Working Paper — March 2026
This paper introduces conversation advertising as a novel category in commercial communication, distinct from broadcast advertising, cold outreach, and first-wave conversational marketing. Conversation advertising is defined by four structural properties: AI agents pay to initiate conversations with human recipients; recipients opt in and set their own price; the conversation is the unit of value; and attention is traded rather than extracted. We develop the theoretical basis for this category, examine its market structure as a two-sided marketplace, and distinguish it from all existing commercial communication formats. The emergence of large language models as capable sales agents, combined with growing regulatory pressure on unsolicited AI outreach, creates the conditions in which conversation advertising becomes both technically feasible and commercially necessary. We argue that conversation advertising is not an incremental improvement on existing formats but a structural innovation: the first commercial communication category in which the human recipient is an economically empowered participant rather than a passive audience.
Every existing approach to commercial outreach involves a structural compromise between reach, conversation, and consent. Broadcast advertising achieves reach but sacrifices conversation — the message cannot listen, respond, or adapt. Cold outreach achieves conversation but sacrifices consent — the recipient did not ask to be contacted and is not compensated for their time. First-wave conversational marketing (Cancel and Gerhardt, 2019; Jaffe, 2007) achieved consent but sacrificed outbound reach — the brand could only converse with consumers who had already arrived at its door.
No existing format combines all three properties. Conversation advertising does.
The enabling conditions for this category have converged in the mid-2020s. Large language models have produced AI agents capable of conducting expert, personalized sales conversations at unlimited scale — creating a new class of commercial actor with unprecedented demand for human conversational engagement. Simultaneously, regulatory environments in the EU and US are tightening around unsolicited AI-to-human outreach, raising the cost of consent-free communication and creating structural incentive to build consent into the channel itself. And the attention economy has produced a recipient population that is simultaneously overloaded with unsolicited messages and undercompensated for the attention those messages consume.
Conversation advertising addresses all three conditions: it gives AI agents a legitimate, consented channel for outbound conversation; it gives regulators a structural consent mechanism rather than a legal patch; and it gives human recipients economic participation in the transactions that consume their attention.
This paper develops the theoretical framework for conversation advertising, examines its market structure, and establishes its position as a genuinely new category rather than a variation on existing formats.
To establish conversation advertising as a new category, we first map the structural limitations of all existing approaches to commercial outreach, and identify the combination of properties none of them achieves.
Broadcast advertising — display, search, social, video — achieves scale but not conversation. The message is fixed before delivery and cannot respond to what it encounters. A recipient with a question, an objection, or a specific situation receives the same message as every other recipient. The conversion gap between a broadcast message and a skilled sales conversation is well documented: professional salespeople close qualified prospects at 20 to 40 percent (Salesforce, 2024); display advertising closes at below 0.1 percent. The difference is not reach or targeting — it is the presence or absence of dialogue. Broadcast advertising has no answer to this gap because the format structurally excludes conversation.
Cold outreach — cold email, cold calling, and increasingly, AI-driven outbound sequencing via platforms such as 11x, Artisan, and Landbase — achieves conversation but not consent. The recipient did not invite the contact, is not compensated for engaging, and has no market mechanism to signal the value of their time. AI SDR platforms have made this problem structurally worse: by dramatically reducing the cost of initiating outreach, they have increased message volume without increasing recipient willingness to engage. Reply rates on cold email have fallen industry-wide as AI-generated outreach has proliferated. The consent problem is not a quality problem that better personalization can solve — it is a structural feature of the cold outreach model.
Regulatory pressure is compounding this structural problem. CAN-SPAM and GDPR were designed for human senders; their application to AI agents sending thousands of messages per day is contested and evolving. Regulators in the EU and US are actively examining AI-to-human commercial outreach. The cold outreach model carries compliance risk that will only grow as AI agent deployment scales.
First-wave conversational marketing — the chatbot and messaging platform model pioneered by Drift, Intercom, and their successors (Cancel and Gerhardt, 2019) — achieved consent but not outbound reach. The consumer had to arrive at the brand's owned channel before any conversation could begin. This inbound constraint is fundamental: the brand can only converse with consumers who have already found it. For most brands, inbound interest represents a fraction of the total addressable market. First-wave conversational marketing solved the interface problem (replacing forms with chat) but did not address the distribution problem (reaching consumers who have not yet expressed interest).
Several proposals have sought to compensate consumers for their attention — most notably Brave Rewards, which pays users in BAT tokens for viewing advertisements. These approaches achieve compensation but not conversation: the recipient is paid to be exposed to a message, not to engage in a dialogue. The value proposition is passive attention, not active exchange. The commercial effectiveness of passive compensated attention has not been demonstrated to exceed that of uncompensated broadcast advertising, because the format does not address the structural limitation of broadcast: the absence of dialogue.
The four properties that together define a commercially effective, ethically sound, and regulatorily defensible outbound commercial communication format are:
No existing format combines all four. Conversation advertising does.
| Format | Outbound reach | Conversation | Consent | Compensation |
|---|---|---|---|---|
| Broadcast advertising | ✓ | — | — | — |
| Cold outreach / AI SDR | ✓ | ✓ | — | — |
| First-wave conversational marketing | — | ✓ | ✓ | — |
| Brave Rewards / attention tokens | ✓ | — | ✓ | ✓ |
| Conversation advertising | ✓ | ✓ | ✓ | ✓ |
We define conversation advertising as a commercial communication format with the following four structural properties:
1. Senders pay to initiate conversations. A sender — a brand, an AI agent, or an individual — pays a market-determined price to initiate a conversation with a recipient. The payment is not for an impression or a click but for the right to begin a dialogue. The conversation is the unit of value.
2. Recipients opt in and set their own price. Recipients join the marketplace voluntarily, with full knowledge that senders will pay to initiate conversations. Each recipient has a notice price — the market-clearing cost to reach them — that emerges from competitive bidding among senders rather than being set by the platform or the recipient unilaterally. A recipient in high demand among senders commands a higher price; a recipient with fewer competing bids costs less to reach. The market prices each recipient's attention.
3. Recipients are compensated for engaging. When a recipient reads and responds to a message, they earn the sender's bid. The transaction is direct: value flows from sender to recipient, with the platform facilitating and earning from the exchange. The recipient is not the product being sold to advertisers — they are the supply side of a market for conversational attention.
4. The conversation is conducted by an AI agent. The sender's side of the conversation is conducted by an AI agent with expert product knowledge and sales judgment — not a scripted chatbot or a form-replacement tool, but an agent capable of genuine dialogue: answering questions, handling objections, qualifying fit, and adapting to what the recipient actually says. The AI agent is the commercial actor that makes outbound conversation advertising feasible at scale.
Conversation advertising as a category is enabled by, and in important ways created by, the emergence of capable AI agents. Two aspects of this relationship merit examination: AI agents as the demand side of the marketplace, and AI agents as the conversation technology that makes the format viable.
The deployment of autonomous AI agents for commercial purposes has accelerated sharply since 2024. A new category of AI SDR platform — including 11x, Artisan, Landbase, and others — has emerged to serve brands' demand for AI-driven outbound sales conversations. These platforms collectively represent a large and growing demand for human conversational engagement: AI agents actively seeking to initiate sales conversations with qualified human prospects at scale.
This demand currently routes almost entirely through cold outreach channels — cold email, LinkedIn messaging, and voice calls to prospects who did not request contact. Conversation advertising offers an alternative routing: instead of AI agents cold-contacting humans, they enter a marketplace where humans have opted in, set their price, and are compensated for engaging. The demand is the same; the channel is structurally different.
Prior to large language models, sustaining high-quality sales conversations at scale required human salespeople. The economics were prohibitive for outbound conversation at meaningful scale: a human salesperson conducts perhaps ten substantive conversations per day. Reaching a market at broadcast scale through human conversation would require an impractical number of salespeople.
Large language models dissolve this constraint. An AI agent can conduct thousands of simultaneous conversations, respond in real time, maintain the context of each individual exchange, and apply consistent sales judgment across all of them. The quality of conversation available from a well-designed AI agent — product expertise, objection handling, adaptive listening — approaches that of a skilled human salesperson in many commercial contexts. The scalability constraint that made outbound conversation economically impractical no longer exists.
Conversation advertising is therefore a category that could not have existed before capable AI agents. It is not a digital-era improvement on an existing format; it is a new format made possible by a new technology.
A conversation advertising marketplace is a two-sided market (Rochet and Tirole, 2003) in which senders (brands and AI agents) and recipients (human consumers) transact through a platform that facilitates matching, manages payments, and maintains delivery infrastructure. The cross-network externalities are bilateral: more senders increase demand for recipient attention, raising notice prices and increasing recipient incentive to participate; more recipients expand the addressable audience, increasing the marketplace's value to senders.
This structure differs fundamentally from conventional advertising platforms, in which the platform sells recipient attention to advertisers and recipients receive no direct compensation. In conversation advertising, value flows directly from sender to recipient. The platform's revenue derives from facilitating the market, not from monetizing the audience.
The central pricing mechanism in conversation advertising is the notice price: the market-clearing cost to initiate a conversation with a specific recipient. Notice prices are not set by the platform or by recipients directly; they emerge from competitive bidding among senders. The notice price for a given recipient reflects the highest active bid among senders who want to reach that recipient, plus a unit increment.
This mechanism has several desirable properties. It is incentive-compatible: senders reveal their true valuation of each recipient's attention through their bids. It is dynamic: prices adjust continuously to reflect changing sender demand. It allocates recipient attention to the senders who value it most, which under standard assumptions is associated with higher relevance and mutual value. And it is decentralized: no party sets prices unilaterally.
Senders target recipients by specifying populations whose characteristics suggest genuine interest in the sender's offering. Targeting dimensions include demographic characteristics, geographic location, topical interest communities (cliques), and profile-based interest signals. The marketplace matches sender messages to recipient queues based on these criteria, subject to bid thresholds.
Targeting serves the interests of both parties: senders reach recipients more likely to find the conversation valuable; recipients receive conversations from senders more likely to be relevant. A recipient who receives a conversation from an AI agent whose product is genuinely relevant to their situation is more likely to engage substantively than one who receives a random outbound message. Targeting is the mechanism through which the marketplace produces mutual value rather than merely transacting attention.
Conversation advertising marketplaces support community-based targeting structures in which recipients organize around shared interests and senders target these communities as a unit. Community hosts — curators who organize and grow recipient communities — earn a revenue share from the engagement of their community members. This creates an incentive for community builders to attract engaged recipients and to maintain community quality, aligning host earnings with the value the community delivers to both senders and recipients.
Conversation advertising offers brands and AI developers a new outbound channel with properties unavailable in existing formats: consented access to qualified recipients, conversation as the interaction mode, and market-priced attention that reflects genuine demand. The channel is particularly well-suited to high-consideration products where dialogue moves the buyer more than impressions, and to AI agents whose conversational capability is fully utilized only in a format where the human can respond.
For AI SDR platforms specifically, conversation advertising represents a consent-first alternative to cold outreach channels that are deteriorating in quality and facing growing regulatory scrutiny. The same AI agents that currently cold-contact prospects can operate in a conversation advertising marketplace where the contact is consented, the recipient is compensated, and the conversation starts from a position of mutual willingness rather than adversarial interruption.
Conversation advertising introduces a pricing mechanism for human attention that does not exist in any current advertising format. In broadcast advertising, attention is not priced — it is estimated in aggregate (CPM for a population) and delivered without individual consent. In conversation advertising, each recipient's attention has a specific market price, determined by competitive demand. This is a structural change in how the advertising industry relates to the humans it is trying to reach.
If conversation advertising scales, its existence will create a reference price for individual human attention that currently does not exist. Whether this reference price influences broader advertising market dynamics — or remains confined to the conversation advertising category — is an empirical question the current analysis cannot resolve.
Market liquidity at launch. Two-sided markets are subject to thin-market problems at early stage: insufficient sender demand fails to generate meaningful recipient earnings, reducing recipient participation, which further reduces sender value. Conversation advertising marketplaces require mechanisms to bootstrap both sides simultaneously.
Conversation quality and equilibrium. The marketplace's value to recipients depends on the quality of conversations initiated by senders. A marketplace that permits low-quality AI outreach — agents with poor product knowledge, generic messaging, or coercive sales tactics — will face recipient disengagement and adverse selection. Platform design must include quality signals and enforcement mechanisms.
AI agent disclosure. Recipients engaging with an AI agent are transacting with a non-human entity. The effect of mandatory disclosure on recipient engagement rates, and therefore on the market's equilibrium, is an open empirical question. The consent structure of conversation advertising — recipients opted in knowing they would receive AI-initiated conversations — is the strongest argument that disclosure is already embedded in the format.
Regulatory treatment. Conversation advertising sits at the intersection of advertising regulation, consumer protection law, and emerging AI communication frameworks. Its consent-and-compensation structure distinguishes it structurally from cold outreach, but regulatory classification has not yet been established in any major jurisdiction.
Conversation advertising is a new category because it combines four properties — outbound reach, genuine conversation, recipient consent, and recipient compensation — that no existing commercial communication format combines. This combination is not a design choice among alternatives; it is the minimum set of properties required to make outbound AI-to-human commercial conversation ethically defensible, regulatorily sound, and commercially effective.
The category is enabled by capable AI agents and created by the convergence of three conditions: AI agents generating unprecedented demand for human conversational engagement; regulatory pressure eliminating the consent-free option; and a recipient population that is overloaded with unsolicited messages and undercompensated for the attention they consume.
Conversation advertising is the first commercial communication category in which the human recipient holds a genuine economic position. That is not a feature — it is the definition of the category. Attention that is traded rather than extracted is a different thing, with different properties, different market dynamics, and different implications for the relationship between brands and the humans they are trying to reach.
Cancel, D., & Gerhardt, D. (2019). Conversational marketing: How the world's fastest growing companies use chatbots to generate leads 24/7/365. Wiley.
Goodrich, K. (2011). Anarchy of effects? Exploring attention to online advertising and multiple outcomes. Psychology & Marketing, 28(4), 417–440.
Jaffe, J. (2007). Join the conversation: How to engage marketing-weary consumers with the power of community, dialogue, and partnership. Wiley.
Levine, R., Locke, C., Searls, D., & Weinberger, D. (1999). The Cluetrain Manifesto. Perseus Books.
PageFair. (2024). Global ad blocking behavior report. PageFair Research.
Rochet, J.-C., & Tirole, J. (2003). Platform competition in two-sided markets. Journal of the European Economic Association, 1(4), 990–1029.
Salesforce. (2024). State of sales report, 7th edition. Salesforce Research.
Simon, H. A. (1971). Designing organizations for an information-rich world. In M. Greenberger (Ed.), Computers, Communication, and the Public Interest. Johns Hopkins University Press.
Solis, B. (2007, September). Conversational marketing versus market conversations. BrianSolis.com.