” in Telegram Ads
Direct answer: a sponsored post with a unique combination of hook + body text + CTA (and button text if different), launched as a separate advertising entity in the account.
Entities: advertising approach, hook option, sponsored message, wave, pilot budget, minimum impressions, main KPI. Text rules - Creative best practices; cell insulation - A/B testing framework.
Formula: budget → max. parallel approaches
Direct answer: max. approaches ≈ floor(pilot media budget ÷ min. spending per approach), where min. approach costs - €2.5–4k (EU/CIS, mixed inventory, August 2026, planning range TG Target).
| Pilot media budget (EU/CIS) | Wave 1 sets (parallel) | Wave 2 options | Total unique creatives (typical) |
|---|
| €6–8k | 2–3 | 2 for the winner | 4–5 |
| €8–15k | 3–4 | 2–3 | 5–7 |
| €15–25k | 4–5 | 3 | 7–8 |
| €25–40k | 4–5 + control group | 3–4 | 8–10 |
| €40k+ / month | 5–6 structured cells | 4+ in the update pool | 10–14 scheduled |
Reserve ~30% of pilot media for wave 2 - don't spend 100% on wave 1 distribution. TCL context - Prices.
Wave structure: how to distribute the number
Consistency of the process - not duplicating numbers from the budget table.
Wave 1 - Finding an Approach (Days 0–10)
3–4 different families of approaches; equal division of the budget; the same pack; reading CPA + CTR for days 7–10.
Wave 2 – Hook/CTA Refinement (Days 11–21)
Disable 1-2 worst approaches; 2–3 options only for the leader; do not introduce a new family unless all have passed the audit criteria.
Wave 3 - scale + fatigue (day 22+)
1–2 update options every 10–14 days if CTR falls; not a full repeated acceleration of 8 approaches.
3–4 approaches→Winner selection→2–3 hook options→Scale + periodic update
Why not “the more the merrier”
Прямой ответ: каждый лишний подход без дополнительного бюджета размывает показы и задерживает сигнал CPA — media buyer оптимизирует активность, а не обучение.
- Moderation load: 12 loads → queue delays → staggered test windows.
- Auction training: little data on the option → false winners by CTR.
- Operating costs: Naming/tracking breaks → immeasurable divisions.
Unique block: minimum impressions and events per creative
| Solution type | Min. signal (approach, EU/CIS) | Don't decide when |
|---|
| Directional shutdown via CTR | 3k–5k impressions | <1.5k views |
| Soft reading CPA (lead/KYC) | 15–25 events per portfolio | <8 events per approach |
| Read CPA (FTD/deposit) | 25–40 events per portfolio; the share of the approach is proportional | <12 FTD total |
| Declaration of scaling | CPA winner stable for 7 days. + min. n | One good day |
Benchmark context - Conversion benchmarks.
Mini-case: 12 creatives vs 4 for €11k (fintech KZ)
Same pack · target CPA KYC-start €24 · 14 days.
| Approach | Active ads | Avg. impressions/ad | Winner announced? | Final CPA |
|---|
| “Spraying” 12 approaches | 12 | ~1,1k | No (noise) | €28 mixed (no scale) |
| TG Target, wave of 4 sets | 4 → 2 → 1 | ~4.2k, then focus | Yes, day 12 | €19 winner (scaled) |
The same general media budget - structure won over volume. Related - creative hook cases.
Practical solutions: volume trade-offs
Decision rules are not a repetition of the wave process.
Budget €8k: 2 vs 4 approaches
4 approaches, if pre-moderation is completed and the lead event is frequent; 2, unless FTD and slow conversion - priority is given to depth over width.
Parallel languages
Two languages = two parallel stacks, not doubling approaches in one cell. Each language stack gets its own limit of 3 approaches.
Moderation Deviation Buffer
Load 5, expect 3-4 active - plan your invoice based on approved ones, not drafts. Rejected creatives are untested creatives.
When to increase the count to 5–6
Only with a €25k+ pilot AND a stable pack AND documented operations for equal divisions - otherwise false accuracy.
Comparison: creative volume culture by channel
| Channel | Typical test volume | Note on Telegram |
|---|
| Telegram Ads | 3–7 per pilot cycle | Text format; moderation + pack compliance limit the account |
| Meta Ads | 10–50+ options | Creative volume engine |
| Google Search | RSA combine elements | Other assembly model |
Telegram vs Google vs Meta.
Metrics: when the number of creatives is “enough”
| Signal | There are enough creatives | Too much |
|---|
| CPA winner ≤ 7 day goal. | Yes - stop adding approaches | — |
| No approach beats control | Fix the pack/funnel first | Adding 5 more |
| Each ad <2k impressions on day 10 | — | Reduce your parallel account |
| Fatigue CTR winner | +1–2 upgrade options | +8 new families |
Common mistakes
- 10+ approaches for €10k “because Meta does 50.”
- Hook edits are considered a new wave of approaches.
- There is no budget reserve for wave 2.
- Top 2 scaling by CTR without checking CPA.
- Ignore moderation deviations in account planning.
- Daily updates instead of structured waves.
- Multiple languages in one cell approach.
Checklist: number of creatives before launch
- The pilot media budget and TCL are documented.
- max. approaches = floor(media ÷ €2.5–4k) calculated.
- Wave count 1 ≤ max.; wave 2 reserve - 30%.
- Each approach = a separate family (with a name).
- Pre-moderation queue with buffer (+1).
- The minimum number of impressions per approach is fixed.
- Min. n of the main KPI is aligned with the A/B framework.
- Shutdown rule: everyone loses → audit, not +creatives.
FAQ
How many creatives should be tested in Telegram Ads?
On a pilot media budget of €8-15k in EU/CIS, TG Target recommends 3-4 families of approaches in wave 1 rather than 10-15 unrelated ads. After the winning approach, 2-3 hook/CTA options in the second wave. Total 4–7 unique creatives for a full training cycle with a typical budget. More is statistical noise; fewer than two approaches are weak hypothesis testing. The number depends on the budget, the rarity of the main KPI and the percentage of moderation completion.
How to calculate the number of creatives from the budget?
Planning formula: max. active approaches ≈ floor(pilot media budget ÷ (min. spending per approach)). Min. cost per approach in TG Target - €2.5–4k media or 3k–5k impressions × expected range CPM before first reading CPA. Example: €12k pilot ÷ €3k ≈ 4 approaches max. in a parallel wave. Reserve 30% of budget for wave 2 options within the winner. Check with TCL from the price guide.
How does the family of approaches differ from the hook variant?
Family of approaches - different value proposition (benefit vs context vs problem→solution): testing 3-4 in wave 1. Hook option - paraphrase of hook/CTA within one approach: 2-3 pieces in wave 2 after the winner. Mixing 8 hook options without a chosen approach is a waste. The taxonomy of hooks is in the guide to best creative practices.
How many impressions do you need per creative?
Directed reading CTR - from ~3,000–5,000 impressions per approach in a comparable pack. Solution for CPA - min. 25–40 major portfolio events; One creative is rarely given a separate threshold CPA, unless there is an equal division of the budget. Don't disable creative at 800 impressions unless there is a disaster CPA with clear tracking. Fatigue refresh is a new option, not +10 micro-edits.
Is it possible to test 10+ creatives at once?
Technically it is possible, but with a media budget of <€25k this is almost always insufficient sampling power: everyone receives little spending, the moderation queue grows, and the buyer loses causality. TG Target sees a pattern: 10+ approaches for €10k → “no winner”, in fact n is too small. The exception is large managed accounts €40k+/month. with a dedicated creative team.
How many creatives are there for iGaming vs fintech?
iGaming/crypto (FTD rare): fewer parallel approaches (3), longer window, more weight for moderation pre-check. Fintech lead/KYC: 4 approaches are possible for the same €12k due to a more frequent event. The limitations of regulated text reduce viable options—don't compensate with the number of prohibited approaches. Vertical benchmarks - in the guide to conversion benchmarks.
When to add new creatives to a running campaign?
Add wave 2 options when: CPA winner ≤ 7+ day goal; or CTR −20% with stable CPM (fatigue). Don’t add 5 new ones every 48 hours “for the test” - you break the training. Update cadence TG Target: 10–14 days or a surge in moderation deviations. New approaches = new brief, not replacement of emoji.
How are the number of creatives and A/B testing related?
Each set in wave 1 = A/B cell at equal division vs control pack (see A/B framework). The number of creatives should correspond to the design of the cells: one lever per test window. 4 sets in parallel = division into 4 parts, not 4 sets + changing the pack in the same week. The framework defines isolation; this article is how many cells the budget will accommodate.
What to do if all 4 creatives fail?
If all approaches CPA > goals with a sufficient sample size - the problem is probably in the package, funnel or offer, and not “6 more creatives are needed.” Step back: channel and landing page compliance audit (see campaign audit). Then a new wave of approaches after fixing the pack. Adding creatives without fixing the funnel and without changing the offer is a typical media buying mistake.
What are the errors in the number of creatives being tested?
Too many approaches for a small budget; hook options to approach winner; accounting for moderation deviations as “tested”; no wave structure; the same attitude towards text editing and a new approach; ignore minimum impressions; scaling all 4 because “one has the best CTR”; there is no pre-moderation queue. Another mistake - 20 creatives without naming and tracking discipline.
The tables of the number of creatives, the formula and the mini-case are relevant for August 2026 (planning TG Target EU/CIS). Adjust € thresholds for a different vertical/event frequency; keep the 1/2 wave structure.