unlocking customer insights through TheyDo’s JourneyAI.

A great development is happening: more and more organizations are prioritizing customer feedback. They’re wisely combining quantitative questions—like NPS or CES scores—with qualitative open-ended questions. This dual approach promises to deliver insight into the *what*, but also the crucial *why* behind their experiences.

But here’s the often-unspoken reality: this goldmine of qualitative data collected rapidly turns into an overwhelming mountain. As surveys multiply and comments stream in, manually interpreting this context-free text into insights and connecting these to specific customer journey moments becomes a huge, time-consuming challenge. Consequently, many teams revert to reporting purely on the quantitative data, leaving that immensely valuable customer input about the “why” untouched. Customers put in the effort to provide this feedback, only for it to be overlooked—a practice far from customer-centric. Does this scenario sound familiar?

3 tips for JourneyAI success

1. Segment your data for targeted input.

Leverage existing segmentations in your qualitative data to limit your upload to the most relevant data and give AI more targeted analysis instructions. For example, open-ended follow-up questions to NPS scores naturally separate negative and positive feedback. Open responses to low scores identify key pain points and unmet needs, while responses to high scores reveal important gains.

By separately uploading the open answers of only the highest and lowest scores and providing AI with clear guidance on how to interpret the uploaded data; extracting pains specifically from negative feedback and gains specifically from positive feedback, you gain some more control over creating relevant and manageable output.

2. Start with an expert-guided foundational journey

Use your deep organizational and product expertise to build a preliminary customer journey map. This base journey ensures the AI builds on what you already know and guides AI to accurately align with your context when uncovering new insights.

For example, creating already building in a high-over insight about a confirmation email and one about a shipping update email helps the AI to distinguish the difference between the two and tailor analysis more precisely to your unique customer journey.

3. Design your journey to deal with ‘clutter’

Raw user feedback can vary greatly in quality and relevance. Some users can write very specific and detailed essays, while others provide a very vague answer or provide feedback on topics that are actually beyond the scope of your specific journey. To support AI in dealing with this, I discovered two elements that you can consider in your journey set-up.

Use separated insight lanes:
JourneyAI allows you to upload data to one or multiple selected insight lanes. This gives you control over where new information is added, also during iterative uploads. When separating insight lanes for pains and gains you can use this feature to ensure your new segmented positive and negative data is only uploaded to the right insight type. This is especially helpful when the data quality of your segmented data differs. For example, I noticed, people were much more detailed about what they disliked than what they liked. Creating separate lanes for pains and gains helped to prevent AI from polluting the specific pains with broader, less detailed positive data. Additionally, you can use this feature to deal with insights about recurring topics that don’t directly relate to the journey. By creating an out-of-scope lane and, in each iteration, uploading data to both the main lanes and the out-of-scope lane, you can ensure to keep irrelevant information from interfering with your core insights while still capturing them for potential future use. 

Create broad ‘insight buckets‘:
During my experiments, I noticed that AI tends to translate very generic comments—such as “it was nice” or “it was slow”—into equally vague and high-level insights. Even when I removed these insights from the journey, as I found them less interesting, they kept reappearing after each new upload. A practical way to address this is to intentionally create a few broad, high-level insights in your journey that act as “insight buckets.”With every upload, these buckets absorb the frequently recurring, generic feedback, preventing the AI from continuously generating new generic insights or contaminating more specific and actionable ones.