Agent-Assisted Revenue is revenue generated from sales in which an AI agent participated at one or more stages of the customer journey without necessarily completing the transaction itself. The agent may have helped the user discover a product, compare offers, narrow down available options, receive a recommendation, or reach the online store.
The metric includes revenue from transactions where the AI agent had an identifiable influence on the purchase, even if the customer completed the order independently on the store’s website, in an app, or through another channel.
Agent-Assisted Revenue = total value of transactions in which an AI agent participated before the purchase
For example, if a user receives a product recommendation from an AI system, visits the online store, and completes the purchase independently, the transaction may be included in Agent-Assisted Revenue.
How Can an AI Agent Assist a Sale?
An AI agent may support the shopping process by:
- identifying the user’s needs,
- searching for suitable products,
- comparing specifications, prices, and purchase conditions,
- creating a shortlist of offers,
- recommending a specific product,
- answering product-related questions,
- checking availability or estimated delivery time,
- directing the user to the online store,
- transferring a selected product to the cart,
- initiating the checkout process.
Agent-Assisted Revenue does not require the AI system to process the payment or complete the checkout. It is enough for the agent to play a meaningful role in product discovery, evaluation, or selection.
Why Is Agent-Assisted Revenue Important?
Traditional analytics may not capture the full impact of AI on sales. A user can begin the shopping process in ChatGPT, Gemini, or another AI system and later return to the store through Google Search, a direct visit, or a mobile app.
In this case, the transaction may be attributed to organic search, direct traffic, or another channel, even though the AI recommendation played an important role in the customer’s decision.
Agent-Assisted Revenue helps answer the question:
How much revenue was generated or influenced by AI-agent interactions, even when AI was not the final touchpoint before the purchase?
What Can Be Included in Agent-Assisted Revenue?
Depending on the adopted methodology, it may include transactions in which:
- the user reached the store from an AI system,
- the agent recommended a product that was later purchased,
- AI participated in comparing available offers,
- the agent transferred a product to the cart,
- the agent initiated checkout and the user completed it,
- the customer confirmed in a survey that AI influenced the decision,
- the transaction was connected to an agent through a referral URL, identifier, or platform data.
The scope of the metric should be clearly defined because not every AI-assisted sale can currently be attributed with complete accuracy.
How to Measure Agent-Assisted Revenue
Measurement may use:
- Referral traffic from AI systems.
- UTM parameters and session identifiers.
- Data from Agentic Commerce platforms.
- Information about products transferred to the cart by an agent.
- Conversion path and assisted-conversion analysis.
- Post-purchase surveys.
- Questions about the source of inspiration or recommendation.
- CRM and e-commerce platform data.
- Attribution models that include earlier AI interactions.
The result can be analyzed as total revenue, a percentage of overall sales, or revenue by AI system, product category, and stage of the customer journey.
Agent-Assisted Revenue vs. AI Referral Revenue
Agent-Assisted Revenue is broader than revenue generated directly from AI referral traffic.
- AI Referral Revenue includes sales completed after a direct visit from an AI system.
- Agent-Assisted Revenue also includes transactions where the agent influenced the decision but was not the final traffic source.
- Agent-Completed Revenue includes transactions directly carried out or completed by an AI agent.
Example
A user asks an AI agent to find a coffee machine priced below €600. The agent compares several models and recommends one of them. The customer does not click immediately but searches for the model on Google the next day and purchases it.
Traditional analytics may attribute the sale to organic search. Because the AI agent helped the customer discover and choose the product, the transaction may also be included in Agent-Assisted Revenue.
Summary
Agent-Assisted Revenue is the value of sales in which an AI agent participated at any stage of the shopping process, from product discovery to checkout initiation. The metric helps measure the broader impact of AI systems on customer decisions, including purchases ultimately completed through another channel.
