Doxa’s AI Edge: Driving Company-Wide Transformation with Caplena

Pictured, from left to right: Susanna De Luca – Director Ipsos Doxa, Chiara Munari - Data Scientist & AI Specialist, Erica Ravarelli – Data Analyst

Founded in 1946, Doxa is Italy’s first market research company and one of the first in Europe. They strongly rely on sector expertise, with dedicated teams to specific industries, offering a wide portfolio of solutions such as brand tracking, product testing, customer experience surveys and social listening. In June 2025, Doxa joined the Ipsos Group: Ipsos Doxa was born in Italy. 

What started as an initiative to leverage machine learning for social listening and web insights has grown into an internal AI powerhouse. Caplena sits at the heart of Doxa’s text analytics workflow across multiple business units.  

Discover their journey.  

Highlights 

• From Weeks to hours to build and update complex topic collections 

• From rule-based to AI categorization, with significant improvement in coverage 

• A hybrid supervised + unsupervised approach powering more accurate, scalable analysis 

• Application to large projects at scale (up to 200K verbatims in a single project) 

• Adoption from one team to the entire company, including international teams 

• New use cases across ad hoc, tracking studies, and social listening

"

Caplena helped us become pioneers, guiding clients toward a more modern, scientific and efficient way to analyze data."

Susanna De Luca

Director Ipsos Doxa 

The Challenge 

Doxa’s journey with text analytics began back in 2017 long before AI became mainstream. One of their largest clients asked them to analyze huge volumes of open-ends and feedback from emails, complaints, WhatsApp messages, NPS responses and more. 

At the time, the team was building rule-based taxonomies with a text analytics tool.

While precise, it was painfully time-consuming. Setting up an initial taxonomy took weeks (sometimes more than a month) and every update required manual intervention. Coverage remained limited, especially for “leftover” verbatims that didn’t fit neatly into rules. Nothing scaled. 

Clients wanted speed and depth. Researchers wanted flexibility. The market was evolving quickly. But AI tools at the time weren’t user-friendly enough for large teams. 

Doxa knew they had to move to a more modern, AI-driven approach, without losing the control they needed for high-stakes clients. 

Transition to Caplena 

The shift began in 2022 when Doxa discovered Caplena during an ESOMAR webinar. After completing a full evaluation of Caplena against several competitors, the team saw a clear difference.

"

It was the start of our ‘love story’ with Caplena. After years of suffering with rule-based systems.”  

Susanna De Luca

Director Ipsos Doxa 

Caplena uniquely combined: 

• The freedom of unsupervised discovery 

• The precision of a supervised, researcher-guided approach 

• A friendly interface that researchers could use easily 

• High accuracy even on messy, multilingual datasets 

• A team that evolves quickly and listens closely 

"

Caplena gave us the scientific blend of supervised and unsupervised AI we needed — powerful, flexible, and fully aligned with how researchers work.”

Chiara Munari

Data Scientist & AI Specialist

How Doxa led Change Management 

Convincing clients and teams to migrate meant reshaping expectations. Some feared losing granularity as Doxa moved from deep, multi-level taxonomies to a more modern structure. Manual coders had to adjust their mindset.  

To build confidence, Doxa ran side-by-side comparisons of manual coding versus Caplena, and the results made the case on their own. 

Coverage improved. 
Accuracy maintained. 
Time dropped dramatically. 
And enormous projects suddenly became achievable. 
 
While there was also a fear of being replaced by the machine, the most resistant manual coders became convinced that using Caplena was easier and quicker. They now have an interface giving them control over what they are doing. They can go back, check, organize, see the results, see an AI quality score. This level of flexibility couldn't be achieved with manual analysis. 

"

Some coders were used to 4-level taxonomies and 150 codes. We had to help them change their mindset — fewer levels, fewer codes, but more accuracy and control. Now nobody miss manual tagging.”

Chiara Munari

Data Scientist & AI Specialist

Results 

A Dramatic Change in Speed and Scale 

The setup phase (building a taxonomy) that once took weeks now takes just days, even hours. Doxa processed a 210,000-verbatim project soon after adopting Caplena — something that would have been unthinkable before. 

New Capabilities for Web and Social Listening 

Doxa now analyzes everything from long-form news articles to YouTube comments or reviews. They created scripts that summarize long texts via GPT before ingestion into Caplena and built automations using Caplena’s API to ingest data and export results into visualization tools like Power BI and PowerPoint. 

From One Team to the Whole Company 

Susanna’s Web & Social Listening team was the first to adopt Caplena. In 2024, the company decided to share the knowledge of Caplena with the other business units.  

Doxa was running a lot of questionnaires. Open-ended questions were only partially analysed because usually there was no time to do it. When time allowed, analysis was done manually.  

Chiara joined the internal AI team and began training researchers across FMCG, Finance, Media, B2B, Innovation, Brand Tracking, as well as the international BVA Family. 

Beyond Coding Specialists: Experience Shared Across Teams 

Some Business Unit representatives who had been curious to explore open-ends coding took the opportunity to try analyzing text hands-on for the first time.  

It was an interesting shift, as they are close to clients and understand their needs well. After some training, they quickly saw how Caplena's powerful yet intuitive platform could support their work, allowing them to run simple analyses on their own and share initial insights directly with clients, while coders continued to lead the main process. 

"

At first, researchers weren’t used to coding or working this way. But once they learned Caplena, they were thrilled at how fast they could work and how quickly they could deliver insights.” 

Chiara Munari

Data Scientist & AI Specialist

A More Modern, Insight-Ready Workflow 

Caplena’s new AI topic assignment and Insight Agent have helped researchers speed up sense-making and storytelling. Instead of manually validating everything, teams focus more on interpreting patterns and generating insights.   

A Competitive Edge in a Crowded Market 

In an era where survey platforms increasingly bundle “quick and cheap” auto-categorization features, Doxa needed a more scientific, supervised approach that clients could trust. Caplena gives them exactly that: a level of precision, explainability and flexibility that embedded tools can’t match. 

"

No other platform offers the same 360° mix of supervised control, unsupervised discovery, scalability and user-friendliness. You always evolve. You listen to us. This is why we stay."

Susanna De Luca

Director Ipsos Doxa 

Future Direction 

Caplena can potentially play a strategic role in the next phase of Doxa's insight operations.  

Above all, Doxa is committed to staying ahead of the curve.  
They see Caplena as a key partner on that journey. As the market evolves at record speed, visibility (and input) into Caplena’s roadmap will help Doxa and Caplena stay closely aligned and innovate together. 

"

Caplena has changed our way of working. Beyond saving money and time, our first benefit is to rationalize our processes, increase accuracy, and propose a modern approach as a stakeholder for our clients.”  

Susanna De Luca

Director Ipsos Doxa 

A “love story” built on precision, partnership, and a shared commitment to raising the bar for insight quality, with humans in the driver’s seat.