Skip to content
Primores AI wiki

Alternative Advertising Channels — Landscape & Ranked Picks Beyond the Meta/Google Duopoly

Alternative Advertising Channels

TL;DR: There is a broad landscape of credible self-serve paid channels beyond Meta/Google/TikTok, and published entry floors mislead in both directions. Telegram’s headline “€2M to advertise” overstates the barrier — a TON-denominated self-serve path now starts near $50. Reddit’s “$5/day minimum” understates the barrier — the real gate is an advertiser approval process that closes self-serve for whole verticals. For a DTC/ecom brand diversifying off Meta, the frictionless first move is Microsoft/Bing (no approval wall, cheap search intent); Reddit has the best raw economics if your category clears policy; Pinterest/Snapchat suit visual DTC and Amazon captures demand. Telegram Ads is a specialist/arbitrage play — genuinely cheap attention, but contextual (channel-level) targeting and access friction keep it out of the default DTC starter kit. Newest layer (July 2026): check attribution transparency before believing any channel’s ROAS — Amazon DSP is opaque (blends modeled + measured into one number), Microsoft folds view-through into the headline by default; separately, Netflix and Amazon Streaming TV opened to low/no-minimum self-serve, and VK Ads is ruled out for EU advertisers on July-2026 sanctions.

This page is the paid-channel catalog and ranking. For the underlying framework — why channels sit on an intent-vs-cost map and why owned-channel diversification is the structural answer — see marketing/channel-economics. For Telegram as an organic/community surface and its geographic channel-fit, see marketing/telegram-marketing-channel.

The load-bearing insight: published floors lie in both directions

The single most useful correction this research surfaced is that the advertised barrier to entry is a poor guide to the real one — and it errs in opposite directions on the two channels people ask about most:

ChannelWhat gets repeatedWhat’s actually trueDirection of error
Telegram Ads”€2 million to advertise”Legacy EUR direct account ~€2M, but resellers start at €1,000–5,000 and a TON self-serve path starts ~$50–60Published barrier overstates — a $50 path exists but is barely known
Reddit Ads”$5/day minimum, fully self-serve”$5/day floor is real, but account + ad approval gates it, and restricted verticals must work directly with a Reddit sales rep or are bannedPublished floor understates — the approval wall blocks many
X (Twitter) Ads”$0 minimum — no spend floor”True that there’s no spend minimum, but you must hold a paid verification subscription (X Premium ~$8/mo, or Verified Orgs $200–1,000+/mo) to be eligible to advertisePublished floor understates — the real entry cost is a pay-to-access subscription, not spend

All three errors point the same way for a practitioner: verify the real access path before assuming a channel is open or closed to you. The cheap number and the scary number are equally likely to be wrong — and the “floor” is sometimes a subscription, an approval queue, or a €2M myth rather than a spend number at all.

Landscape taxonomy

Verified anchors carry their verification verdict. Channels where no claim survived adversarial verification are marked [unverified — open] rather than asserted, per this wiki’s evidence discipline.

ChannelAccess modelTrue entry floorTargetingBenchmark (2025–26)Best-fit
Reddit AdsSelf-serve dashboard, gated by approval$5/day per campaign ($25 lifetime) — behind account approvalSubreddit / interest / keyword$2–6 CPM; CPC unverified (see benchmarks)DTC with a research-driven audience, if category clears policy
Microsoft Advertising (Bing + Audience Network)Self-serve, no approval wallNo hard minimum (~$20–50/day practical)Search keyword + LinkedIn profile (company/role/industry; +job seniority Jun 2026)CPC ~$1.54, 33–42% below Google (3-0)Demand capture; older/wealthier base (70% aged 35–65; 38% $100K+ HHI)
Amazon Ads / retail media3 models: self-serve DSP / managed (~$50K/mo) / APISelf-serve DSP minimum dropped (unBoxed 2025)Shopper / product / audienceSignal needs ~$10–15K/mo (directional, 1-2)Demand capture for physical goods
Podcast advertisingNetworks + programmatic/DAIProgrammatic $5–15 CPM (cheapest); flat-rate for small showsShow / genre / (limited) audienceHost-read mid-roll $25–40; pre-roll $15–25; post-roll $10–20 (3-0)Trust-transfer, considered purchases
Quora AdsSelf-serve, no sales gate (3-0)$5/day (3-0; “no minimum” refuted 0-3)Question / topic / keyword intent[CPM/CPC — open]High-intent research queries
Snapchat AdsSelf-serve$5/day floor ($20–50/day rec.; new accounts ~$50/day ramp cap) (3-0, primary)Interest / behavioral / lookalike~$8.60 CPM (peaks $12.84 Q4); ~$0.51 CPCYounger visual DTC
Pinterest AdsSelf-serve, no verification/approval gate (3-0, primary)No enforced hard minimum (3-0, primary; aggregator-only ~$15/day new-account beta)Interest / keyword-search / audience / Actalike / demo / geo (3-0)~$3–4.67 CPM; $0.50–0.70 ecom CPCVisual/aspirational DTC; catalog/shopping ads for ecom
X (Twitter) AdsSelf-serve (ads.x.com) but pay-to-access: requires paid verification — X Premium ~$8/mo / Verified Orgs $200–1,000+/mo (3-0, primary)No spend minimum ($0 floor) (3-0; “$0.01/day” refuted) — the access cost is the subscriptionInterest / keyword / follower-lookalike / DPA retargeting (taxonomy partly open)[CPM/CPC — open]Real-time / topical / tech — weigh brand-safety (4% of marketers rate it brand-safe vs 39% Google)
Telegram Ads (official)Bifurcated: legacy EUR direct / resellers / TON self-serve~$50–60 (TON self-serve) → €2M (legacy direct)Contextual: channel / topic / language / geo — no interest/behavioralMin CPM 0.1 TON (~$0.31–0.50) (3-0)Audiences that cluster in known public channels; CIS/EU proximity
CTV — Roku / Hulu (self-serve)Self-serve managers (Roku Sept 2024; Hulu → Disney Campaign Manager)$500 hard floor, no revenue requirement (3-0, primary)Household / content / demo$15–40 CPM std / $35–65 premium; video-only 15/30sBrand + reach — now genuinely low-floor self-serve
CTV — NetflixSelf-serve within The Trade Desk (S&P 500+ marketplace)No minimum spend (3-0) — but needs a TTD/Kokai seat; not open signupHousehold / content (via TTD)[CPM — open]Reach at scale for existing programmatic buyers
CTV — Amazon Streaming TV (Prime Video / Twitch / Fire TV)Self-service, any Brand Registry seller$0 self-service min (2-1); DSP tiers separate (~$10K rec / $50K managed)Household / content / Amazon shopper signals[CPM — open]Physical-goods DTC already on Amazon
Streaming audio — SiriusXM / Pandora / SoundCloudVia Amazon DSP (AdsWizz SSP) — select-advertiser, not open self-serveNegotiated (PG / preferred deals); no min or CPM disclosed (3-0)Audio audience (through Amazon DSP)[CPM — open]Audio reach for Amazon DSP buyers
Spotify / audioSelf-serve Ads Manager, auction$250 lifetime (~$15/day) (3-0, primary; “no min” refuted 1-2)Audience / playlist / podcast context$15–25 CPM (audio $10–35)Audio brand + reach
Nextdoor AdsSelf-serve (NAM: Quick / Advanced Create)~$1/day / “a few pounds”, no minimum (3-0, primary)Hyperlocal geo + demo + interest + retargeting + lookalike[CPM — open]Local & service DTC
Native / DSP (Taboola, StackAdapt; Outbrain open)Self-serve + managedTaboola $10/day tech (rec. 10×CPA / $50); StackAdapt no public min, **$5K/mo de-facto** (3-0)Contextual / audience$0.10–0.90 CPC; $5–12 CPMContent-led / editorial arbitrage
Newsletter — beehiiv Ad NetworkManaged / sales-gated (advertiser side)$15,000 entry min ($15–30k typical, 4–6 wk) (3-0, primary)ML contextual (audience-fit + category tags)CPM (per 1k opens) or CPC (~$2 ex.) (3-0)Brands buying curated newsletter reach at scale
Newsletter — Paved / SponsyMarketplace / self-serve (claimed)[unverified — open] (Paved “self-serve, no-min” refuted 1-2)Newsletter audience[unverified — open]Niche direct sponsorships

Note on “minimum spend.” Sources routinely conflate the hard platform floor ($5/day or less on most alt-social) with the recommended test budget to exit the learning phase ($20–150/day). Both are useful, but they are different numbers — the floor tells you if a channel is accessible; the test budget tells you what a fair read costs. Don’t confuse “cheapest legal spend” with “enough to learn anything.” Note that CTV/audio “hard floors” ($500 Roku/Hulu, $250 Spotify) are scoped to the self-serve path — managed/reseller buys can go lower, and premium/interactive formats sit behind managed buys (~$50K+).

CPM / CPC benchmarks (2025–26)

Live pricing, from a dedicated verification pass. Read every figure with its credibility flag — ad pricing is volatile, seasonal, and mostly secondary-sourced, so these are triangulated ranges, not quotes. Credibility tiers: 🅐 large-sample benchmark (Gupta Media first-party tracker / WordStream N-stated studies) · 🅑 triangulated agency/blog (≥2–4 independent sources agreeing; no platform publishes official CPMs for audio/CTV/native) · 🅒 single/weak source.

ChannelCPMCPCCredibilityNote
Pinterest~$4.67 all-verticals · ~$3.00 retail/ecom$0.50–0.70 (retail)🅐 CPM (Gupta) · 🅑 ecom cutCheapest mainstream social CPM
Native DSP (Taboola/Outbrain/Teads)$5–12$0.10–0.90 (discovery ~$0.30–0.60)🅑Cheapest CPC entry; competitive verticals $2–5 CPC
Reddit$2–6 standard (~$3.50 central; full range $0.50–15)[open] — see note🅑 CPM · CPC unverifiedPrior “$0.59 CPC” could not be re-triangulated; two replacements refuted
Microsoft/Bing[CPM open]~$1.54🅐 (prior pass)33–42% below Google
Snapchat~$8.60 blended, spiking to $12.84 (Oct)~$0.51🅐 (Gupta)Highest-CPM major social; strong Q4 seasonality
Podcastpre-roll $15–25 · host-read mid-roll $25–40 · programmatic $5–15—🅑 (prior pass)Placement-tiered
Streaming audio (Spotify)$15–25 standard ($10–35 broadly, ~$20 avg)—🅑Programmatic audio $5–15 (cheapest audio tier)
CTV self-serve (Roku/Hulu)$15–40 standard AVOD/FAST (~$25) · $35–65 premium—🅑Roku Ads Mgr $25–35; Hulu $15–40 (~$25–32)
Telegram0.1 TON min (~$0.30–0.50); Tier-1 ~2–2.5 TON—🅐 primary (prior pass)TON-denominated
X (Twitter)[open][open]—No verified pricing (only candidate refuted)
Quora[open][open]—Nothing surfaced
Nextdoor[open][open]—Nothing surfaced
Baseline — Google Search—$5.26🅐 (WordStream, N=16,446)The intent-priced comparison
Baseline — Meta~$8.19—🅐 (Gupta)Same tracker as Pinterest/Snapchat → internally comparable

The three cost tiers. Against Google ($5.26 CPC) and Meta ($8.19 CPM): cheap discovery — Pinterest ($4.67 CPM, $0.50–0.70 ecom CPC), native DSP ($0.10–0.90 CPC / $5–12 CPM), Reddit ($2–6 CPM) — is where the CPM arbitrage lives; mid/premium social — Snapchat, the highest-CPM major social; high-attention inventory — streaming audio ($15–25 CPM) and CTV ($15–65 CPM) carry the steepest CPMs because they sell attention, not clicks. The through-line: the access-cheap channels (Pinterest, native) are also the CPM-cheap ones; the newly-self-serve premium inventory (CTV/audio) is cheap to enter but expensive per impression.

Honest caveats. Only the Gupta social CPMs (Pinterest, Snapchat, Meta) and the WordStream Google/Reddit figures are large-sample. Everything in CTV, audio, and native is triangulated agency/blog data — no platform publishes official CPMs there. Most figures are US / all-verticals; only Pinterest has a clean DTC-ecom cut, and all-vertical aggregates understate competitive DTC verticals. Snapchat’s CPM nearly doubled Jan→Oct 2025 — treat any single CPM as a snapshot. Refresh before publishing if >1 quarter old.

Attribution transparency — which channels credit themselves for sales they didn’t cause

A channel’s reported return is only as trustworthy as how it counts conversions. The load-bearing distinction — from the glossary/engaged-view-conversions work — is whether a channel folds view-through / modeled credit into its headline conversion number (inflating apparent ROAS on high-demand sites) or reports it separately. This pass verified two channels against primary docs; the rest of the roster is not yet audited on this axis and is flagged honestly below.

ChannelView-through in the headline number?Default VT windowModeled conversions?Verdict
Amazon DSPYes — and worse: blended(varies)Yes — off-Amazon conversions are modeled, and “directly measured and modeled conversions are reported together as a combined result”Opaque (3-0, primary) — you cannot separate tracked from estimated in the headline figure
Amazon Sponsored (Brands / Display / Store ads)Yes, but split-reporting available<14 days (new, exact undisclosed) since Jan 1 2026PartialSemi-transparent (3-0, primary) — the Jan-2026 “shopping-signal enhanced” model tightened view credit (view-through ROAS may read 15–30% lower); a legacy “Purchases (all views)” 14-day metric is retained for comparison
Microsoft AdvertisingYes, by default — folded into the “All conv.” column1 dayUET-basedConfigurable (3-0, primary) — more honest than Amazon (view-through is separable, excluded from “All conv. CPA”, de-duped against clickers, and an account-level checkbox turns it off) — but still view-inclusive unless you opt out
Reddit · Pinterest · Snapchat · X · Telegram · Roku/Hulu CTV[not audited this pass]——Open — verify each channel’s VT-window default and whether it’s in the primary conversions column before trusting a reported ROAS

The practitioner rule: before you believe any alt-channel’s ROAS, check three things — (1) is view-through folded into the headline number, (2) what’s the default attribution window, (3) can you export click-through-only. This is the same discipline as the YouTube/EVC case: the platform’s default is tuned to flatter itself. Amazon’s Jan-2026 tightening is a live example worth tracking — it’s a rare case of a platform reducing its own credited return. (Re-check within a quarter; all three findings are dated within weeks of this pass.)

Ranked shortlist — for DTC/ecom diversifying off Meta

Weighting: real ease of entry (not just the published floor) · targeting quality · audience scale · arbitrage (underpriced attention, low advertiser competition) · fit for a physical-goods DTC product.

Tier 1 — try first

1. Microsoft Advertising (Bing + Audience Network) — the cleanest first move precisely because it has no advertiser-approval wall. CPC ~$1.54 (33–42% below Google), captures bottom-funnel search intent Meta can’t, an older/wealthier demographic that over-indexes for many DTC categories, and LinkedIn profile targeting for anything B2B-adjacent. Import existing Google Ads campaigns and you’re live in an afternoon. Verified 3-0.

2. Reddit Ads — cheap CPM and real arbitrage (standard CPM $2–6 vs Meta’s ~$8.19 is the solid, triangulated figure; subreddit targeting maps to genuine purchase-research communities; near-zero DTC advertiser competition) — but gated, not open-access. (One honesty note: the widely-quoted “$0.59 Reddit CPC” is single-sourced and could not be re-triangulated in a dedicated pricing pass — lean on the CPM advantage, not a precise CPC.) The practical caveat your Meta experience won’t prepare you for:

  • Account approval takes 1–3 business days — automated email/payment/website-URL checks plus manual team review before you can run anything. Ads then face 24–48h review (longer for restricted categories). (Reddit Ads Help — official)
  • Whole verticals cannot self-serve. Alcohol, dating, gambling, financial services, crypto, and pharma require working directly with a Reddit sales rep, licensing, and/or mandatory disclosures. Others are banned outright: weapons, tobacco/vaping/nicotine, and “dupe”/“inspired-by” products.
  • Enforcement escalates: rejection → warnings → spending restrictions → domain/account suspension.

So Reddit is try-first for its economics only if your category clears policy. If you’re in a restricted vertical, Reddit’s self-serve floor is effectively closed and it drops out of Tier 1 for you entirely — the arbitrage is real but conditional. This is the mirror image of the Telegram access story: here the cheap published number hides the barrier.

Tier 2 — visual DTC + demand capture

3. Pinterest / Snapchat — self-serve, low floors, visual formats that suit product-led DTC and planning-stage intent (Pinterest especially). Snapchat’s floor is verified $5/day (primary; $20–50/day recommended, new accounts under a soft ~$50/day ramp cap that lifts after the first billing cycle). Pinterest is now verified too (primary): fully self-serve, no verification/approval gate, no enforced hard minimum, with a genuinely useful targeting set for ecom — keyword/search-intent targeting (Pinterest behaves like a search-and-discovery engine), plus interest / audience / Actalike lookalikes / demo / geo, and catalog-driven shopping ads (standard, video, personalized collections, carousel). Its AI product Performance+ is just a toggle, not an entry barrier. And it’s cheap — ~$3–4.67 CPM with a $0.50–0.70 ecom CPC (see benchmarks), the lowest CPM of any mainstream social channel. The one caveat: aggregators report a ~$15/day minimum on brand-new accounts during setup (not in official docs).

4. Amazon Ads / retail media — for physical-goods DTC, the place to capture existing demand rather than create it. Self-serve DSP minimum dropped at unBoxed 2025, so access is no longer gated by a hard floor — but a real optimization signal still needs ~$10–15K/mo (directional). Three access tiers (self-serve DSP / managed ~$50K/mo / API) let you match commitment to scale.

Tier 3 — specialist / arbitrage

5. Telegram Ads — cheap attention, specialist access. Not a default first-choice for a typical Western DTC brand, for two structural reasons: (a) targeting is contextual/channel-level (topic, specific channels, language, geo) with no interest/behavioral layer — alien to a Meta-trained buyer optimizing on audience signals; and (b) the access model is confusing (see deep-dive below). Where it wins: audiences that concentrate in identifiable public channels, and CIS/EU-adjacent markets where Telegram’s density is real (see marketing/telegram-marketing-channel for the geographic case). The friction is the arbitrage — few Western DTC advertisers have crossed it, so attention is underpriced.

Also credible — niche-fit and second-wave

Second research pass verified these enough to place them, though most still lack a live CPM figure:

  • Nextdoor — genuinely low-friction and under-used: self-serve, ~$1/day floor (no minimum spend), and — unusually for a low-floor channel — audience/behavioral targeting (hyperlocal geo + demographic + interest custom audiences + retargeting + lookalikes), not just contextual. The obvious fit for local and service DTC (home, local retail, trades). (3-0, primary.)
  • Spotify / audio — self-serve Ads Manager, auction-based, $250 lifetime (~$15/day) floor (this confirms the figure the first pass couldn’t). A brand/reach audio channel, not direct-response-first. (3-0, primary.)
  • CTV went self-serve — and the two biggest holdouts opened up. Contrary to the “CTV = big budgets” reflex, Roku Ads Manager (Sept 2024) and Hulu Ad Manager / Disney Campaign Manager run true self-serve at a $500 hard floor, no revenue requirement (video-only 15/30s). New this pass: Netflix is now buyable with no minimum spend through The Trade Desk’s Sellers & Publishers 500+ marketplace (previously PG/PMP-only) — the catch is it’s self-serve within TTD, so you need a Trade Desk seat, not an open signup (3-0). And Amazon Streaming TV (Prime Video, Twitch, Fire TV channels) is open to any Brand Registry seller at a $0 self-service minimum (2-1), with the DSP tiers (~$10K rec / $50K managed) sitting separately above it. (Correction: Freevee is gone — Amazon shut the standalone app Aug 2025 and folded it into Prime Video’s “Watch for Free.”)
  • Streaming audio beyond Spotify. SiriusXM Media’s streaming-music inventory (Pandora, SoundCloud US) became buyable through Amazon DSP via the AdsWizz SSP (Sept 2025) — but it’s a select-advertiser rollout on negotiated PG/preferred deals, not open self-serve, and no minimum or CPM has been disclosed (3-0). Access = an Amazon DSP seat, not a standalone console. [pricing — open]
  • X (Twitter) — eyes open. Fully self-serve at a $0 spend floor, with Dynamic Product Ads for ecom retargeting — but two real frictions: (a) you must hold a paid verification subscription to advertise at all (see the both-directions table above), and (b) brand safety is severely depressed — only 4% of marketers rate X brand-safe vs 39% for Google (Kantar), and a net 26% planned to cut X spend in 2025. Not blocked, but a channel to enter deliberately, not by default. (3-0, primary + Kantar.)

Higher-commitment — second-wave

  • Native / DSP (Taboola, StackAdapt) — self-serve-capable but a real commitment: Taboola has a ~$10/day technical minimum (but recommends 10× your CPA goal, ≥$50/day, for optimization signal); StackAdapt publishes no hard floor but sits at a de-facto ~$5K/mo self-serve entry. Content-led/editorial arbitrage. Outbrain’s mechanics stay [unverified — open]. (3-0.)
  • Podcast — the clearest cheap entry among the commitment channels (programmatic DAI at $5–15 CPM, verified) if you want a trust-transfer channel.
  • Newsletter sponsorship — read the access model first. beehiiv’s Ad Network is not the low-floor self-serve channel people assume: its advertiser side is fully managed and sales-gated, with a hard $15,000 entry minimum ($15–30k typical first buy over 4–6 weeks), priced CPM (per 1,000 opens) or CPC (~$2 example) with ML contextual matching. That’s an enterprise-scale commitment, not a $5/day test. Paved/Sponsy (the more self-serve, lower-floor marketplaces) stay [unverified — open] — and Paved’s “self-serve, no-minimum” claim specifically failed verification, so confirm before assuming.

Evaluated this pass — sanctioned or not-yet-verified

New candidate channels were checked; most did not produce surviving verified claims and are placed honestly rather than asserted:

  • VK Ads — do not use from the EU. On July 13, 2026 the EU imposed an asset freeze on VK Company (operator of VKontakte) plus a prohibition on EU persons/companies making funds available to the entity (Council of the EU, 3-0). Because the sanction attaches to the legal entity, an EU/Baltic advertiser paying for VK Ads would be providing funds to a sanctioned entity — legally prohibited. Classify: sanctioned — do-not-recommend from the EU. (Directly relevant to Baltic/CIS-adjacent advertisers who might otherwise consider it for Russian-speaking reach.)
  • Yandex Direct — [unverified — open]. The circulating claims that Yandex is “on the OFAC list” (refuted 0-3) and “no longer available to US/EU advertisers” (refuted 1-2) did not verify — so do not assert either the sanction or free availability. Needs its own pass against primary Yandex docs + a current EU-sanctions check before it earns a catalog entry.
  • Apple Search Ads — [unverified — open]. The commonly-quoted Basic (CPI, $5–10K budget band) and Advanced (CPT ~$1.56) specifics both failed verification (0-3 / 1-2) — likely stale after Apple restructured its ASA tiers. Warrants a dedicated primary-source pass.
  • Twitch / Discord / LinkedIn / Waze — [unverified — open]. Surfaced but not verified this pass. Directional signal only (e.g. Discord Ads/Quests appears sales-gated with six-figure floors; LinkedIn carries the highest CPMs of any major platform) — none placed as fact. LinkedIn’s targeting is already accessible cheaply via Microsoft Advertising (see Tier 1), which may be the better route to LinkedIn-profile data than LinkedIn’s own ad manager.

Deep dive: Telegram Ads as a paid product

Telegram is the channel most often misdescribed, so this section is deliberately precise about what survived verification versus the folklore that failed it.

Pricing — TON, not EUR

The official Telegram Ad Platform prices CPM in Toncoin (TON), not fiat. Minimum CPM is 0.1 TON (~$0.31–0.50 depending on the TON/USD rate; the USD figure floats). CPMs vary sharply by geography — English-speaking markets run higher (~2–2.5 TON, roughly €3.50–5.00), cheaper markets like Kazakhstan far lower (0.2–0.7 TON). A Cost-Per-Result option and Telegram Stars funding (with a ~30% discount) exist alongside CPM. (Min-CPM floor verified 3-0 against the official getting-started page; geo figures 2-1.)

The old “€2 CPM floor / €3–6 realistic CPMs” fiat model that circulates in guides is outdated and refuted — do not use it.

Access — the bifurcated model that explains the underutilization

Three coexisting paths, which is why the channel confuses advertisers:

  1. Legacy EUR direct account — the famous ~€2M top-up to be an official direct advertiser. Real, but increasingly not the relevant number for a new advertiser.
  2. Resellers / agencies — meaningful testing in the €1,000–5,000 range.
  3. TON self-serve — a low-friction path that has emerged by 2026, entry ~$50–60, bypassing both the €2M gate and reseller markup. Telegram’s own getting-started page lists no public minimum deposit — only the 0.1 TON minimum CPM — consistent with this path existing.

Separately, the cheap “Telegram” inventory that circulates in performance-marketing circles runs on third-party Mini App networks (RichAds, PropellerAds), not the official platform — entry from ~$150, formats from $0.015/click and $3 CPM. This route bypasses the official platform’s access barrier and its gambling ban, which is why cheap iGaming inventory lives here (see marketing/telegram-marketing-channel). (Third-party $150 path verified 3-0; €2M/reseller split 2-1.)

Targeting and format — what we can and can’t assert

A dedicated primary-source pass (July 2026) resolved the format spec and split a claim the earlier round had lumped together:

  • The 160-character text cap is real — ads.telegram.org/guidelines states verbatim: “The maximum length of the ad text is 160 characters including spaces.” (Confirmed 3-0, primary.) This was previously on the do-not-cite list bundled with “text-only”; the character cap is true, the “text-only” part is what’s false.
  • But the format is not text-only. A Telegram ad carries a title field + the ≤160-char text + one button, and media is supported — the “text-only, no media” framing failed 0-3 and stays do-not-cite.
  • Targeting is contextual, not channel-only. You buy against channels / topics / language / geo — not individual interest/behavioral profiles — but the over-restrictive “channel-level only, nothing else” claim also failed 0-3; the topic/language/geo layers exist.

Still open: the full enumerated targeting parameter set as of 2026 wasn’t exhaustively re-listed against the live creating-ads flow, and whether the legacy €2M EUR direct account still runs in parallel with TON self-serve remains unconfirmed.

Why mainstream advertisers underuse it

Three compounding reasons, all consistent with the evidence: (1) awareness gap — the €2M figure is what everyone “knows,” so the $50 TON path goes unseen; (2) targeting-model mismatch — contextual/channel buying doesn’t fit a workflow built around Meta’s audience signals; (3) no discovery surface — Telegram has no algorithmic feed pushing your ad to the right audience, so you must already know which channels your buyers inhabit (the keyword × region scout phase in marketing/telegram-marketing-channel). None of these is a quality problem — they’re friction. For an advertiser willing to cross the friction in the right geography, that’s the arbitrage.

Do-not-cite (claims that failed verification)

These plausible-sounding, widely-repeated claims were refuted 0-3 (or 1-2) and must not be carried as fact:

  • ⚠️ Telegram “text-only, no media” (0-3) — CORRECTION (July 2026): the 160-character text cap IS real and confirmed 3-0 against ads.telegram.org/guidelines (“maximum length of the ad text is 160 characters including spaces”). Only the “text-only / no media” framing is false — ads carry a title + the ≤160-char text + one button + media. Cite the 160-char cap; do not cite “text-only.”
  • ❌ Telegram “only channel-level targeting — no topic/language/geo” (0-3) — the over-restrictive version; the layers exist
  • ❌ Telegram “€2 CPM floor, €3–6 realistic CPMs” fiat pricing (0-3) — it’s TON now
  • ❌ Telegram “€1 per 1000 impressions (€0.5 in Politics)” (1-2)
  • ❌ Telegram “$1,500 agency vs €2M direct” and “€3,000–5,000 reseller minimums” as the definitive split (0-3)
  • ❌ Reddit “CPM 5.1× cheaper than Meta (eCPM $2.29 vs $11.73)” (0-3) — Reddit’s advantage is CPC-ordering, not a proven CPM discount
  • ❌ Reddit “$0.20–1.20 CPC, 260M weekly active users” (0-3)
  • ❌ Amazon “$900/month to start, $0.75–2.50 CPC” (0-3)
  • ❌ X/Pinterest “$0.01/day floor, same as Google” (1-2) — the “no enforced minimum” shape holds; the specific $0.01 figure did not
  • ❌ Quora “no hard minimum daily budget” (0-3) — the $5/day floor stands (confirmed against Quora’s own pricing page)
  • ❌ Spotify “no fixed hard minimum” (1-2) — the $250 lifetime floor is operative
  • ❌ Paved “self-serve, no minimum spend on PPC” (1-2) — unconfirmed; do not assume Paved is a no-floor self-serve channel
  • ❌ X “$0.01/day floor, same as Google” (refuted) — the correct answer is no spend minimum; the real entry cost is the paid verification subscription
  • ❌ X “you can advertise with just a credit card, no Premium needed” (0-3) — a paid verification subscription is required to be eligible
  • ❌ X “$5–10 CPM” (1-2) — the only X pricing candidate; did not survive. X CPM/CPC remains genuinely unknown here
  • ❌ Reddit “$0.10–0.80 CPC (consumer) / $0.20–0.50 (retail)” (0-3) — do not quote a Reddit CPC; only the $2–6 CPM is verified
  • ❌ Snapchat “$6.43 June / $10.48 Oct CPM” (0-3) — the verified path is ~$8.60 blended, $12.84 Oct (Gupta)
  • ❌ Native DSP “$0.10–0.50 CPC ‘cheapest entry’ / $3–7 CPM” (0-3 / 1-2) — verified ranges are $0.10–0.90 CPC / $5–12 CPM
  • ❌ CTV “$20–35 / $35–55 / $55–80 CPM tiers” (1-2) — slightly inflated; verified is $15–40 standard / $35–65 premium
  • ❌ X (Twitter) “$5–9 CPM / $0.30–2.50 CPC” (autotweet.io) (0-3) — single weak blog, no primary/triangulated corroboration; X CPM/CPC remains genuinely unknown
  • ❌ Apple Search Ads “Basic CPI $5–10K budget band” (0-3) and “Advanced CPT ~$1.56” (1-2) — likely stale after Apple’s ASA tier restructure; do not quote until re-verified against Apple docs
  • ❌ Yandex “on the US OFAC list” (0-3) and “no longer available to US/EU advertisers” (1-2) — the specific sanctions characterization did not verify; do not assert Yandex is sanctioned or freely available
  • ❌ Amazon Streaming TV “$0 entry, previously $50K-only managed” (1-2) — the “$0 self-service min for Brand Registry sellers” holds, but the “previously $50K-only” contrast is overreach
  • ❌ Amazon Freevee as a live channel — shut Aug 2025, folded into Prime Video’s “Watch for Free”
  • ❌ SiriusXM / Amazon audio “self-serve” — it’s a select-advertiser, negotiated (PG/preferred) integration, not open self-serve

Honest gaps (open, not asserted)

A fifth pass (2026-07-28) closed the CTV/audio-access and attribution-transparency cells and resolved the Telegram format spec. What remains genuinely unverified after five passes:

  • CPM/CPC pricing gaps — STILL open after a second dedicated pricing pass. X, Quora, Nextdoor, Reddit CPC, Bing CPM, and StackAdapt did not close: the only X candidate (autotweet.io) was refuted 0-3, and no A-tier figure surfaced for any of the others. These are a genuine, persistent hole — secondary sources simply don’t publish trustworthy numbers here. The honest conclusion: effective CPMs on these channels are only knowable from a live test, not desk research. Everything else has a credibility-flagged benchmark above.
  • Attribution transparency audited for only 2 channels. Amazon (DSP opaque; Sponsored semi-transparent post-Jan-2026) and Microsoft (view-inclusive-by-default, configurable) are verified against primary docs. Reddit, Pinterest, Snapchat, X, Telegram, and Roku/Hulu CTV remain un-audited on this axis — their view-through-window defaults and headline-column behavior need a dedicated pass.
  • New candidate channels need their own passes: Yandex Direct and Apple Search Ads (both had circulating specifics refuted this pass); LinkedIn, Twitch, Discord, Waze (surfaced but unverified). VK is resolved — sanctioned, do-not-recommend from the EU.
  • X native targeting taxonomy — access, pricing gate, formats, and brand-safety are verified, but the current targeting-option set (keyword/interest/follower-lookalike/custom-audience/geo) wasn’t enumerated.
  • CTV beyond Roku/Hulu — mostly closed this pass: Netflix (TTD, no min), Amazon Streaming TV ($0 Brand-Registry self-service), and SiriusXM/Pandora (Amazon DSP/AdsWizz, select-advertiser) are now placed. Still open: CPMs for all three, Netflix formats, and Outbrain mechanics/floors.
  • Pinterest new-account minimum — the aggregator-reported ~$15/day setup minimum on brand-new accounts (not in official docs); a direct account-creation test would settle it.
  • Paved / Sponsy access model and minimums (the low-floor self-serve newsletter cell) — unconfirmed.
  • Exact current Telegram targeting parameters + format spec, whether the legacy €2M EUR account still runs parallel to TON self-serve, and effective (not floor) CPMs on Telegram vs Reddit vs Bing for a DTC product — only a live test answers the last one.
  • Reporting/attribution transparency for most channels — not systematically verified.

Key Takeaways

  • Published entry floors mislead in both directions. Telegram’s €2M overstates (a $50 TON self-serve path exists); Reddit’s $5/day understates (an approval wall gates it). Verify the real access path before assuming open or closed.
  • Microsoft/Bing is the frictionless first move — no approval gate, CPC 33–42% below Google, bottom-funnel search intent, older/wealthier base, LinkedIn targeting.
  • Reddit has the best economics but is conditional — cheap CPC and real arbitrage, if your category clears policy. Alcohol, dating, gambling, finance, crypto, and pharma must work with a Reddit sales rep; weapons, tobacco/vaping, and dupes are banned.
  • Telegram Ads is a specialist/arbitrage play, not a default DTC channel — cheap attention, but contextual (channel-level) targeting and a confusing access model. Best in CIS/EU-adjacent markets and where buyers cluster in known channels.
  • Distinguish the platform floor from the test budget. $5/day tells you a channel is accessible; $20–150/day is what a fair read costs.
  • Most repeated Telegram “facts” are wrong — the 160-char/text-only, fiat-CPM, and €2M-only framings all failed verification. Use the TON-pricing, contextual-targeting, three-path-access model instead.
  • CTV is no longer big-budget-only — and Netflix + Amazon just opened. Roku and Hulu run self-serve at a $500 floor; Netflix is now buyable with no minimum via The Trade Desk, and Amazon Streaming TV (Prime Video/Twitch/Fire TV) at $0 for Brand Registry sellers. The barrier moved; the reflex hasn’t. (SiriusXM/Pandora audio also opened — but via Amazon DSP, select-advertiser, not open self-serve.)
  • Check attribution before you believe a channel’s ROAS. Amazon DSP is opaque — it blends modeled + measured conversions into one headline number you can’t split. Microsoft folds view-through into “All conv.” by default (configurable off). Amazon tightened view-attribution Jan 1 2026 (view-through ROAS may read 15–30% lower) — a rare platform reducing its own credited return. The rest of the roster is un-audited on this axis; verify the view-through-window default before trusting a number. Same discipline as the glossary/engaged-view-conversions / YouTube case.
  • VK Ads is off the table for EU advertisers. EU sanctions (July 13 2026) froze VK Company’s assets and prohibit EU persons from making funds available to it — so paying for VK Ads from the EU/Baltics is legally prohibited. Yandex and Apple Search Ads stay [unverified — open] (circulating specifics failed verification).
  • Nextdoor is the sleeper for local/service DTC — self-serve, ~$1/day floor, and rare among low-floor channels in offering audience/behavioral targeting (interest, retargeting, lookalikes), not just contextual.
  • Read the newsletter-network access model before budgeting. beehiiv’s Ad Network is managed and sales-gated with a hard $15,000 minimum — not the $5/day self-serve test people assume. Paved/Sponsy may be lower-floor but are unverified.
  • X’s “$0 floor” hides a pay-to-access gate — you need a paid verification subscription to advertise, and its brand-safety reputation is the weakest of any major platform (4% vs Google’s 39%). A third case of the same lesson: the “floor” is often not a spend number.
  • Pinterest is a verified, low-friction ecom channel — no approval gate, no hard minimum, and rare among alt-social in offering keyword/search-intent targeting plus catalog-driven shopping ads. Under-rated for product-led DTC.
  • On pricing, cheap-to-enter and cheap-per-impression line up. Pinterest (~$3–4.67 CPM) and native discovery ($5–12 CPM, $0.10–0.90 CPC) are the CPM-cheapest alternatives; Snapchat is the priciest major social; CTV/audio ($15–65 CPM) are cheap to enter but expensive per impression. All benchmarks are snapshots — refresh quarterly.

Sources

Second pass (2026-07-14) — primary sources:

Third pass (2026-07-14) — Pinterest & X, primary sources:

Fourth pass (2026-07-14) — CPM/CPC benchmarks:

Fifth pass (2026-07-28) — CTV/audio access, attribution transparency, new candidates (primary where available):

By Andrej Ruckij