Margot Mueller
"I help product teams ship. I focus on what users need, how teams work together, and clear processes so we can deliver."
This isn't a typical resume. Below I walk through how I'd approach a new opportunity at Anthropic: personas, prioritization, and alignment.
Creating Structure from Ambiguity
The JD says: "comfortable with ambiguity and skilled at creating structure where none exists." That's my sweet spot!
Here are three concrete examples from my work at Stitch Fix, Opendoor, and Levitate:
The Process
1. Map the Problem
What's broken? What's the cost of the current state? Skip the full audit — talk to 3 people who feel the pain. Use MCP to pull as much context from across the internal org as possible.
2. Prototype the Extreme
Don't design the system — build the simplest version that tests whether the behavior shift you want is even possible. Start at the edges.
3. Measure and Kill or Scale
Did it work? If no signal at the extreme, stop. If yes, iterate fast based on usage, not assumptions.
Real Examples from my prior work
Stitch Fix
Product Operations
When I joined, styling feedback was scattered across client notes, stylist Slack channels, and support tickets. No single view of themes (fit, style, inventory, algorithm)—so we couldn't prioritize or route to the right team.
I built a feedback synthesis system: (1) weekly stylist roundtables with a consistent template and owner, (2) categorized feedback dashboard (fit, style, inventory, algorithm) to surface themes, (3) routing rules—product/algorithm issues → Product, operations → Ops.
Time-to-insight dropped from weeks to days. Inventory and styling algorithm decisions became data-driven; we could see which themes showed up most and route them to the right owners.
Opendoor
Product Operations
Market pricing feedback lived in 6+ regional spreadsheets and email threads. Pricing leads had no single view of accuracy (list vs. sale price) or market signals, so decisions were reactive and slow.
I created a pricing feedback loop: (1) standardized intake form for field teams in 12 markets, (2) Looker dashboard tracking list-to-sale variance and days-on-market by market, (3) monthly pricing review with 8 regional leads and a single decision log.
Pricing accuracy improved 12% within 6 months. Time to respond to market shifts dropped from weeks to days; regional leads could see their market in context of the rest of the portfolio.
Levitate
Product Operations
When I joined, product feedback lived in 12+ Slack channels, email threads, and support tickets. No single place to see requests, no rules for who owned what, and no way to prioritize beyond "who yelled loudest."
I built a Voice of Customer system from scratch: (1) weekly CS/Sales sync with a simple template and owner, (2) Airtable-based feedback dashboard with tags (feature request vs. bug vs. question), (3) routing rules—feature requests → Product, bugs → Engineering—so nothing fell through the cracks.
~70% reduction in manual triage. Request visibility and data-driven prioritization improved; we could show which themes came up most and tie them to segments (e.g. enterprise vs. SMB).
The Product Ops Role
The system that makes product decisions faster and less wrong.
Inputs
- What CS and Sales are hearing
- Who uses what, where they get stuck
- What competitors are doing
Rituals
- Monthly business review
- Product–Sales sync every two weeks
- Launch checklists
Outputs
- Launch playbooks so CS is ready
- Internal updates: what shipped and why
- Clear success metrics before we build
The Opportunity: AI-Native Wealth Management
Private wealth management is ripe for disruption. Services like Range.com charge $6,000/year for what is essentially: document parsing, tax scenario modeling, goal tracking, and periodic human advice.
Claude already excels at all of these tasks. With the right product packaging, Anthropic could capture significant market share in the $1.2T wealth management industry—while genuinely helping people make better financial decisions.
I'm a Range customer today—I live the current model and see the opportunity from the user's side.
Range.com (Today)
- Price$6,000/year
- AvailabilityWait for human CFP
- Document HandlingLimited document uploads
- Advice QualityGeneric advice
Claude (Potential)
- PriceClaude Pro: $240/year
- AvailabilityInstant, 24/7 availability
- Document HandlingNative PDF/document analysis
- Advice QualityPersonalized, context-aware
Understanding the User: Meet the Personas
The Tech Executive
VP of Engineering, $400k total comp, RSUs vesting quarterly
Pain points:
- • Equity compensation complexity
- • Multi-state tax implications
- • No time to research
Current solution:
Expensive CFP ($5k/year) + scattered spreadsheets
"Just tell me what to do with my RSUs"
The Dual-Income Household
Two professionals, combined $350k, two kids, saving for college
Pain points:
- • Coordinating 401ks, 529s
- • Tax optimization across two employers
Current solution:
TurboTax + hope
"A plan that actually accounts for both of us"
The Small Business Owner
Owns an LLC, $250k revenue, wants to maximize retirement contributions
Pain points:
- • SEP IRA vs Solo 401k confusion
- • Quarterly estimated taxes
- • Bookkeeping
Current solution:
Accountant once a year, otherwise DIY
"Help me not screw up my taxes"
What It Could Look Like
Recent
Workspace
Tax-loss scan found 3 positions with unrealized losses. Sell orders are teed up and ready for your review—could offset ~$4.2K in gains.
I have 2,500 RSUs vesting next month. My marginal tax rate is 35%. Should I sell immediately or hold?
|
Metrics That Matter
We run a controlled experiment with ~100K users in the financial vertical. The main question: does Claude become the stickiest platform for their financial workflows? These are the metrics I'd use to answer that—engagement first, then conversion and retention.
Engagement (North Star)
Goal: make Claude the stickiest platform. Are users coming back to financial workflows?
Feature Stickiness
Are financial features habit-forming, not one-and-done?
Feedback to Iterate
Once the financial experience is live, the priority is making it easy for customers to give feedback—and feeding that signal back into the system so it self-improves, with human-in-the-loop judgement where it matters.
Easy feedback everywhere
In-flow thumbs, optional comments, and lightweight prompts so users can say what worked or didn't—without leaving the conversation.
Signal back into the system
Feedback gets categorized, prioritized, and routed. Product and ops see trends; the model and prompts get tuned based on what users actually do and say.
Human in the loop
High-stakes or ambiguous cases get human review before they change behavior. We improve fast, but we don't automate judgement calls that need a human.
Why This Role, Why Now
I want this role because I truly think Anthropic will be 50% of the economy over the next 10 years. I want to be part of building that.