Puneet Gangrade

The Silent Killer in MarTech: Data Debt Everyone talks about tech debt, but very few talk about data debt. Data debt happens when companies accumulate years of fragmented, duplicative, or poorly governed data across platforms. The result? - CDPs and CRMs that don’t talk to each other. - “Single customer view” dashboards that are anything but single. - Teams spending more time cleaning and reconciling data than actually using it. Why this matters: AI, personalization, and analytics are only as good as the foundation beneath them. If your data is riddled with gaps and contradictions, your MarTech stack can’t deliver on its promise. A few ways to start addressing data debt: 1. Audit: Know where your data lives and who owns it. 2. Prioritize: Not all data is worth saving. Focus on what drives customer and business outcomes. 3. Govern: Define rules for collection, storage, and integration before adding another tool. Data debt isn’t just a back-office problem, it’s a growth limiter. I’ve seen organizations pour millions into new MarTech tools, only to end up with fancy dashboards powered by broken data. The stack looks impressive, but the insights can’t be trusted. Fixing data debt doesn’t get the spotlight, but it pays off in every area that matters: more accurate personalization, cleaner reporting, smoother integrations, and ultimately, better customer experiences. Leaders who treat data debt like a strategic priority, not a technical chore, will have a real advantage. When was the last time you audited the data that fuels your MarTech stack, not just the tools themselves? #MarTech #EmergingTech #DataStrategy #DigitalTransformation #MarketingTechnology

9

1 Comment

The Shift to AI-Native Analytics in MarTech The MarTech landscape is undergoing a significant transformation as organizations move away from traditional dashboards towards AI-native analytics platforms. Companies like Astrato Analytics are leading this change by offering cloud-native business intelligence (BI) solutions that enable real-time data exploration and intelligent decision-making workflows. This shift is driven by the need for more agile, predictive, and automated insights to stay competitive in today's fast-paced market. https://lnkd.in/gQHEEfC3 How is your organization adapting to the rise of AI-native analytics in your MarTech stack? #MarTech #AINativeAnalytics #DigitalTransformation #EmergingTech

21

2 Comments

The size of the model is a critical consideration for agentic search, as most search applications require low latency and high throughput. To date, agentic search models, particularly agentic RAG models, have been large due to the demands of reasoning, planning, instruction following, multi-hop reasoning, tool usage, and response synthesis. This paper investigates whether smaller models, with fewer than one billion parameters, can be effectively applied to agentic search. The authors introduce Distillation-Guided Policy Optimization (DGPO) for training such models and propose a benchmark, Agentic RAG Capabilities (ARC), to evaluate their performance. Their results demonstrate that compact models can achieve strong performance on agentic RAG tasks. https://lnkd.in/gqQscdTY

7

From Strategy to Activation: Getting Your Data AI-Ready Continuing the conversation from the Snowflake World Tour, let's build off the foundational strategy with data activation. Most companies want to activate AI, but their data just isn’t ready for it. Dirty, decentralized, and inoperable data stops innovation before it starts. If your AI strategy is set, it’s time to turn to your data strategy. Here’s how to get your data ready to fuel innovation and growth: Clean it - Audit what data exists and what’s missing. - Eliminate duplicates and outdated records. - Set up continuous hygiene processes so bad data doesn’t creep back in. Centralize it - Consolidate your data into a unified, governed environment. - Define clear ownership, permissions, and metadata standards. - Eliminate silos so teams can operate from a single source of truth. Activate it - Connect clean, centralized data to the AI workflows that drive value. - Use it to train models, personalize experiences, and surface insights faster. - Track outcomes, refine what works and retire what doesn’t. You can’t build an intelligent organization on an unintelligent data foundation. AI readiness isn’t about more data, it’s about better data. If AI is the engine, is your data clean enough to fuel it? #DataStrategy #AI #Snowflake #DigitalTransformation #DataGovernance #Analytics #EmergingTech

9

2 Comments

The next chapter of agentic advertising is being written at the infrastructure layer. I recently shared some thoughts in Ad Age on why the future of AI in advertising isn’t just better models but building the infrastructure and workflows that allow AI to work within real advertising environments. I also share some of the thinking behind our recently launched Amazon Ads MCP Server. If you’re exploring where AI agents fit into advertising today, take a read. I’d love to hear your thoughts: https://lnkd.in/g7R6fHbQ.

161

1 Comment

In the consumer world, we want a single interface that can write a poem, plan a vacation, and debug code. In that open world, scale is the only strategy that works. But enterprise workflows don’t live in an open world. Most B2B problems like parsing invoices, routing tickets, classifying clauses operate in closed systems. They have well-defined inputs, explicit outputs, and hard failure modes. In my latest article, I argue that the future of enterprise AI isn't always about getting bigger, it's often about model specialization. The most effective architectures I’m seeing are cascading. They use SLMs for routine volume, and escalate to LLMs only when deep reasoning is required. Read the full breakdown below. https://lnkd.in/gkyJbFWb

57

6 Comments

Our latest post on Analytics@Meta covers how to create good governance around experimentation and how to get new experimentation ideas: "Decision metadata is the backbone of our experimentation culture. It provides valuable insights into experiment outcomes, enabling us to refine products, identify areas for improvement, and make informed decisions. One use case is to identify opportunities at a large scale. By examining experiments from a various team or set of teams, one can identify (a) what shipped and had the most efficient tradeoffs, and (b) what tradeoffs we made for experiments that didn’t ship but perhaps should have. With high quality decision metadata, AI models can even be leveraged to provide tailored suggestions based on previous experiment outcomes." See full article here: https://lnkd.in/g3hUXDFF #Meta #DataScience #Experiment #AI

66

2 Comments

Now this is an eval! OpenAI has released GDPval (great name). It includes real tasks, with real inputs and outputs, across 44 occupations covering $3 trillion in annual US wages. All of the tasks were created by experts in their respective domains. The best model right now, Claude Opus 4.1 (good for OpenAI being transparent about this), does 47% of the tasks as well or better than a human (so, roughly $1.5 trillion worth of tasks). And even when the models weren't 100% successful, they still accelerated work, doing in minutes what took humans 7-9 hours. They're going to continue improving and expanding the eval. Also, they're using what they're learning from it to improve model training (I'm assuming this includes RL), and have already seen an impact. Buckle up.

29

6 Comments

Great post from Nate Carter: "Marketers are demanding to know how data is sourced, how it’s validated and whether it’s grounded in real-world signals rather than anonymous digital exhaust. Data providers that can clearly document their inputs and processes will rise in influence; those that can’t will fade into the background noise of the open web." #datamarketing, #b2bdata, #digitaladvertising, #AI

26

2 Comments

AI & Market Research - Can AI take surveys for us? I haven't posted about too many uses of AI with Market Research to date, however I think this case study should get a mention. A recent study by PyMC Labs and Colgate‑Palmolive finds that large language models (LLMs) can closely simulate human responses in consumer product surveys—hitting ~90% of human test–retest reliability across 57 surveys (9.3k human responses). How it worked: Instead of asking a bot for a number, the team asked for a short written answer with a why. That text was then compared to five Likert “anchor” statements using embedding similarity (their Semantic Similarity Rating / SSR method). Result: 90% of human test-retest reliability. They mirrored distributions of responses across demographic segments such as age and income. Researchers emphasise that this doesn’t mean LLMs think like people, but their outputs can align with human behavioural patterns under defined conditions Want to try this? 1) Start with a low-risk, high-volume concept test you already run. 2) Write clear Likert anchors (1–5) as plain sentences. 3) Prompt an LLM for a short rationale per concept and then map that text to anchors using embedding similarity. 4) Create a small human holdout to benchmark; only proceed if your synthetic vs human gap is tight. 5) Be transparent: label synthetic vs human, document prompts, and keep humans in the loop for final calls. Imagine being able to test new tourism or retail products or ad concepts instantly across synthetic “audiences”, before spending on full panels. Huge potential for insight and speed — as long as we keep the human layer where it matters most: judgement, context, and creativity. This isn’t about replacing people; it’s more about "simulate → iterate → validate" so teams can test more ideas, faster, and spend human budget where judgement matters most. Full story below 👇 #AI #MarketResearch #Insights #DigitalTransformation #SyntheticData #Tourism #Innovation https://lnkd.in/eRBWtRcs

6

1 Comment

This post from the Instacart Engineering Team is superb: "Scaling Catalog Attribute Extraction with Multi-modal LLMs" - https://lnkd.in/gKdNmnRS It sparked my interest as it leverages AI for an 'age-old' problem: understanding/extracting entities & their attributes. 8+ years ago, during my time at TNW , we had a similar need: Let our editors write, but help us understand in the meantime which topics (companies, CEOs, startups, investors) drive the most interest. Manually tagging isn't the best option if you have a 50K+ page archive. At Postmates, closer to Instacart's business, we wanted to build another 'catalog'. Many restaurants in your neighborhood offer burgers, but we didn't realize that menu item X could be similar to restaurant Y as there was no appropriate data architecture. These architectures can inform SEO, Email, you name it. The blog post provides detailed guidance on how to extract the entity (burger) more easily and derive crucial attributes (flavor, size), which is where true personalization comes in. This unlocks better personalization in Product & Marketing.

22

4 Comments

AI isn’t replacing your operations, it’s making them better. Forrester Consulting shows how a composite org achieved a three-year 207% ROI and $38M net present value with Google AI in Google Cloud Customer Engagement Suite.

12