Generative AI at the service of marketing: how to optimize content production at scale
Back to all articles
Havea, a leader in natural health, enhances its marketing with Ekimetrics’ generative AI to produce personalized and compliant content while accelerating its time-to-market.
Modernize, digitalize, optimize ROI… businesses have high expectations when it comes to artificial intelligence. For this reason, they have been investing massively in generative AI since 2023… with varying degrees of success. Only 11% of artificial intelligence projects generate a large-scale business impact.
Delivering this large-scale impact is our core business at Ekimetrics. In this context, a leader in natural health turned to us to take advantage of generative AI to accelerate its digital transformation and support its growth. It wanted to strengthen relations with its longstanding pharmacist clients and explore new channels such as e-commerce. The fruits of a close collaboration between Ekimetrics’ and Havea’s teams and developed in only three months, our generative AI solution made it possible to massify the production of high-quality, large-scale marketing content. It was the topic of a round table with Pierre Biousse, Head of GenAI offer at Ekimetrics, and Michaël Cosentino, Strategy and Business Development Director at Havea Group.
Why did you develop a generative AI use case for the mass creation of textual and visual content?
- To accelerate the digitalization of our brands, better meet consumers’ expectations, and create a competitive advantage.
- In the context of internationalization of the group, to accompany the deployment and penetration of new markets while adapting to the cultural and regulatory constraints specific to each region, thanks to personalized content.
- To accompany the explosion of the number of consumer/partner touchpoints while strengthening marketing capacities in a complex context of multi-brand, multi-audience, and multi-digital touchpoints while respecting strict regulatory standards.
3 steps for a successful generative AI project
- Identify the priority use case: This process can be long. Two essential questions need to be asked:
- What is my main business objective?
- What critical business process can AI most help transform to get maximum value?
An in-depth audit of all the business processes and their associated impact allows us to map all the use cases and select the one that will bring the most value to the company.
- Put in place a robust data governance to reinforce the quality of data, determine its ownership, and ensure the long-term viability of the solution. It’s as much a question of securing and structuring access to the data as derisking the accuracy and reliability of the outputs produced.
- Carry out in-depth change management: As AI projects are transformative at every level of the company, ensuring both the adoption of the tool and the new content production process is a key lever for success. Combining a program of change management and intuitive UX is essential to maximize the adoption of a generative AI project for large-scale content. The co-construction approach between Ekimetrics’ and Havea’s teams has been at the heart of this project. Ekimetrics has orchestrated a proactive and targeted upskilling of teams, ensuring leaders’ engagement. Together, the two partners have created a community of champions, mobilized to accompany the operational teams and guarantee a smooth and sustainable adoption of the solution.
4 key points before starting:
- Don’t underestimate data governance.
Guaranteeing the reliability and compliance of the content, particularly in a sector that is subject to strict regulatory requirements, like the health industry, data governance is a foundation that makes it possible to:
Define data ownership: Clearly identify those in charge of data management and empower each stakeholder to guarantee rigorous structuring in line with regulatory standards.
Guarantee the quality and relevance of the content: the product database must be 100% reliable to avoid errors in the generated content; this implies making the information reliable in advance by training those in charge of the database.
- Ensure the reliability and compliance of the generated content thanks to technical levers and human post-event monitoring.
Several prerequisites are essential to do this:
- Intent routing optimization
The content creation tool is calibrated to understand the context of the questions.
It directs the answers to the relevant databases, reducing the risk of errors or confusion between different pieces of information.
- Database segmentation
The databases are segmented to avoid hallucinations between different sources.
Each use case is connected to a dedicated database to guarantee accurate and relevant answers.
- Creation of specialized agents
Expert agents are created for each specific domain: product, active ingredients, and pathologies.
Each agent uses business rules and specific data to generate content that is accurate and compliant with regulatory requirements.
- Integration of prompt libraries
Prompt libraries help users generate specific content (e.g., for social media or product descriptions.
They also display source references to allow quick and efficient verification of the generated information.
Even when robust governance and technical tricks are deployed, human follow-up is necessary to validate the generated content and to guarantee its accuracy, relevance, and compliance. Human monitoring is essential to avoid errors in a highly regulated sector like the health industry. The “machine” doesn’t do everything.
- Plan for corrective maintenance and follow-up after deployment.
A rigorous run must follow the initial build to maintain consistency and effectiveness over time.
- Don’t neglect the adoption of the solution within teams.
If the audit and technical development steps can be performed in record time – three months between the identification of the use case and delivery of the tool in this case – the real challenge lies in the adoption of the solution within teams. It requires evangelization and targeted training programs, such as prompt training or training in generative AI tools like 3D image creation.
The benefits of a successful generative AI project:
- Increased productivity and quality across the entire value chain of marketing content production by accelerating certain steps, allowing teams to concentrate on steps with higher added value.
- Acceleration of time-to-market from ideation phases for new innovations to textual and visual marketing asset production.
- Enhanced consumer service thanks to the rapid generation of a first draft of replies to client requests (product information, advice), allowing a reduction in response time and improving consumer service satisfaction.
- Digital transformation: Generative AI is a real lever to accelerate a company’s digitalization and increase its capacity to meet consumers’ expectations.
