Home
Résumé (PDF, opens in a new tab)
← All work
Professional workAds, marketplaces & recommendations

Programmatic Loyalty: aligning advertiser goals and customer incentives.

Launching Programmatic Loyalty: a recommendation and optimization platform that translates advertiser goals into commission rates, budget pacing, and customer incentives.

View résumé (PDF) ↗

Product flow

  1. 01Advertiser spend + target ROAS
  2. 02Commission, pacing + incentive decisions
  3. 03Campaign and customer outcomes

The takeaway

Product leadership across marketplace economics, recommendations, and real-time data.

Platform annual revenue
$500M+
Conversion lift contributed to
+20%

Sanitized reconstruction from product requirements and interview materials. Proprietary interfaces, customer data, and implementation detail are omitted.

The problem

Sales and operations teams were manually configuring campaign economics across hundreds of advertisers. Spend goals, target return on ad spend, incentive allocation, and delivery pacing moved through multiple handoffs, making launches slow and the system difficult to scale.

My ownership

I took Programmatic Loyalty from idea to launch: validating the problem, aligning leadership on an MVP, defining the advertiser workflow and system requirements, and leading phased delivery with engineering, data science, design, sales, and revenue operations.

The product translated advertiser spend and target ROAS into recommended effective commission rates, pacing strategies, and customer incentive allocations. I also defined personalization for first-time buyers, new-to-merchant customers, and lapsed audiences.

A decisive product moment

Should fresh intent signals directly change live campaign decisions at launch?

More responsive optimization could help an underdelivering campaign recover faster. It could also blur the boundary between trusted pacing guidance and request-time behavior, making spend changes harder to explain and debug.

Put fresh intent in the live path

Higher responsiveness, with more request-time complexity and less predictable pacing behavior.

Keep the live path deterministic

Chosen

Use precomputed pacing state online, preserve a clean system boundary, and fall back conservatively when state is stale.

Why

I treated budget discipline and ROAS predictability as the product promise that could not move. We kept heavy settled-sales-based optimization offline and made the online layer fast, explainable, and conservative.

What changed

The team launched a system that could react quickly enough for campaign delivery while remaining predictable for merchants and operable for internal teams.

System view

From advertiser goal to measurable outcome

A simplified view of the product and data loop I defined with engineering and data science.

  1. 01Advertiser intentMonthly spend, target ROAS, promotion context
  2. 02Offline guidanceCommission recommendation, pacing plan, profitability guardrails
  3. 03Live decisioningEligibility, ranking, incentive bucket, delivery surface
  4. 04Customer experienceRelevant offers across app, email, push, and other placements
  5. 05MeasurementSpend, ROAS, conversion, margin, pacing, and system health
Foundation

Kafka ingested events · Redis served low-latency decision state · Snowflake remained the settled-sales and analytical source of truth

Leadership scope

How I made the work move

My role was to create shared decisions across the people building, evaluating, operating, and using the product.

01Engineering
Defined system boundaries, failure behavior, state transitions, API requirements, and phased milestones.
02Data science
Aligned the advertiser inputs, recommended outputs, forecasting logic, and evaluation measures.
03Design
Turned complex campaign choices into a usable setup flow, then iterated from beta behavior and drop-offs.
04Sales & revenue operations
Used discovery, sandbox demos, advertiser edge cases, and role-specific dashboards to make the product useful and sellable.
05Leadership
Built the opportunity case, agreed on the MVP boundary, and used a phased pilot plus shared success metrics to earn trust.

Selected product artifacts

The structure behind the decisions

Sanitized reconstructions show how I framed scope, contracts, and measurement without exposing internal systems or customer data.

MVP boundary

Ship the shortest complete path from goal to activation.

Included
Campaign creation, targeting, spend and ROAS input, recommendation preview, activation
Deferred
Post-launch reporting and campaign editing
Reason
Validate the highest-leverage workflow before expanding the operating surface

Decision scorecard

Make speed-versus-control tradeoffs explicit.

Primary outcomes
ROAS attainment, budget utilization, conversion
Marketplace guardrails
Margin health, pacing deviation, overspend risk
System guardrails
Latency, fallback rate, stale-state behavior

Professional outcomes

What changed

  • Programmatic Loyalty scaled to $500M+ in annual revenue.
  • Personalization and cohort-targeting work contributed to a 20% lift in conversion.
  • Online optimization used low-latency state while settled financial outcomes remained the source of truth.
Technical context
  • Recommendation systems
  • Kafka
  • Snowflake
  • Redis
  • Personalization

Up next

Better creative workflows. Shared decisions. Infrastructure that keeps up. ↗

Let’s talk about the product decisions behind the work.

Get in touch ↗