Sarthak Guleria

Incentives are easy to add and hard to remove.
Most systems fail because they reward the wrong behavior for too long.

• + A system that worked (temporarily)
  MilkyWay is a liquid staking protocol. The points program drove participation quickly — contributing to $250M+ in staked assets. But the real work started once patterns emerged and abuse vectors became visible.
• + Where incentives started getting gamed
  Certain behaviors looked healthy on dashboards but felt wrong when mapped to actual usage.
• + A trade-off I underestimated
  Short-term activity versus long-term trust isn't a neutral choice. It always compounds.
• + What I do differently now
  Incentives need decay, friction, and iteration. Static rewards don't survive contact with users.

Growth is rarely blocked by distribution.
It's blocked by friction nobody wants to own.
At MilkyWay, I grew the community from zero to 70K+ followers.

• + Initia testnet: what actually drove participation
  Clear tasks, low ambiguity, and visible progress mattered more than hype.
• + Where onboarding friction showed up
  Drop-offs almost always pointed to UX issues, not lack of interest.
• + Why community growth only works with product clarity
  Engagement without understanding creates noise, not users.
• + What I stopped doing despite pressure
  Some growth levers create activity but destroy signal. I stopped pulling them.

Payments don't forgive bad assumptions.
Users feel mistakes immediately.

• + Why payments UX exposes product flaws
  Confusion shows up as hesitation, retries, or silent churn.
• + What most fintech landing pages hide
  Fees, limits, and edge cases are buried because they're inconvenient.
• + Where users lose trust first
  Not during failure — during ambiguity.
• + Why clarity beats features
  Fewer options, clearly explained, outperform flexibility without context.

Some decisions looked right on paper and failed in reality.

• + A points multiplier that backfired
  Increased activity masked worsening behavior quality. The metrics looked healthy, but the users it attracted weren't the ones who stayed.
• + A referral incentive that created noise
  It brought attention without intent. The numbers went up, but signal-to-noise ratio tanked and feedback became harder to trust.
• + Listening too literally to feedback
  Users describe symptoms better than causes.
• + What this taught me about second-order effects
  Every lever creates shadows you don't see immediately.

I build browser-based tools that model DeFi strategies before you risk real capital.
Stress-test the numbers, understand the risks, then decide.

A case study in increasing DeFi participation without notification spam.
Most protocols solve low participation by pushing more alerts. We built a dashboard that let users pull what they needed instead.

• + Situation
  MilkyWay had incentives across multiple protocols: LP positions on Thena and Pancakeswap, lending on Echelon, staking opportunities with variable APRs. Users kept asking "where should I put my tokens?" but we couldn't answer without spamming them with every new pool. The real challenge: keeping 390k+ wallet holders informed about opportunities without training them to ignore our messages.
• + Task
  Increase participation in milkLST and MILK DeFi positions across partner protocols while maintaining trust and reducing support volume. Success meant users discovering opportunities themselves rather than waiting for announcements.
• + Action
  Built a live DeFi dashboard aggregating TVL, APR, reward tokens, and platforms in one view.
  
  
  Key decisions: Showed real-time APRs instead of estimated ranges. Highlighted new opportunities without hiding established ones. Let users filter by risk preference. Created step-by-step guides for each strategy type.
• + Result
  Featured pools saw measurably higher participation. Support tickets asking "where should I stake?" dropped noticeably. Users started deploying across multiple protocols instead of clustering in one. The dashboard became our go-to answer for "I have MILK, now what?"
  
  What this taught me: Push notifications create urgency. Dashboards create autonomy. For long-term products, autonomy compounds better than urgency.

A case study in making cross-functional work visible without slowing anyone down.
Fifteen people. Three departments. Everyone shipping. The problem wasn't effort — it was that nobody had a shared view of what was happening across the team.

• + T — Tasks
  Fifteen people. Three departments. Everyone shipping. The problem wasn't effort — it was that nobody had a shared view of what was happening across the team. Not because they didn't care. Because when you're moving fast, coordination feels like overhead until it suddenly isn't.
• + R — Review
  The gaps were predictable in hindsight. Blockers had no clear owner. Deadlines slipped quietly instead of visibly. Tasks that touched multiple teams were the most at risk. The fix needed to work without asking everyone to change how they worked.
• + A — Architecture
  Built a three-layer system. Task breakdown across Marketing, Engineering, and Growth. A Jira hierarchy with consistent naming so everything traced back to a parent. A Master Track Sheet on top — plain language, no Jira expertise required.
  
  
  The part I spent the most time on was the escalation flow for blockers. Most coordination systems ignore this and then wonder why blockers stall quietly for weeks.
• + C — Coordination
  Onboarded team leads, not every individual. Two weeks to adoption. At peak the system was tracking 50+ tasks across all departments.
• + E — Execution
  Deadlines became more reliable. More importantly, timeline estimates got more honest — because dependencies were visible before commitments were made, not after.
  
  What this taught me: The best coordination system is the one the team actually uses. That usually means designing for the least Jira-fluent person on the team, not the most.