The intention economy: why the idea matters again

AI is bringing an old question back into focus: who controls the relationship between what we want and what the market offers us? For Doc Searls, the intention economy was a project to give customers greater independence. The same term is now used to warn of a market for signals about their plans. The promise of an assistant that “understands what we want” sidesteps the question of whose interests it serves. That question also arises while we are still working out what to ask for.

What is the intention economy?

Doc Searls introduced the term intention economy in an article published in Linux Journal on 8 March 2006. He challenged a digital economy organised around sellers: attract people's attention, then try to turn them into buyers. His proposal gave customers the means to state what they were looking for and invite competing offers. Suppliers would respond to demand expressed by the buyer (Searls, 2006).

His 2012 book, The Intention Economy: When Customers Take Charge, developed this vision. Customers would have their own tools for managing supplier relationships and the information they shared. They would set the terms on which they were willing to engage. That independence would also make it easier to switch suppliers without having to rebuild their customer identity from scratch.

The term describes a proposed way of organising markets. Searls sets out the tools and relationships that would give buyers more control over exchanges. A company gaining a better understanding of its customers serves a different purpose: improving its own ability to sell.

What changes when buyers state their intentions?

Consider a hypothetical procurement manager looking for a maintenance contract across several sites. She knows which equipment needs cover and requires a defined response time. Following Searls' approach, she sends her requirements to several suppliers on terms she chooses. They receive the information they need to prepare their bids, and she compares their commitments.

The quality of this exchange depends on how much control she retains over the information she shares. She decides which suppliers can access the equipment details and for what purpose. A request for a quotation opens a negotiation with a defined scope. If those details then feed a commercial profile used elsewhere, the recipient has put the information to a use the buyer may never have intended.

Searls' original proposal rests on the customer's ability to organise the relationship from their side of the market. Detailed information helps suppliers respond when it circulates on the buyer's terms. They compete through the offers the buyer receives and assesses.

Why has AI brought the idea back into focus?

Writing in Harvard Data Science Review in 2024, Yaqub Chaudhary and Jonnie Penn describe an emerging market for signals of intent. A conversation with an assistant reveals something about a user's plans and the reasons for their hesitation. The authors examine the scope for collecting and trading that information, and for using it to steer choices. Their account draws on technical research and industry announcements to examine how such a market could develop.

A conversation takes place while the request is still taking shape. Someone starts with a problem, answers questions and weighs the proposed solutions. The assistant contributes by reframing the problem and making some options more prominent. Whoever designs that exchange has a role before the person has settled on a choice.

Chaudhary and Penn (2026) develop this argument further by examining forms of power associated with conversational interfaces. They stress that the interaction runs both ways: systems also prompt users and guide the discussion. Adapting their language to the individual gives their designers additional ways to influence how a proposal is received.

The connection with cognitive economics lies in this contribution to reasoning. A tool that prepares a comparison helps select the criteria and present the information. Alongside an answer, the user receives a way of approaching the decision.

Two visions, with different answers to who controls demand

On 30 December 2024, Searls responded to Chaudhary and Penn in a post titled The Real Intention Economy. He challenged their use of his term for the threat they described and asked them to choose a different name. He pointed to years of work on tools designed to serve customers. His objection concerned the project the term stood for, and who would exercise control within it.

Chaudhary and Penn acknowledge earlier accounts that favour consumer independence. Their analysis examines another possibility: companies exploiting signals of intent collected through their own infrastructure. Both visions recognise the value of this information, but they envisage different relationships between the people who generate it and those who use it.

[[INSERT TABLE 1 HERE: Two visions of the intention economy]]

Casting the debate as the attention economy giving way to the intention economy obscures the disagreement between the authors. Gathering signals about people's plans also extends existing targeting practices. The technical change matters where it alters either party's ability to shape the exchange and influence the decision.

What does the evidence show about conversational persuasion?

Salvi and colleagues (2025) studied online debates involving 900 participants assigned to different experimental conditions. Participants exchanged arguments with either a human or GPT-4, with or without their opponent having access to sociodemographic information about them. In the personalised condition, GPT-4 secured greater post-debate agreement than human opponents in the baseline condition. The experiment measured changes in opinion following short exchanges.

A correction published in September 2026 clarified an important finding: the direct comparison between personalised and non-personalised GPT-4 did not reach the study's threshold for statistical significance. The observed advantage over humans therefore cannot be attributed exclusively to personalisation. The findings support the system's persuasive capacity in these debates while clarifying what the researchers can say about the contribution of personal data.

Turning that capacity into a market for intentions requires further conditions, including access to data and commercial agreements between organisations. A persuasive system provides a means of influence. Whose interests it serves depends on its objectives and the choices made by those who deploy it.

An inferred intention is still an interpretation

Chaudhary and Penn (2024) also question the assumptions involved in treating intention as something that can be computed. Conversational traces are used to assign categories and make predictions. What those traces mean depends on modelling choices. Classification turns selected clues into a representation of the user's plans.

Asking what a service costs may be part of an exploratory comparison, with no decision to buy. People also consider options they later abandon through discussion. Treating that exploration as a firm commitment turns a hypothesis about their plans into an apparently settled request. The precision of the profile then conceals what the person is still trying to work out.

Useful assistance makes room for revision. Users clarify what they want to pursue and correct the intentions attributed to them. This connects with the debate over libertarian paternalism: steering a choice in someone's interests also raises the question of how that person defines those interests.

What organisations choose to delegate

In the procurement example, the tool prepares a comparison of maintenance contracts. Whoever oversees its deployment chooses the sources it consults and the default criteria. If annual price determines the ranking, a cheaper offer comes first even when its geographical coverage is a poor fit for the sites. The settings embody a trade-off before the procurement manager reviews the result.

The question of intention then becomes part of how the organisation makes decisions. The buyer has a professional objective, her employer sets constraints, and the tool's supplier has its own business model. The task is to identify potential conflicts between these interests and establish how they are handled. A recommendation paid for by a supplier is a different kind of advice from a comparison based solely on the buyer's criteria.

Leaders deploying an assistant can give professionals the authority to change criteria and challenge recommendations. They also establish how those corrections reach the people responsible for the tool. Control is exercised in these everyday decisions, when users secure a change to a proposal or choose another way to address their problem.

The idea matters again because these opportunities to act are being redistributed. Assessing an assistant solely by its usage rate overlooks the trade-offs built into its answers. Adopting it also commits the organisation to rules about how people frame their requests and ensure that what they are trying to achieve is taken into account.

Frequently asked questions about the intention economy

Who introduced the concept of the intention economy?

Doc Searls introduced the term in Linux Journal in 2006 and developed it in a book published in 2012. Chaudhary and Penn adopted it from a critical perspective in 2024. Their texts describe different projects, which explains Searls' explicit objection.

Is the intention economy only about AI?

No. The original proposal concerns tools that allow customers to express their requirements and manage supplier relationships. Their independence rests on controlling the terms of the exchange. AI opens up new possibilities both for assistance and for exploiting the information gathered.

Is it a psychological theory of intention?

No. The term describes economic relationships around the expression and use of people's plans. The psychological question of how intentions translate into action concerns the conditions needed for a behaviour to occur. The COM-B model examines those conditions, including the opportunities to act provided by the environment.

References

• Searls, D. (2006; 2012). The Intention Economy. Linux Journal, 8 March 2006: https://www.linuxjournal.com/content/intention-economy. Developed in The Intention Economy: When Customers Take Charge, Harvard Business Review Press, 2012: https://www.oreilly.com/library/view/the-intention-economy/9781422184028/.
• Searls, D. (2024). The Real Intention Economy. 30 December 2024. https://doc.searls.com/2024/12/30/the-real-intention-economy/.
• Chaudhary, Y. and Penn, J. (2024). Beware the Intention Economy: Collection and Commodification of Intent via Large Language Models. Harvard Data Science Review, Special Issue 5. DOI: https://doi.org/10.1162/99608f92.21e6bbaa.
• Chaudhary, Y. and Penn, J. (2026). New parameters of power: On LLM-based manipulation and control and the spectre of strategic AI. Big Data & Society. DOI: https://doi.org/10.1177/20539517261426450.
• Salvi, F., Horta Ribeiro, M., Gallotti, R. and West, R. (2025). On the conversational persuasiveness of GPT-4. Nature Human Behaviour, 9, 1645–1653. DOI: https://doi.org/10.1038/s41562-025-02194-6. Updated following the correction published on 3 September 2026: https://doi.org/10.1038/s41562-026-02588-0.

About the author

Maxime Rabéchault is Founder and President of UNREST, a change enablement consultancy grounded in behavioural science. He applies behavioural science to facilitate organisational transformation. He also teaches at IAE Paris-Sorbonne.