Gnanagurubaran Venkatesvaran

Google has been showing far fewer AI Overviews in the Google Search results. "Google has gradually been reducing AI in search results from 84% to under 15%." The data is in the comment section. #google #ai #aioverview #searchresults #googlesearchresults #data

٨

٣ تعليق

Having an increasing number of power users is the dream for companies. But most of us use DAU/ MAU to calculate engagement rates, which blurs out the variance. A better way to go about it is the L30 or L7 (a term coined by the Facebook Growth team)- where you can decide the action you think is important for your app/ product's growth and track its activity for 30 days or 7 days. This is based on what is most relevant for your product. See if the engagement gets better over time or during some part of the month. And then map out if it was triggered because of an experience or feature launch. This helps track actions that lie deeper in the funnel rather than just app opens. The thing to figure out here is whenever through the 30 days of the month your users are most engaged- are you able to monetize them well?💲 - Do you have core activities you want your users to perform? Once you have this answered, you can then run a combination of say, walkthroughs, contextual in-app messages or spotlights, challenges, or even a loyalty system that gives them points when they perform the core action. Think about forming re-engagement loops. Capitalizing on international events like the world cup often also plays out well. Nudge helps you sort onboarding, activation, engagement, conversion, and retention within your app and launch native experiences that help you target the core activity you want your users to perform and increase the number of power users.

١٤

١ تعليق واحد

Core algo used by most commerce apps to Up-sell: Upselling itself can be of the following kinds: 1. Quantity Upsell: Instead of buying 1 kg rice, nudge the user to buy 5 kg from this platform. This can only work for certain categories as in case a user only needs / plans to eat 1kg of tomato, it's difficult to convince them to purchase more. This hence works on things which are firstly non perishable and secondly of a nature where they are regularly used in the household. Examples of this would be Maggie, Lizol, Toilet Paper etc How: Most platforms nudge the users to buy a higher quantity SKU by ensuring offers on such pack is better than that of lower SKU. The apps also ensure this is communicated very clearly to the user, as can be seen in the image. When the users observe very clearly that on a 1 kg pack they get no off but a 5 kg pack they save 6%, they end up making the choice which maximises their savings 2. Brand Upsell: Instead of buying Nescafe coffee, the user can be nudged to buy Davidoff. This is much harder than Quantity upsell as it's nudging the user to try an entirely different and more expensive brand. How: A lot of platforms do this via either running a free trial or by brand specific offers combined with recommendation i.e. the TRI method as mentioned below: Step 1 - Trial: A user who has always had Nescafe needs to experience the richness of Davidoff. That's the first step In even building the need/desire for the product. Step 2 - Recall: Post this, placing the product in the right places in the user journey e.g. higher up in there Search results will help in building the recall. Step Step 3 - Incentive : Finally some kind of offer which is only applicable for this brand will go far into making the purchase actually happen 3. Quality Upsell: Mostly applicable to FnV(Fruits and Vegetables) , this is the nudge towards organic, homegrown or fresher produce. How: Giving a lot of information about how fresh/healthy this alternative is. It's also necessary to have the right placement on such products to ensure they get the needed visibility. Again as can be seen in the image, on the search of 'Tomato', the results show me organic and hybrid on the top, followed by desi. #product #commerce #productmanagement #pricing #innovation

٢٧٢

١٦ تعليق

Google is rolling out a new report called 'Advertising segments' in #GA4. Advertising segments are GA4 audiences that are synced with Google advertising products (like Google Ads) for remarketing and ads personalization. The Advertising segments report shows list sizes i.e. how many users are available for ads personalization in the last 30 days. The sizes of advertising segments may differ from the number of users in GA4 audiences used for measurement.

٥

٢ تعليق

Google Search Console (GSC) data hasn't been updated for the past 05 days. While it's normal for occasional delays, this seems like an unusually long period. Google, this issue needs to be fixed soon. #google #gsc #data #searchconsole

١٥

٤ تعليق

🚀 Wanna craft Google Ads that Convert Like Magic?! 💰✨ Well, crafting Google Ads is an art. It an art of reading people’s mind and then generating a compelling copy that turns clicks into conversions. Don’t know how to do it? Let me set a roadmap to creating Google Ads that work like a magic. I am going to give some tips. Well, they are more like hints, the rest you’ll have to figure out. So, here is what to focus on while creating a guaranteed selling copy: - Laser-Focused Keywords - Compelling Ad Copy - Irresistible Call-to-Action (CTA - Landing Page Magic - A/B Testing - Ad Extensions - Mobile Optimization And don’t forget you are not just trying and getting clicks – your focus is to convert those clicks into valuable actions. 🌟💼 #GoogleAdsConversion #DigitalMarketingMagic #AdvaSpark

١

١ تعليق واحد

I am sure most of you must have read about this on X, Reddit, and countless news amplifiers on the internet and social media. Having read about the issue, here are the key learnings- Responsible AI should be baked into every step of the process, and I am confident that Google is committed to this while building Gemini et al. However, whenever we use Retrieval-Augmented Generation (RAG) systems to pull in external information and use it as context in Large Language Models (LLMs), it changes the game significantly by adding an unknown variable. Quality checks of the retrieved content are often overlooked. To address this, it's crucial to establish a contract distinguishing serious from non-serious content. When building a RAG application, filtering the retrieved content should be a top priority. This involves taking a second opinion to verify the reliability and seriousness of the information. Even when your knowledge base or vector database contains high-trust content (like an enterprise repository), you might avoid such issues only until someone internally corrupts the content as a joke. A rudimentary way to mitigate this is to use the LLM itself to create an agent that scans the content, ensuring it adheres to the serious content contract. I am sure our friends at Google are coming up with even smarter and more efficient ways of achieving this (Agentic systems are good but present challenges with latency, which is critical for operations like Google Search). Also, filtering is not enough. Funny content can still sneak in, so it is equally important to build an observability stack that notifies you before your customers. One simple mechanism is lazy probing of generated content using another agent, where latency is not a constraint. This will at least flag that something is not working as expected, and appropriate mitigations can be done. Lastly, there is a need to add the AI muscle to your respective application security teams and give them skill set to think about the pitfalls of not only the API behavior but also about nuances due to data. I have written plenty about red teaming in the past, which is a good start in this direction. Along with red teaming, roles like security engineers and pen testers also need to be given necessary tools to manage GenAI systems. In short, just like you wouldn't take culinary advice from a prankster suggesting glue for pizza, we shouldn't let our AI models fall for jokes either. Let's build AI smart enough to spot a prank when it sees one! #ArtificialIntelligence #MachineLearning #AI #ResponsibleAI #AIEthics #AIDevelopment #AIQuality #RAG #AIApplications #ContentModeration #AITrust https://lnkd.in/gU3dae5W

٢٣

The last 10 years of business transition - Phase 1 - “Amazon for X” Phase 2 - “Uber for X” Phase 3 - “CloudSaaS for X” Current phase - “AI for X” What remains constant throughout these phases are- 1. Every phase has unlocked higher productivity & efficiency for X. 2. People who can understand and articulate the value addition of each phase can thrive. 3. Domain expertise of “X” is becoming increasingly less of a deciding factor for career progression. #AIforSales #AIforB2BSaaS

١٩

As an analyst, how can you make a marketing team happy? Solve attribution for them and give them all the data they lost because of consent. Come on, let's be realistic; no one can do this. But you can do something that still helps them a lot. Let's assume the marketing team usually comes together every Monday to check the current campaign performance—or every two weeks or once a month. During this meeting, it's a waste of time if all data is scattered across different platforms and needs to be aligned first. The better version is a centralized report with all campaign data, one or two attribution models, funnel performance for all campaigns, revenue impact, and finally, the Return on Ad spent. This sounds good and is a valuable asset. It takes much work to build. It's often done manually in a spreadsheet with much effort and usually copy number errors. You can try it in GA or Looker Studio, but frequently, essential things are missing (all cost data and attribution models). So - the final solution is to build it in your data warehouse. Well, that would require quite some resources. There is a different way. With Mixpanel's marketing attribution and the new ad spent data integration, you can build this report in a day. In this video, I show you how you can do this, step-by-step and with no video-friendly shortcuts. https://lnkd.in/d9zMf6ig

١٢

🤐 You can't scream at your call center Agent any more and SoftBank Group Corp. is behind it. It's AI reduces mental strain on call center staff by modulating angry customer calls. Key Achievements: → Over 100 emotional phrases trained → Significant stress reduction → Enhanced call handling efficiency Advantages over Traditional Methods: → Transforms aggressive tones into calmer speech → Detects overly long or abusive conversations → Issues warnings to prevent prolonged distress Benefits: - 🎤 Voice modulation without altering content - 🛡️ Protection against emotional burnout - 📈 Improved customer service efficiency PS: AI handles complaints efficiently, but a human touch for apologies is irreplaceable. ♻️ Repost this to spread the word about this game-changing technology. ↓ ☝️ Follow for more AI-News » https://lnkd.in/esb3PH9W ✌️ Check out our AI Workshops and supercharge your AI Game (so far 90+ successful individual workshops done for companies and corporations) https://lnkd.in/dXGDz58E 🤘 Test our Gen AI platform creamAI: http://www.creamai.de #ai #genai #callcenter #customerexperience #innovation

٢٠

٢ تعليق

1/n on Product-Market Fit (PMF) 📈 I have struggled with this in the past - Popular definitions of PMF is esoteric and elusive! ❎ People say - “You feel it when it happens”, “You know it when it happens”, “Fishes are jumping onto the boat” What does it even mean? You can use this rule to determine if your product has reached PMF ⬇️ #product #PMF #EarlyStage #Category

١٠

Metrics all product managers should know ↙ ➯ DAU, WAU, MAU. ➯ Mean & media. ➯ Sticky factor. ➯ Conversion. ➯ Retention. Make sure to save the carousel. ↳ So you can always to refresh your memory. If you want to learn about each metric in detail. Register on my website and begin learning right away. Enroll for free. (Link in the comments)

٢٨

١٨ تعليق

Let's Understand A Very Important KPI For Gaming Indursty Churn Rate Churn Rate: This is the percentage of players who stop playing the game within a given timeframe. Why is Churn Rate Important? A high churn rate indicates that you're losing customers or players faster than you're acquiring new ones. This can have a significant negative impact on your business growth and revenue

٢

This one metric which almost no one tracks, is easily you're lowest hanging fruit for increasing conversions Because there is certainly a trend between users reading your product descriptions and converting But what's more, there is also trends between opening lines, tone and content structure that impacts how many users read your product descriptions A simple piece of tracking from Descriptomizer will give you all of this information for free Reach out if you're curious to see

٦

What if I use my GA4 data in Mixpanel? There are rumors that people are not so happy with the GA4 frontends. And they might be true. My biggest issue with GA was always customization. It got better in GA4 (yes, there are things that got better), but it's still too limited for me to build powerful data user experiences for other teams. And here is the deal: Most people in your company check analytics data 1-2 times a month. They have other things to do than spend every day crunching datasets. I design the experience for them, and I want to make it as easy as possible. A marketing team might check in on initiative performances every Monday. I can build one report that works best for them in these meetings (with the right metrics and proper drill-downs). And yes, they can then go from there and dive deeper. But they need a starting point. A lot of people have posted here about how they have built custom Looker Studio reports for that. Yes, that is one way. But I miss the part that marketing can go deeper; for this, they would need different reports. So, I build a different alternative: Syncing your GA4 data into Mixpanel. In Mixpanel, I can build a custom user experience, and my users can use these to go deeper by themselves. By using GA4 data, I don't need to change my tracking implementation, and I could even filter and transform some of the events when some data is not tracked correctly. Check out the video to learn more: https://lnkd.in/dKHTHNfv

١٤

Unpopular/ Underrated KPI for 0-1 journey: Building a comprehensive knowledge base from day 0. Let me explain, Two important KPIs that most of us go after/should go after in a 0-1 journey: Building product & Building converting pipeline. Third KPI: Unpopular one- Knowledgebase updation. How well we maintain our knowledgebase, - currently we use Notion One of the weekly progress checks that my co-founder and I focus on is to try to check is how well we are maintaining the documentation of our knowledge that we are accumulating daily and organising it into our notion workspace. Huge compounding with incremental daily steps. We started off with one page, currently we have 8 pages, 6 subpages. This is one learning I picked up from Rajith Shaji during our early days of building Volopay. Im glad to have a co-founder who pushes me on this strictly. If you are already doing it, would love to know the best practices that you follow or if you know any founder/ individual whos killing it, please do tag them, Ill learn from them or steal the template from you 😄

٤٨

٢ تعليق

BASIC: What makes Advantage+ Audience different from Open Targeting? In terms of usability, they aren't vastly different except.. ➡ In Open targeting - You set minimal criteria like geographic and age range, then let Meta freely target anyone within those parameters. This approach is hands-off, allowing Meta's algorithms to operate with maximum flexibility. ➡ Advantage+ Audiences: Takes this concept further by allowing two levels of input: ↳Audience Controls: Define essential criteria such as geographic locations, age minimums, languages, and exclusion lists. These are strict requirements that Meta will honor. ↳Audience Suggestions: Offer guidance to Meta’s AI with suggestions based on demographics, interests, behaviors, etc. These suggestions are not mandatory but serve to direct the AI’s initial targeting efforts. Now, what's better? In my experience it differs from use-case to use-case, for example - - Lead Campaigns + Open Targeting = Better MQL's 📈 - Purchase Campaigns + Advantage = Better Purchase Volume 💰 BUT Really can't call one out - test test test! Thoughts? #performancemarketing #metaads

٥

Hi LinkedIn Fam! Super stoked to be launching Raijin.ai - AI Analysis Assistant on Product Hunt in 2 more days! If synthesizing tons of user interview, usability test feedback or general notes and articles is a massive pain and you want to reduce half the time your team spends on manual analysis, then you will love what we do. Raijin.ai uses LLMs and Generative-AI to automatically cluster and perform thematic analysis with high accuracy. Craziest part: we can even write a first cut of a findings report for you, based on your own data sources and unique context. It's been a fun journey building this and shout-out to some amazing friends spending time outside work to do usability tests or look at our demos for us in the last few weeks and catching catastrophic bugs before our launch: Joanna Soh, Darryl Tan, Idris Murat Atakli, Wilmer Tay, Diana Tupilus, Ju Hu Soh, Cecilia Wong 王 施 施 , Taryn D'Souza and more... Stay tuned: https://lnkd.in/gcbTZdhv #uiux #productmanager #customerdiscovery #continuousdiscovery #userresearch

٥٢

١٢ تعليق