Shailee Shah

Here’s a fun project 👇 I connected Claude with Zomato’s live data to identify supply gaps and uncover opportunities for new restaurant entrants. How the location data works — not pin codes, but address IDs. Zomato's API doesn't work by pincode. Instead it uses specific saved delivery addresses from your Zomato account. Each address has a unique ID, and the API returns restaurant results serviceable to that exact delivery point (latitude/longitude based). What will be accurate is serviceability per address, restaurant ratings and vote counts, menu prices, and ETAs. These are all live, real data points. What won't be accurate is market-wide demand. The restaurant results only tell you supply — which restaurants can deliver where. They don't tell you how many people are actually ordering, which cuisines are most popular city-wide, or what the revenue potential of each zone is. That data lives inside Zomato's internal systems and isn't accessible via the MCP. Bottomline - If someone is making a business decision — like whether to open a cloud kitchen — the supply gap analysis is a useful real starting point. However, data via MCP wouldn't tell you about demand. For that you'd need Zomato's business analytics platform or third-party market research. #MCP #insights #claude #marketresearch

45

5 Comments

Did you know that Google Analytics GA4 can automatically convert currency values? Here is a post touching on the challenge I faced and the discovery which allowed for a simple solution.

6

It is wonderful to see measurement and marketing science become more mainstream in advertising. Modern measurement brings MMM econometrics thinking, database marketing experimentation rigor, and digital marketing attribution speed together for a unified and more accurated-together system!

22

In the past couple weeks, I read a post on Avinash Kaushik's blog about the shifting from SEO to AEO (Answer Engine Optimization). I was really curious about this, so I investigated a bit.   Google affirms that the AI Overviews are driving higher quality clicks to websites - https://lnkd.in/dZty8T_q. Although they make this claim, media and publishers are going through really tough times regarding the demonetization of their pages. The traffic volume is going down and they attribute this to AI completely taking the traffic awayy from their pages - https://lnkd.in/dr3tE77F - Pew Research confirms this, although Google claimed that the base of research done by Pew isn't sufficient to draw ant conclusion - https://lnkd.in/dyTUhA_a. The past week, I had the opportunity to hear firsthand from a publisher based in Portugal that they are facing difficult times. Their traffic levels are decreasing, and producing content is becoming more difficult because of this. But if we don’t have content, we don’t have access to information and therefore, no AI 🧐   There must be a change in this matter so that media and publishers can thrive by writing their content but this content must be correctly handled by AI. Some changes are also being made on Channel Grouping in GA4 data. But it's not enough. What can we do now? Follow Carl Fernandes's guidelines - https://lnkd.in/dDbpTCMv and build strong measurement governance so we can track our business and make the right decisions. We are facing a fast-paced environment, and the best we can do is have solid foundations that can handle fast changing times 🤠

7

In US hospitals dont get paid on the number of days a patients stay in the hospital but based on the cause of admission. this can be a good use case for hospitals and even add Gen AI to do data anaylsis what type of patients stayed longer and what measures can be taken to provide those service early

6

GA4 isn’t just analytics. It’s the new chase. In the twilight, analytics looked like Tom & Jerry — endless pursuit, no catch. Vanity metrics entertained but rarely built credibility. At dawn, GA4 rises hand‑in‑hand with AI (Recruiter AI Orientation — sourcing, screening, and language tools). Funnels don’t just report activity; they drive outcomes. Measurable impact has become the new credibility. By the zenith, achievement orientation defines the winners. Skills, foresight, and outcomes are the new currency of trust. Recruiters already reflect this shift — moving toward skills‑based hiring, quality of hire, and data‑driven impact. GA4 + AI isn’t optional; it’s the language of digital credibility. TIP means business. #MATCHWINNER 🏆 #Intent #Impact

6

📊 Understanding Signal vs. Noise — Key Takeaways from Dr Simon Jackson's Webinar Today, I attended a highly insightful session on experimentation and statistics, and it helped clarify some powerful concepts that guide decision‑making in data-driven environments. Here are my top learnings: 🔹 Signal vs. Noise in Experiments Every experiment tries to separate real impact (signal) from randomness (noise). Three factors influence this balance: • Sample Size — smaller samples create more noise. • Observed Effect — larger differences are more likely to be true signals. • Variance — higher natural variation increases noise. 🔹 P‑Value Simplified I loved the intuitive breakdown: p‑value is the probability that what you’re seeing is just noise. • A p‑value of 0.9 → 90% chance the difference is noise. • A p‑value of 0.1 → only 10% chance it’s noise; the effect may be real. • P < 0.05 → our standard for calling results statistically significant. 🔹 Common Experimentation Pitfalls • Novelty effect — initial surge in metrics due to user curiosity, not actual impact. • False Positives — mistaking noise for signal. • False Negatives — missing real improvements. 🔹 Power Analysis: Planning the Right Sample Size Learnt how sample size is influenced by power, effect size, and variance — all crucial to detect meaningful signals reliably. 🔹 Test Duration Matters Experiments may need different durations depending on whether user participation is consistent or variable across time. Overall, this session reinforced the importance of designing experiments thoughtfully and interpreting results responsibly. Data doesn’t just tell a story — we must ensure it tells the right one. #DataScience #Experimentation #ABTesting #Analytics #ProductInsights

22

4 Comments

Leading Product Analytics from last 10+ years, I have seen these Product Analytics Myths That Hold Teams Back Product analytics isn’t failing because of tools. It fails because of how we think about it. A few common myths I see repeatedly: • “Tracking events = analytics” Events without context don’t explain behavior. • “More data means better insights” Focus beats volume. Clarity beats dashboards. • “Analytics is only for PMs” The real impact comes when design, engineering, growth, and leadership use it together. • “Dashboards tell the full story” They show what happened — not why. • “Analytics = reporting” Reporting looks back. Product analytics should shape what you build next. • “We need perfect data first” Waiting for perfect data delays learning. • “Set it up once and forget it” Analytics should evolve as your product and users do. The real value of product analytics isn’t in charts — it’s in better decisions. Curious: which of these myths have you seen most often in your teams? #productanalytics

42

5 Comments

A must read to stay ahead of data insights on trends and hot topics in D&A and AI - terrific work from Gartner colleagues Anurag Raj and David Pidsley.

7

What if shopping didn’t require typing, reading, or navigating complex apps? Excited to share Vaani — Meesho’s voice-based shopping assistant, built to make e-commerce simpler, more natural, and accessible for the next 500 million users in India. #ai #meesho #bharat

81

1 Comment

📢 Amazon Prime Video Introduces Ads in India via Amazon DSP – A New Era for Streaming Advertising Starting June 17, 2025, Amazon Prime Video will incorporate limited advertisements into its streaming content in India, aligning with global trends and expanding its ad-supported streaming model. This strategic move opens up significant opportunities for advertisers, as Amazon leverages its Demand-Side Platform (DSP) to offer targeted advertising solutions. The enhanced Amazon DSP provides: Precision Targeting: Utilizing Amazon's vast first-party data, advertisers can reach specific audiences based on shopping habits, content preferences, and more. Full-Funnel Optimization: With tools like Performance+, campaigns can be optimized across the customer journey, from awareness to conversion, using AI-driven insights. Comprehensive Insights: Advertisers gain access to detailed demographic, shopping, and streaming behavior insights, enabling more effective campaign strategies. For Prime members in India, the introduction of ads comes with an option to maintain an ad-free experience by subscribing to an additional plan priced at ₹129 per month or ₹699 per year. This development signifies a pivotal shift in the streaming landscape, offering brands a unique opportunity to engage with a highly engaged audience through Amazon's integrated ecosystem. #AmazonPrimeVideo #DigitalAdvertising #AmazonDSP #StreamingAds #MarketingInnovation #IndiaAdvertising #PrimeVideoAds

6

***Want to build impeccable data sense? Consuming data passively w/o training your mind to question and understand data doesn't help. Great data sense helps you glean insights from data better, lets you focus on the big picture and helps you catch erroneous data easily*** So how do you do it? **1. Does the data point you are looking at make physical sense or not Let's say you are looking at a funnel that has 5 steps and the conversion data for the funnel is being presented as 20%, immediately you should feel that the number is most likely wrong. Why? because data cannot escape the physics of the problem. Imagine the individual step conversions of the funnel are at a super optimistic 70%, even then the overall conversion will still be (70%)^5 = 16.8%. With highly optimistic step conversions you still ended up with 16.8% what is the chance that your overall conversion is 20%...hmm...very very unlikely. **2. Always compare data to some baseline. Example - let's say some one tells you that our Mobile app funnel conversion is 9%, now how do I reason if this is good, bad or ugly? Well, the way to understand this data is to compare it with another number which makes sense. Eg. you can compare the mobile app conversion data with Desktop (DT) conversion data, since mobile is a smaller form factor you would expect App conversion to be lower than Desktop conversion. If say your DT conversion is 11% and App conversion is 9% you are doing relatively well since you are sitting at 0.8x of DT if your DT conversion instead is 15% hmm not so well. 3. If you don't have granular data to compare with, you should still be able to reason about a data point by comparing it with **macro numbers** you know. Eg. let's say you read somewhere that a food delivery company's ARPU/month is Rs 375. Do you think this number makes sense or not? Let's say applying pareto you have a sense that 30% of a food delivery company's MAU (call it M) places order and the avg monthly order from this base is let's say 1.5 (bookers place orders 1.5 times on an avg in month). Again based on your own personal experience you know roughly order size/order is Rs 700-750 so to speak. So ARPU becomes = 0.3M * 1.5 * 750/M = Rs 337 which is similar to what is quoted, so the number quoted makes sense. In fact by reasoning like this you also now understand the imp levers that drive this number. 4. Always consume data with an active mindset. If some one quotes to you GDP of India is 4 trillion $s, immediately your mind should go to oh what's a good baseline to compare it with - what is China's GDP? What is GDP of some developed countries like US. You should do mental maths like - hmm..what does per capita GDP look like - with 1400 million population of India. Follow me for high quality content, not product philosophy but practical insights - https://lnkd.in/gHweK_q7 AND also here https://lnkd.in/ggUKftFq

36