AMA: Nebius Head of Marketing, Media & Entertainment, Abhishek Ratna on Market Research

I think there's a natural order of progression here. Without good customer insight and solid product knowledge, messaging and positioning won't resonate with audiences. Without a solid foundation in messaging and positioning, product collateral and documentation will be weak. The most impactful thing a small PMM team can do (and I, by the way, have been a PMM team of one) is to understand the product and understand the customers first. That would be the biggest credibility driver within your org ...Read More

Honestly? My best CI doesn't come from a single source. It comes from triangulating across layers. Sales calls are my first stop. The reps at Databricks, Protopia and Labelbox heard objections I never would have found in a Gartner report. I'd sit in on calls, read Gong transcripts, and mine CRM notes for patterns. That's where real buyer language lives. From there I layer in community signals: Reddit threads, Hacker News, LinkedIn comments, G2 and Gartner Peer Insights reviews. On greenfield opp ...Read More

This is the exact situation I am in for the past seven years in AI marketing :) I'll share examples from Protopia and Atlassian with Rovo Studio. The trap is trying to wait for the category to stabilize before you position. It won't. So instead, I focus on the problem that's stable even when the solution space is moving fast. Buyers may not know what "agentic AI" means yet, but they absolutely know they're drowning in manual workflows and worried about AI governance. Lead with that. On research ...Read More

A few places I actually use, not just the ones that sound good on a list: Sales team. They hear objections in real time. I set up a shared Slack channel at multiple companies specifically for competitive intel from the field. 93% of PMMs say sales is their top CI source, and in my experience that tracks. AI Tools. PMM Sherpa, Perplexity and Gemini. Win/loss data. Structured interviews with buyers who chose someone else. This is underused and underrated. Competitor's own content. Their job postin ...Read More

Today the website is still your always-on pitch. But seven out of ten B2B buyers have already formed a strong opinion before they ever talk to sales. The website is where that opinion gets confirmed or killed. At Protopia I rebuilt the web presence from scratch with that in mind. Every page had a job: either teach the problem or prove we could solve it. Cut the fluff. Over the next three years? The website as a destination is going to matter less. Buyers will increasingly get synthesized answers ...Read More

The signal I trust most: specificity and unprompted repetition. If a customer says "this is really useful," that's noise. If three different customers in different companies, without being asked, describe the same workflow problem in almost identical language, that's signal. At Google, when we were rebooting tensorflow.org, we ran UX sprints and customer interviews. The feedback that changed our direction wasn't the loudest. It was the pattern: developers kept abandoning the same pages at the sa ...Read More

The mistake I see most often is leading with the new product's features instead of the customer's next problem. The better frame is your customer hired your existing product to do a job. What job are they still struggling with, adjacent to that one? The upsell lands when it feels like a natural extension of value they're already getting, not a separate purchase decision. At Labelbox I built 24 solutions across 6 verticals. The ones that sold fastest weren't the most technically impressive. They ...Read More

Partner marketing without PMM alignment is just co-branded noise. At Databricks I ran joint campaigns with Azure, AWS, Tableau, and others. The ones that drove real pipeline shared one thing: a unified problem narrative. We weren't promoting two products. We were telling a story about a customer problem that neither company could fully solve alone. The alignment mechanism that works: PMM defines the joint ICP and the specific use case the partnership unlocks. Partner marketing executes the campa ...Read More

Who has this problem badly enough to change behavior? Not who might benefit, but who is actively feeling the pain today. At Labelbox, this was the question that determined which of the 24 solutions we launched first. What does the buyer believe before they meet us? Their existing mental model is your real starting point. If you ignore it, your messaging lands in a vacuum. Why hasn't this problem been solved already? This is the "why now" question. The answer tells you what's changed in the marke ...Read More

The reactive trap: you're synthesizing what already happened. The proactive move: you're explaining what it means before anyone else has framed it. At Protopia, when a major AI governance regulation dropped, I wasn't writing "here's what this regulation says." I was writing "here's the three decisions your AI team needs to make in the next 90 days because of this regulation." That's a different posture. The framework I use: observation → implication → decision. Don't stop at the observation (the ...Read More

The instinct is to lead with the finding. Don't. Lead with the shared goal. "We're trying to understand why conversion from trial to paid is stalling. Here's what the research showed." Now the finding is in service of a problem everyone already cares about, not a challenge to someone's assumption. The other thing that helps: let the customer do the talking. A direct quote from a buyer carries more weight than my interpretation of the data. I'll often play a clip from a customer interview in the ...Read More