> ## Documentation Index > Fetch the complete documentation index at: https://docs.useadmesh.com/llms.txt > Use this file to discover all available pages before exploring further. # CPM vs CPX > Why CPX exists and how it differs from traditional CPM advertising ## Why CPX Exists **CPX (Cost Per Exposure)** was created because **CPM (Cost Per Mille)** doesn't work in AI conversations. In traditional advertising, CPM measures how many people *saw* an ad. But in AI conversations, there is no "page load" or banner view. Instead, **AdMesh tracks exposure when a recommendation appears contextually inside an answer** — a verified moment where the user is reading about something relevant. **That's why CPX exists:** it bridges the gap between *visibility* and *intent*, measuring the real value of exposure inside AI-driven user interactions. > CPX = the new currency of visibility for conversational AI environments. *** ## The Problem with CPM ### REAL PROBLEM #1: CPM rewards volume, not quality CPM charges you per 1,000 exposures, regardless of: * Whether the user actually saw the ad * Whether the user had any intent to buy * Whether the ad was relevant to the user's current context **Result:** You pay for low-value exposures that don't drive results. ### REAL PROBLEM #2: CPM doesn't measure real attention CPM counts: * Page loads (even if user never scrolled) * Banner views (even if user ignored it) * Feed exposures (even if user scrolled past) **But it doesn't measure:** * Whether the user actually read the ad * Whether the ad was contextually relevant * Whether the user had purchase intent ### REAL PROBLEM #3: CPM creates perverse incentives Because CPM rewards volume, publishers optimize for: * More page loads * More "scroll depth" tracking * More feed units All because CPM rewards **volume**, not **quality**. You pay \$0.003 per impression — yes. But 60–80% of those exposures are **low-value**. That's why CPA and CPC outperform CPM in many verticals. *** ### REAL PROBLEM #4: CPM is meaningless in AI environments AI chats and agents don't have: * pages * feeds * banners * scroll depth * viewable areas So you can't use CPM because there's no "impression surface." CPX fixes that by counting: > "Did the AI *surface* the brand inside the answer in a meaningful, verifiable way?" *** ## The Intent Model: CPX CPX (Cost per Exposure) measures verified, human-seen exposures tied to an explicit intent signal inside an agentic environment. Each CPX event is: * Signed by verifiers (platform + network) * Linked to a serve\_token * Auditable end-to-end Advertisers pay only for real, verified interactions — not assumptions. *** ## Why Create a New Metric? Many advertisers ask: *why change?* CPM has worked for decades and is easy to understand. But AI-driven discovery breaks the assumptions that made CPM valid. In a conversational context: * There are no "pages" or "exposures" to count. * What matters is whether an AI system truly *exposed* a verified recommendation to a user with intent. * CPX gives advertisers confidence that every dollar tracks to a verified, human exposure — not a background impression. So CPX doesn't replace CPM out of novelty — it replaces it out of necessity. *** ## Corrected Explanation **Simple Version:** CPM charges advertisers for every impression — even if the impression had low attention, poor visibility, or no user intent. The price is prorated (for example, a $3 CPM means $0.003 per impression), but the problem is **what gets counted as an impression**, not the math. In AI conversations, there are no pages or banners, so CPM does not make sense. CPX measures **verified exposures inside conversations**, which is a more accurate and intent-aligned unit. *** ## Comparison Table | Aspect | CPM (Traditional) | CPX (AdMesh) | | -------------------- | ----------------------------------- | -------------------------------------------------- | | **What it measures** | Per 1,000 page exposures | Per verified exposure in AI response | | **Based on** | Banner views or ad slots | User intent and query relevance | | **Tracking method** | Cookies and viewability tags | Privacy-safe, server-side exposure validation | | **Optimized for** | Websites and feeds | AI chats, search, and agents | | **Measures** | Reach, not relevance | Visibility within context | | **Inflation risk** | Often inflated by passive exposures | Fired only when recommendations are actually shown | | **Intent alignment** | No | Yes — tied to user query | | **Verification** | Assumed | Verified and signed | *** ## Key Differences ### 1. Measurement Unit * **CPM**: Counts exposures (page loads, banner views) * **CPX**: Counts verified exposures (AI recommendations shown with intent) ### 2. Context Awareness * **CPM**: No context — same price whether user is browsing or actively seeking * **CPX**: Context-aware — price scales with relevance to user's query ### 3. Verification * **CPM**: Assumes impression occurred (may not have been seen) * **CPX**: Verified exposure — confirmed that recommendation was shown ### 4. Intent Alignment * **CPM**: No intent signal — user may have no interest * **CPX**: Intent-aligned — user is actively asking for something related ### 5. Environment * **CPM**: Designed for web pages and mobile apps * **CPX**: Designed for AI conversations and agentic environments *** ## When to Use CPX vs CPM ### Use CPX When: * ✅ Advertising in **AI conversations** (ChatGPT, Claude, Perplexity, etc.) * ✅ You want **intent-aligned** placements * ✅ You need **verified exposures** (not assumed exposures) * ✅ You value **contextual relevance** over volume * ✅ You're targeting **decision-ready audiences** ### Use CPM When: * ✅ Advertising on **traditional websites** (banner ads, display) * ✅ You need **high-volume** reach * ✅ You're running **awareness campaigns** (not conversion-focused) * ✅ You have **simple measurement needs** (exposures only) * ✅ You're advertising in **non-AI environments** *** ## Example: CPM vs CPX in Practice ### CPM Scenario (Traditional Web) ``` User visits website → Page loads → Banner ad shown → CPM charged ($0.003 per impression) → User may or may not have seen the ad → No intent signal — user was just browsing ``` **Result:** You pay for an impression that may not have been seen or relevant. ### CPX Scenario (AI Conversation) ``` User asks: "What's the best CRM for small teams?" → AI surfaces your CRM recommendation → CPX charged ($0.05 per exposure) → Verified exposure — user is actively seeking a CRM → High intent — user is decision-ready ``` **Result:** You pay for a verified, intent-aligned exposure to a qualified user. *** ## Summary CPX doesn't replace CPM because it's "better" — it replaces it because **CPM doesn't work in AI environments**. * **CPM** = Volume-based, impression counting for web pages * **CPX** = Quality-based, verified exposure counting for AI conversations Both have their place, but for AI-native advertising, **CPX is the only metric that makes sense**. *** ## Related Documentation Deep dive into CPX calculation and mechanics. See how CRS affects CPX pricing.