Sentiment analysis of online reviews to validate survey insights for an SaaS company

  • Triangulated unsolicited user feedback with internal survey findings

  • Identified gaps in perception between survey and review data

  • Reinforced strategic decisions with data-driven validation


CLIENT CHALLENGES

  • A SaaS company wanted to assess user satisfaction in a competitive and evolving market
  • Internal surveys captured general sentiment and feature feedback, but lacked unbiased, real-world perspective
  • The goal was to validate internal insights, uncover blind spots and enhance the roadmap for product development and customer experience improvements

OUR APPROACH

  • Step 1: Data collection and processing
    • Extracted five years’ worth of publicly available online product reviews

    • Cleaned and structured data using natural language processing (NLP) techniques for further analysis

  • Step 2: Sentiment and theme analysis
    • Applied lexicon-based techniques such as VADER and SentiWordNet to classify sentiment as positive, neutral or negative

    • Used topic modelling (LDA) to extract core themes to identify key drivers of satisfaction and dissatisfaction

  • Step 3: Survey validation
    • Conducted a short follow-up survey to cross-reference top issues and validate sentiment proportions

    • Compared insights from online reviews with survey responses to assess alignment and misalignment

IMPACT DELIVERED

  • Validated alignment between internal survey insights and external review sentiment
  • Identified critical areas such as “performance reliability” and “customisation options” requiring product improvement
  • Equipped stakeholders with clear, triangulated insights to prioritise user experience enhancements
  • Supported data-driven decisions in roadmap planning, reducing reliance on single-source feedback
Thank you for sharing your details

Your file will start downloading automatically

If it does not download within 1 minute,

Share this on

What we have done

Web traffic and engagement diagnostics for an online education platform
What we are proud of

Identified key content formats and channels driving high-intent learners

Validated survey insights through web traffic and behavioural analysis

Cross-platform validation of social media metrics for a personal-care brand
What we are proud of

Identified optimal content formats and ideal posting times based on both data sources

Verified that self-reported behaviours were consistent with actual social media engagement

Decision tree analysis for a consumer credit firm
What we are proud of

Mapped key decision factors influencing credit card adoption

Optimised marketing efforts through data-driven targeting

Van Westendorp pricing analysis for a Canadian health and wellness brand
What we are proud of

Identified the optimal price range for a new organic supplement

Balanced affordability perception with premium brand positioning

Conjoint and latent class analysis (LCA) for a Southeast Asian (SEA) smart wearables brand
What we are proud of

Segmented users into four distinct buyer personas

Achieved 27% uplift in product preference alignment across SEA markets

Kano and factor analysis for a European online fashion
What we are proud of

Enhanced customer satisfaction

Simplified survey data

Maximum difference scaling (max-diff) analysis for a Lactam insurance provider
What we are proud of

Identified top value drivers for insurance plan selection

Improved plan uptake by 17% across target LatAm markets