
Varun S.
Marketing Data Analyst
SUMMARY OF EXPERIENCE
• ~3 years of data and business analysis experience with strong focus on sales and marketing analytics. In his most recent role he was leading data initiatives for UNICEF as Business Analyst at Millennium Organisation, optimising sales performance and outreach strategies across India.
• At Media.net, Varun analysed ad performance metrics and revenue generation across multiple dimensions, wrote complex Hive queries to extract business-critical data, and generated monthly reports for major providers whilst conducting company-wide infrastructure cost analysis.
• He has previously built AI-powered automation tools including a fully automated crypto trading bot using GPT-4 and RAG-based chatbot for knowledge management, whilst saving 6+ hours daily through automated reporting systems at Millennium Organisation.
👍 What we loved about them
• Challenges and approaches things from first principles – In the take-home exercise, the client believed call volume was declining; Varun's analysis concluded that wasn't quite right – demand shifted temporarily but didn't disappear, and overall performance was strong. We felt he had the intellectual courage to reach a different conclusion than the one he was expected to confirm.
• He thinks about how customers actually behave – When asked how he'd convince a skeptical client, his answer was grounded in the customer journey: people searched for the old name, couldn't find it, called a different store. He went beyond citing numbers – he told a story about human behaviour that made the numbers make sense. We really appreciated his ability to connect data patterns to real-world actions.
• He did extra work that wasn't required – The correlation analysis between during-transition and post-transition performance was entirely self-initiated. He wondered whether stores that weathered the rebrand well would continue to outperform, so he tested it. The log transformation to handle asymmetric percentage changes showed his statistical awareness.
• He's genuinely self-aware about his gaps – Even before we pointed out that his memo lacked data, Varun mentioned it and didn't get defensive or make excuses. He immediately acknowledged "the numbers could help" and that he'd underestimated what a founder/owner of the agency would need. Overall, we found his calm, self-assured but grounded demeanour quite refreshing.
ℹ️ Things to be aware of
• He's available to join immediately.
• He has strong technical foundations in Python, SQL, Hive, Power BI, Excel, and AI/ML tools (GPT-4, LangChain), with expertise in prompt engineering and product thinking. While he's highly competent at building dashboards, he finds more purpose and joy in analytical work that's less repetitive.
• He left his most recent role after a year because he felt his learning had plateued and his team weren't open to hearing/actioning his opinions. He’s now looking for an opportunity where his work has tangible impact, he feels more seen and heard, and he’s challenged intellectually.
💁♀️ Where he may need support
-
Although he lacks extensive experience with LinkedIn and Bing ads, his proficiency in Google ads suggests a high adaptability to new platforms.
👩💻 Technical interview performance
Objective
This candidate was invited to a 60-minute follow-up interview to assess their technical capabilities in more detail. During this interview, we assessed their critical-thinking skills, technical expertise, and overall conversational skills.
Technical abilities
• Data understanding & structuring [9/10] – His data architecture was thoughtful from the start. He merged the location file with call data, then segmented both stores (rebranded vs non-rebranded) and time periods (pre/during/post rebrand). Critically, he applied the time segmentation to non-rebranded stores as well, which allowed him to identify the demand shift pattern. While data cleansing, he caught and corrected multiple data quality issues without being prompted: city name abbreviations ("VL" for "ville", "HL" for "hill"), rebrand dates showing 2024 instead of 2025, and channel naming inconsistencies. He was one of the few candidates who noticed the August partial data issue (26 days) and factored it into his interpretation. When we asked how he would handle incomplete periods, he immediately said he'd normalize to daily averages and extrapolate, which is the right instinct for ongoing reporting.
• Numerical accuracy [9/10] – His core numbers were absolutely spot on. The GMB drop from 570 to 21 for rebranded stores, the increase from 134 to 1,573 for non-rebranded, the store-level breakdown (24/32 growth, 8 decline) – these all align with independent verification of the source data. Furthermore, he calculated per-store averages rather than just raw totals, which is methodologically correct when comparing groups of different sizes. His correlation analysis (0.78 raw, 0.36 log-transformed) was a proactive addition. He explained the log transformation rationale clearly – percentage losses are capped at -100% while gains are unlimited, so raw correlations can be skewed by outliers.
• Analytical reasoning [8.5/10] – His opening explanation in the interview laid out the core insight – calls dropped for rebranded stores, increased for non-rebranded stores, meaning demand shifted rather than disappeared. When we asked how he would convince a skeptical client, his answer was strong – he would explain the customer journey (people searched for the old name, couldn't find it, called a different store instead), show that non-rebranded stores absorbed the demand, and emphasize that customers weren't lost, they just went elsewhere temporarily. The correlation analysis was self-initiated and operationally relevant – he wanted to understand whether stores that weathered the transition well would continue to outperform afterward. When we asked about the Jun-Aug drop in calls, he immediately mentioned checking August data completeness and identified Google LSA and website as the channels that dropped. He recognized LSA declining was concerning given its importance.
• Communication & storytelling [7/10] – His verbal explanations were clear and logical. He could walk through his methodology, explain his reasoning, and answer follow-up questions without getting lost or contradicting himself. When we asked for the "bottom line" message, he was direct – "No, you should not be worried. Move to more digital marketing campaigns. Google LSA changed the game. Focus on the channels that are working." He also correctly stated that 2025 call volumes were higher than previous years when asked directly, showing he understood the YoY growth context even if it wasn't prominent in his memo. That said, his executive memo could have been better – he admitted he could have emphasized the channel shift more and included supporting data. His rationale ("I thought this was more for a founder, just to get an overall view") suggests he underestimated the need for evidence in executive communication.
• Judgment, assumptions & recommendations [7.5/10] – Varun was transparent about the assumptions he made while completing the exercise. When discussing call volume as a success metric, he volunteered that he was assuming more calls equals more revenue without conversion data, and suggested call duration as a quality indicator he could have explored. His confidence level framework was also sensible – high confidence when numbers were clear, medium when assumptions were involved, low-medium when there were multiple layers of speculation, which showed us appropriate intellectual humility. More specifically, his future rebrand playbook covered the key points: update marketing channels immediately (especially GMB), ensure you're on the right platforms (emphasizing LSA), expect some drop during transition but ensure recovery afterward.
Areas of growth
• Where he falls short is in documentation and communication as his memo lacked the data and specificity needed to make it a standalone deliverable, something he acknowledged during the interview. He has strong analytical instincts but needs to translate those insights into more polished client-facing outputs.