Utilizing Customer Data Analytics

Most problems in GTM can be solved with one thing: a better understanding of your customers. Sounds simple, right? It’s not. Especially if your systems are siloed and you don’t have an analyst on your team. Here’s the good news: It’s something AI is phenomenally good at – if it has the right data. If you’ve used ChatGPT’s deep research feature, you know how good it is at analyzing vast amounts of data and producing detailed insights. If you’ve used HubSpot, you know that it unifies structured and unstructured data and has full context across the customer journey. Today, I’m super excited to share that HubSpot is the first CRM to launch a deep research connector with ChatGPT. That means you can now do advanced analysis of context-rich customer data – and immediately take action on those insights. The result? Better experiences for your customers. Better outcomes for your business. Here are some use cases: - Marketers, you could ask for a list of 'VP' and 'C-Level' leads who opened an email in the past week but have no next activity scheduled – then run a hyper-targeted follow-up campaign. - Sales leaders and reps, you could pull up a summary of why your high-potential deals are stalling – then create an email sequence that addresses the most common objections. - Customer success leaders, you could generate a list of at-risk customers and a unique renewal plan for each one – then take proactive steps to retain them. I can’t wait to see what our customers achieve with the power of deep research and unified data, right in their flow of work. If you’re a HubSpot customer with a paid ChatGPT plan (Enterprise, Team, Pro, Plus, or Edu), you can get started now!

Bronze, Silver, Gold is a solid pattern. The concern is what we put at the end of it. Veronika Heimsbakk's article on Modern Data 101 has become a quick favourite, addressing this right at the source: the medallion transformation. A Gold layer full of clean, well-modelled Parquet tables still has a fundamental limitation: relationships are outside the data. They exist in join logic, dbt models, or SQL scripts that someone wrote eighteen months ago, which break when a schema changes upstream. When a business user asks, "show me everything we know about this customer across all our systems," the answer is: go talk to the analyst who knows which tables to join. 𝐓𝐡𝐞 𝐜𝐨𝐫𝐞 𝐢𝐬𝐬𝐮𝐞 𝐢𝐬 𝐡𝐨𝐰 𝐰𝐞 𝐢𝐝𝐞𝐧𝐭𝐢𝐟𝐲 𝐭𝐡𝐢𝐧𝐠𝐬.  Every record in most data platforms has an ID that means something only inside one system. A customer_id in the CRM, an account_id in billing, a client_ref in the external registry, all pointing at the same entity with no native way to traverse those relationships without custom join logic stitching them together. We've been building pipelines that transform data structurally for decades, while the semantic layer (what things actually mean and how they relate) gets pushed off to the application layer or left undocumented entirely. 𝐆𝐥𝐨𝐛𝐚𝐥 𝐔𝐧𝐢𝐪𝐮𝐞 𝐈𝐑𝐈𝐬 What changes when you give every entity a stable, globally unique IRI in your Silver layer isn't just metadata hygiene. It's the difference between data that describes things and data that connects things. That IRI becomes a join key that works across all your systems without you having to write the join. Map your Silver DataFrames against a shared ontology in your Gold layer, publish as RDF, and the relationships are in the data itself. The SPARQL queries that become possible at that point are a different category of question entirely. Consider impact analysis, entity resolution, and provenance all in one query: "find every entity related to financial compliance that touches this customer record, trace back to source, and tell me which systems contribute what." 𝐇𝐨𝐰 𝐃𝐂𝐀𝐓 𝐟𝐢𝐭𝐬 𝐫𝐞𝐚𝐥𝐥𝐲 𝐰𝐞𝐥𝐥 𝐡𝐞𝐫𝐞 Instead of designing a custom schema to describe your datasets, you get a W3C standard ontology built for exactly this purpose. And because your catalog metadata is RDF and your data is RDF, they are in the same graph. The catalog stops being a separate application pointing at your data and becomes part of the knowledge graph itself. Let's not ask, "Should we use a knowledge graph?" More fundamentally, ask, "What does fully enriched, connected, query-ready data look like? 🔖 𝐅𝐮𝐥𝐥 𝐭𝐞𝐜𝐡𝐧𝐢𝐜𝐚𝐥 𝐰𝐚𝐥𝐤𝐭𝐡𝐫𝐨𝐮𝐠𝐡: https://lnkd.in/gSbC_Vxt #DataArchitecture #KnowledgeGraphs #SemanticWeb #DataEngineering #Medallion

One of the most practical AI use cases in eCommerce right now isn’t a chatbot or a fancy personalization layer. It’s predicting a shopper’s future LTV before you spend the budget, and routing spend toward the people most likely to buy again. This is what I learned recently from Pecan AI which is quite interesting to me. And because most teams can’t do that today, they keep allocating budget evenly and running broad promos, hoping it works. 𝐏𝐞𝐜𝐚𝐧 𝐂𝐨-𝐏𝐢𝐥𝐨𝐭 changes the workflow: • You define the goal (e.g. “Predict 90-day LTV by channel and creative”) • It builds the predictive model for you • Then outputs ranked audiences and campaigns to scale, cap, or test, pushed directly into the tools you already use (ad platforms, CRM, email) No dashboards. Just actionable predictions. 📚 𝐄𝐱𝐚𝐦𝐩𝐥𝐞 𝐭𝐡𝐞𝐲 𝐬𝐡𝐚𝐫𝐞𝐝: A DTC apparel brand had strong AOV but low repeats from a few ad sets. Pecan flagged those cohorts as low predicted LTV, capped spend, and shifted budget to a lookalike built from high-LTV buyers → ROAS went up and discount costs dropped. This is the kind of AI that actually drives growth, not just adds another layer of complexity. Demo link → https://hubs.la/Q03BJHTF0 #AI #ecommerce #predictiveanalytics #martech

I JUST looked at the data from recent customer calls, and 70% started the same way… "We need to understand what's being said about our industry/brand/competitors." Our data backs this up (Talkwalker). Mentions of "Social Listening" are up 145% year-over-year, with engagement around the topic up 84%. But this goes beyond traditional brand monitoring. The revenue leaders I’m speaking to are asking questions like: → What are prospects saying before they even reach out to sales? → How is our messaging landing compared to competitors? → What objections are surfacing in social conversations? GTM teams are starting to realize that while they're debating campaign performance in quarterly reviews, their buyers are having very public conversations about purchasing decisions… right now. So what can we learn from the world's largest focus group (5bi+ to be more precise)? Everything. Market shifts before they hit the mainstream. Competitive positioning gaps. Real buyer objections. Pipeline signals your sales team can actually act on. When we saw how GTM teams were using social intelligence we knew this was the future. The companies winning today aren't the ones talking the loudest on social. They're the ones listening the closest.

🪞 Weekly Reflection...I made a classic mistake when I first started working with customer feedback. I was so eager to get Enterpret up and running at Figma that I pulled in the easiest data sources first—surveys, social feedback, things that were quick to integrate. It felt like momentum, but I later realized I wasn’t getting a full picture of our customers. This week, I came across a diagram that I would have prevented me from making this mistake. 1️⃣ What people say (surveys, support tickets) 2️⃣ What they think and feel (sentiment analysis, call transcripts) 3️⃣ What they actually do (product usage, behavioral data) 4️⃣ Why they do it (mixing behavior + qualitative insights) Fortunately, Jack Divita encouraged us to expand our data sources. Once we layered in behavioral data and metadata, we could answer way bigger questions. Like: How much potential revenue is tied to this feature request? 💡 Lesson learned: Fast isn’t always better. When you’re working with customer feedback, prioritize the data that helps you make better decisions—not just what’s easiest to pull in. I made this mistake early on, and I still see many teams go through the same journey. If you want to shortcut that learning curve, we'd love to show you how Enterpret helps teams get to deeper customer insight faster. Book a quick demo here: https://lnkd.in/gKpVYEQ7 🙏 to Helio for the simple, yet powerful diagram, which was inspired by Hannah Shamji's original work.

You do not fix this by reopening 11 months of Delta partitions and hoping Spark finishes before the CFO asks questions. That is how you create another incident while fixing the first one. Here is a potential approach to handle this: [1] Treat it as a data incident first Before fixing tables quietly, notify consumers. Bad data existed in production. Queries may have produced wrong numbers. ML models may have trained on incorrect features. Finance reports may change after correction. Analytics, ML, finance, and product teams need the affected date range, impacted tables, and fix timeline. A silent backfill is not enough when people already made decisions from bad data. [2] Stop assuming closed means final The root bug is this assumption: “Closed partition” means “final data.” It does not. Track event_time, ingestion_time, source_updated_at, correction_batch_id, and record_version. Fresh hourly data can keep flowing through the normal path. Corrected historical data should enter through a separate backfill lane that can update old partitions safely. [3] Rebuild only the affected slice I would not reprocess everything. I would isolate the 6-week window and build corrected data side-by-side: - ingest corrected records into staging - dedupe by business key - keep the latest valid version - validate counts, aggregates, and key metrics - compare old vs corrected outputs The live pipeline keeps running while this happens. [4] Repair silver, then recompute gold For Delta, silver should support controlled upserts. I would avoid one massive MERGE. Instead: - process partition by partition - batch by date or key range - make writes idempotent - checkpoint progress - keep an audit table of changed records Then rebuild gold tables only for impacted windows. If a metric uses a 30-day rolling window, recompute 6 weeks plus the lookback period. [5] Add alerts for historical changes The system should alert when old truth changes. Add historical mutation alerts, bronze/silver/gold reconciliation, freshness checks by source_updated_at, lineage for impacted dashboards/models, and dataset versioning for ML runs. The winning design is: - immutable bronze - upsertable silver - recomputable gold - correction-aware backfills - consumer notification - impact tracking In short: Do not build a pipeline that only understands yesterday. Build one that knows history can change, repairs only the affected slice, and warns people before wrong data becomes business truth.

If you are applying to Data Science jobs, knowing Statistics is not enough. You need to know how to a͟p͟p͟l͟y͟ Statistics to real world problems. This will be tested in your case study interviews. Let’s walk through a case study interview together: *𝘒𝘦𝘦𝘱 𝘪𝘯 𝘮𝘪𝘯𝘥 𝘵𝘩𝘦𝘳𝘦 𝘪𝘴𝘯’𝘵 𝘰𝘯𝘦 𝘳𝘪𝘨𝘩𝘵 𝘢𝘯𝘴𝘸𝘦𝘳 𝘵𝘰 𝘵𝘩𝘪𝘴 𝘲𝘶𝘦𝘴𝘵𝘪𝘰𝘯. 𝘠𝘰𝘶𝘳 𝘢𝘯𝘴𝘸𝘦𝘳 𝘮𝘪𝘨𝘩𝘵 𝘭𝘰𝘰𝘬 𝘷𝘦𝘳𝘺 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘵 𝘧𝘳𝘰𝘮 𝘮𝘪𝘯𝘦. ——— 𝗤𝘂𝗲𝘀𝘁𝗶𝗼𝗻: Your product team has identified reducing customer churn as a key opportunity for increasing revenue. They have asked you, their Data Scientist, to uncover the key drivers of customer churn. How would you approach answering this question? ——— 𝗧𝗵𝗲𝗿𝗲 𝗮𝗿𝗲 𝗮 𝗳𝗲𝘄 𝘄𝗮𝘆𝘀 𝘁𝗼 𝗮𝗽𝗽𝗿𝗼𝗮𝗰𝗵 𝘁𝗵𝗶𝘀 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻: We’re going to use logistic regression because the results are easy to interpret and communicate. 𝗦𝘁𝗲𝗽 𝟭: 𝗜𝗱𝗲𝗻𝘁𝗶𝗳𝘆 𝘀𝘂𝗰𝗰𝗲𝘀𝘀 𝗺𝗲𝘁𝗿𝗶𝗰𝘀 & 𝗽𝗼𝘁𝗲𝗻𝘁𝗶𝗮𝗹 𝗱𝗿𝗶𝘃𝗲𝗿𝘀 For churn, we could use the metrics: - Monthly churn rate - Customer lifetime value Next brainstorm potential churn drivers: - Pricing issues - Poor customer support - Missing product features - Bugs or technical problems - etc 𝗦𝘁𝗲𝗽 𝟮: 𝗘𝘅𝘁𝗿𝗮𝗰𝘁 & 𝗲𝘅𝗽𝗹𝗼𝗿𝗲 𝘁𝗵𝗲 𝗱𝗮𝘁𝗮 Understand the data that we have, and prepare the data for the model. Some steps to consider are: - How do we handle missing values? - How do we handle outliers? - Should we create any new variables? 𝗦𝘁𝗲𝗽 𝟯: 𝗕𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝗹𝗼𝗴𝗶𝘀𝘁𝗶𝗰 𝗿𝗲𝗴𝗿𝗲𝘀𝘀𝗶𝗼𝗻 𝗺𝗼𝗱𝗲𝗹 We will build a standard logistic regression model, using these steps - Remove highly-correlated variables - Select variables to include in the model - Evaluate model performance 𝗦𝘁𝗲𝗽 𝟰: 𝗜𝗻𝘁𝗲𝗿𝗽𝗿𝗲𝘁 𝘁𝗵𝗲 𝗿𝗲𝘀𝘂𝗹𝘁𝘀 𝗼𝗳 𝘁𝗵𝗲 𝗺𝗼𝗱𝗲𝗹 & 𝗺𝗮𝗸𝗲 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀 Based on the model, we can identify the features which are most predictive of churn. Then we’ll have to provide actionable recommendations — i.e. 𝘀𝗽𝗲𝗰𝗶𝗳𝗶𝗰, 𝗯𝘂𝗶𝗹𝗱𝗮𝗯𝗹𝗲 𝘀𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝘀 that can be implemented to address a problem. 𝗦𝘁𝗲𝗽 𝟱: 𝗖𝗼𝗺𝗺𝘂𝗻𝗶𝗰𝗮𝘁𝗲 𝗿𝗲𝘀𝘂𝗹𝘁𝘀 The key here is that your partners are convinced of your recommendations. - Write reports or emails with your findings - Set up meetings to review your findings and recommendations - Follow-up regularly to ensure that your recommendations land ——— ♻️ If you found this useful, please repost it!

Salesforce just solved the problem every CRM leader pretends does not exist. Your customer profile is a lie. And now there is no excuse not to fix it. The Informatica acquisition is not a data governance play dressed up in a press release. It is the missing layer that turns Data Cloud from a unification engine into an actual source of truth. Here is what most ANZ enterprises get wrong about customer profiles. They assume if the data is in Salesforce, it is accurate. It is not. It is duplicated, inconsistent, and stitched together from systems that were never designed to agree with each other. I spent years inside MuleSoft watching this exact problem play out. Enterprise after enterprise building beautiful API-led connectivity, moving data in real time, and still ending up with three different versions of the same customer sitting in one CRM. MuleSoft solved the flow. It still does. Real-time connectivity, event-driven integration, and API orchestration at scale. Informatica solves the identity. Master data management, data quality rules, lineage, and deduplication. The layer that determines whether 'John Smith' in your pipeline is one customer or four. Under Data 360, these capabilities now sit natively together. MuleSoft moves the data. Informatica cleans and governs it. Data Cloud unifies it. One architecture. One profile you can actually trust. The organisations building Agentforce workflows on top of ungoverned profiles are about to learn the most expensive lesson in AI: agents do not question the data. They act on it. You cannot build an intelligent enterprise on a customer profile held together by hope and manual deduplication. #Salesforce #Informatica #MuleSoft #DataCloud #ANZ

How to Setup Salesforce Personalization for Marketing Cloud Next Ready to deliver "Just for You" experiences on your website? Salesforce Personalization takes your marketing beyond emails by using behavioral and demographic data to show personalized banners, product recommendations, and content to each website visitor in real-time. This technical setup guide walks through deploying Foundation Data (Personalization Point, Decision, Personalizer, and Log DMOs), configuring Calculated Insights for daily personalization metrics, installing the Personalization Pipeline Intelligence Dashboard to monitor performance, and enabling the Attribution Model to track revenue events (Browse, Add to Cart, Purchase). The prerequisite is a Data Graph, and the payoff is real-time personalization powered by Data Cloud customer profiles. If you're moving beyond basic email personalization to full website experience optimization, this is your roadmap. https://lnkd.in/gmzA82rs