If you've ever wondered what it would look like to rebuild a marketplace from the ground up, with AI at the center rather than bolted on as an afterthought, you're about to find out.

Masha joined Groupon last year to lead the Merchant Experience team, and she brought with her nearly three decades of building across fintech, healthtech, adtech, AI hiring, and digital media. She has been a startup founder, a hands-on engineer, and everything in between. What drew her to Groupon was a specific challenge: taking a merchant from "I want to be on Groupon" to "my campaign is live" in minutes, not weeks.

Her team owns the full merchant lifecycle, from the moment a business discovers Groupon through to campaign creation, compliance, quality scoring, and ongoing operations. And the tools they are building to power that experience, including the Truth Layer, AI Deal Creation, and a three-layer evaluation framework, are not roadmap items. They are live, iterating, and generating data today.

In this edition, Masha talks about what it really takes to build foundational infrastructure under delivery pressure, why she believes everyone is a builder, and what the "aha moment" for a merchant on Groupon actually looks like.

"AI at the center, not bolted on as a feature, but as the operating model."

Can you tell us about your journey at Groupon and your scope as VP of Merchant Experience? What does your world look like as you lead the team responsible for how our merchants interface with our platform?

I joined Groupon last year to lead the Merchant Experience team. We own everything a merchant touches, from the moment they discover Groupon through landing pages, to signing up, creating their first campaign, getting it live, and then managing their business on the platform. That's the full lifecycle: discovery, onboarding, campaign creation, compliance, quality scoring, and ongoing merchant operations. My team spans AI services, Merchant Center, deal lifecycle workflows, and compliance agentic automation.
Before Groupon I spent nearly three decades building products and software across media, real estate fintech, AI hiring, adtech, healthtech, and digital media overall. I was a startup founder, a hands-on engineer, and everything in between. What brought me here was the opportunity to rebuild a merchant experience from the ground up with AI at the center, not bolted on as a feature, but as the operating model.

Your current focus is heavily strategic, centered on building the core pillars that will power our future merchant tools. How do you define your mission when you are tasked with architecting the 'big picture' for the entire marketplace?

Make it so a merchant can go from "I want to be on Groupon" to "my campaign is live and generating revenue" in minutes, not weeks. The current median time to launch a campaign takes days. That's not a technology problem, it's a decision-load problem. Merchants are small business owners running between appointments. They don't have time to figure out pricing strategy, write marketing copy, understand fine print requirements, and navigate compliance. We should do all of that for them, intelligently, using everything we know about their market, their category, and what actually sells.

Specifically, you are spearheading the AI Deal Creation and the Truth Layer. For those outside your team, can you break down what these are and why they are the essential building blocks for Groupon’s next chapter?

The Truth Layer is our category intelligence system. We call it Maestro - Category Playbooks Ground Truth. For every type of service we sell on Groupon, massage, laser hair removal, facials, nail services, you name it, we've built a playbook that captures what actually works. What price points convert. What deal titles attract clicks. What fine print is legally required. What licensing the merchant needs. What the refund patterns look like. This isn't AI making things up. It's AI grounded in real data about real categories.

AI Deal Creation uses the Truth Layer. When a merchant signs up and tells us they run a waxing studio in Chicago, we already know what services to recommend, what discount structure works for their market, what the deal description should say, and what compliance documents they need. We generate a complete, publish-ready deal in under 60 seconds. The merchant reviews it, approves it, and they're live.

And every deal that gets generated goes through our evaluation framework, a three-layer quality system with deterministic checks, LLM judges scoring across 24 dimensions, and quality gates that catch problems before a bad deal ever reaches a consumer. Think of it as QA for AI: we can't manually test every permutation of what AI generates, so we built AI that evaluates AI, at scale, in real time.

Why are these foundational? Because without the Truth Layer, AI-generated deals are just prompt wrappers, they look good but they don't perform. Without the eval framework, you're shipping blind. With both, every deal is grounded in what actually converts for that specific category in that specific market, and verified before it goes live. That's the difference between a demo and a product.

Building long-term architectural rigor requires immense focus. As a leader, how do you balance the pressure for immediate results with the Extreme Ownership needed to ensure these foundational projects are built to last?

I don't see them as competing. The Truth Layer and AI Deal Creation are how we deliver immediate results. Every deal we generate is an experiment. Every merchant who signs up and goes through the AI flow gives us data. We're not building a platform and then launching it. We're launching small, measuring fast, and iterating.

We've invested in building a pretty slick operating system that allows us to accelerate our AI-based delivery. It connects our strategy, our data, our playbooks, our analytics tools, and our team workflows into one place. When I need a status update, I pull live data from BigQuery, Asana, and PostHog in one command. When we need to validate a deal, the eval framework runs automatically. When someone new joins, they clone the repo and have the full context in minutes. The foundation isn't separate from the speed, it's what will make the speed possible.

The Truth Layer is designed to be a single source of truth for everyone—from Sales to Content. How do you manage the leadership challenge of building a product that requires alignment and adoption from so many different departments?

Honestly, it's hard. Cross-department alignment is one of the hardest things to get right at any organization. What I've learned is that you can't present your way to alignment or adoption. I pull the right people together and show, don't tell. We roll up our sleeves and work together. We try to get closely aligned across the organization and spend a lot of time on shared context, making sure everyone is looking at the same data,the same goals, and now the same playbooks.

I embody extreme ownership and demand that from my team as well. We write a lot of async updates that help us stay aligned without requiring check-in meetings. So when someone in Sales will have a question about why a deal is priced a certain way, the answer is in the playbook. When compliance needs to understand why a deal was recommended for approval, the audit trail is in the workflow. Once we see an AI-generated deal outperform a manually created one, the alignment conversation is over.

"Groupon isn't a discount platform. It's a customer acquisition engine that actually works."

In a year’s time, when AI-guided deal structures are the standard, what will be the 'aha!' moment for a merchant joining Groupon? What is the ultimate impact you want to have on their growth and success?

They'll type their business name, and within a minute, they'll see a complete deal ready to go live, title, description, images, pricing, fine print, all generated from what we know about their market. They'll look at it and think, "This is exactly what I would have written if I had a marketing team." And they didn't have to fill out a single form or make a single decision they weren't sure about.

The second aha moment comes 30 days later, when they see actual customers walking through their door. Not clicks. Not impressions. Real people with a voucher. And 75% of those customers were new to them, they would never have found the merchant without Groupon. That's when the merchant gets it: this isn't a discount platform. It's a customer acquisition engine that actually works.

"My rule is: If I can't do it, I can't explain it. I build."

What’s one belief about leadership or execution that you hold strongly - even when it’s uncomfortable?

Everyone is a builder. I don't believe in product managers who write specs for other people to build. I prototype, I pull data, I write code, I review designs. My CEO does the same. If you can't build a proof of concept yourself, you can't evaluate whether an idea is worth pursuing fast enough. This makes some people uncomfortable because it blurs the lines between roles. But the lines were already blurring, AI just made it impossible to pretend otherwise.

Finally, what’s one thing about how you work or think that most people wouldn’t guess from your title alone?

How hands-on I am. My rule is: if I can't do it, I can't explain it. I build. I've built AI agents, trained a model, wrote evaluation frameworks, launched prototypes. I can launch a business, run a sales org, build a marketing campaign, or sit with engineers and geek out about prompt architecture and Temporal workflows. This week I built 5 POCs, including a sales call analysis tool and an objection handling simulator, while participating at an offsite in Chicago and interviewing PM candidates. People see "VP" and assume I'm in meetings all day. Most of my day is building.

Inspired? Come build with us