These days, cloud service providers have widely implemented the Pay-Per-Query Commercial Model, enabling public access to search services. Data owners utilize leased cloud infrastructure to provide data search services, with billing calculated based on the actual number of searches. Although this business model offers flexibility, convenience, and cost-effectiveness, it also faces the problem that they can obtain query results without consuming query times by colluding paid users with non-paying ones. Attribute-Based Keyword Search (ABKS), due to its access control features, can verify users who haven’t paid the fees and prevent their query behaviors. However, it also causes the cloud server to interact with the charging server before each query. To address this issue, We propose the Conditional Attribute-Based Keyword Search (CABKS) paradigm. Based on this paradigm, we have put forward a practical CABKS scheme applicable to Pay-Per-Query Commercial Model. This scheme guarantees that users can search the ciphertext only when they hold valid credentials generated by a specific server. By decoupling the access control structure from the keyword ciphertext, it effectively prevents unauthorized users from consuming cloud server resources. Our scheme satisfies the indistinguishability of ciphertexts under the Generic Group Model. In terms of efficiency, compared with other similar ABKS schemes, as the number of attributes increases, our scheme has the smallest linear growth in both time consumption. Therefore, our scheme demonstrates excellent adaptability and applicability for the Pay-Per-Query Commercial Model, making it a highly suitable choice for practical implementation.