Why selling raw AI volume is a strategic mistake for an agency
Florent Dabernat
July 11, 2026
2 min read
In this article
In 2026, 88% of organizations already use AI in at least one function, but only 5.5% get real economic benefit from it. That gap reveals exactly where an agency's real value sits today, and where it no longer does.
An adoption divide in France
The French market shows a clear organizational divide: 56.6% of professionals use AI daily, but 8% of companies formally ban it internally, what's now called "shadow AI." Overall, 88% of organizations use AI in at least one function, but only 7% have deployed it at organizational scale, and only 5.5% derive a real, measured economic benefit from it.
Three strategic pivots for agencies
Three pivots have been identified for agencies wanting to stay relevant in this context. The first: working with very uneven client maturity levels, offering governance for the more cautious and orchestration for the more advanced. The second: shifting from a content-producer role to that of an orchestrator of hybrid workflows enriched with proprietary data, rather than selling raw volumes of generated content. The third: proving irreplaceable human value, sector expertise, brand consistency, strategic interpretation.
The mistake explicitly identified by this market study: selling raw AI production volume. An agency that positions itself solely on that ground gets mechanically crushed by tools available for €20 a month, which technically do the same thing without the support.
What it changes in terms of margins
Agencies that have pulled off this repositioning report gross margins of 65 to 75% on their content services, up from 40 to 50% before AI integration, and up to 85% margin on productized AI agent offerings. A shared goal among the most advanced agencies: reach 60% or more of revenue as recurring income, through monthly retainers rather than one-off projects.
A concrete example illustrates this model well: a restaurant reception AI agent, sold for $399 a month, deployed with 35 clients, represents about $167,000 in annual recurring revenue from that single product alone. It's this kind of logic, recurring product over one-off project, that durably sets a repositioned agency apart from one still selling raw production volume.
Frequently asked questions
-
Why is selling AI content volume becoming a bad strategy?
- Because raw production is now accessible directly to the client through low-cost tools. An agency's value needs to shift toward orchestration, strategy and consistency, skills that are harder to automate.
-
What is the "shadow AI" mentioned in the study?
- It's undeclared or unmanaged use of AI by employees, in companies that formally ban it, a phenomenon that reveals a gap between internal policy and actual practice.
-
How can a small agency aim for recurring revenue rather than one-off projects?
- By turning part of its offer into a product or service maintained over time (monitoring, updates, governance of an AI-ready system) rather than a one-shot deliverable billed once.