In 2015, I had the opportunity to pitch Starbucks on a new type of loyalty platform. One that would leverage AI to deliver individualized gamified offers designed to nudge valuable new behaviors. Starbucks loved our vision and committed $25m — we raised another $5m from an institutional investor.

Over the next 6 years, we deliver more than $1b of incremental value to Starbucks, productized an AI-powered loyalty platform, and brought on additional Fortune 100 companies, including United Airlines. We grew the team to 130 employees and hit $25m in revenue (ARR). Along with my role on the executive leadership team, I led a 10 person design team across product, marketing and R&D.

In August 2022, Formation was acquired by BCG.

Marketing promo for Formation.

Origin Story

The vision for an AI behavioral nudging product began while I was consulting for Kaiser Permanente. They brought me in to help a Kaiser innovation team develop new ways to address the problem of patient adherence. More than 50% of patients prescribed a medication or health regime do no adhere, representing almost 10% of health care costs in the United States. I worked with their team to develop a gamified reward system for establishing and reinforcing new healthy habits.

I led the personalization practice at BCG and presented insights from the Kaiser work to the executive team at Starbucks. Partnering with the BCG data science team, we pitched a new type of AI-driven behavioral nudging program. I brought together a small team to build a working prototype, which led to securing an investment from Starbucks.

Early thinking about components of behavior influence.

A Platform For Behavioral Change

The work at Kaiser formed the core of our product thesis. But it was insights from BJ Fogg, director of the Stanford Behavior Design Lab, that gave us clear, actionable tools for driving new behaviors. The Fogg Behavior Model is the cornerstone of Formation platform, and serves as a core optimization framework.

The core Formation construct:

  1. Use past purchase behavior to determine high probability actions
  2. Use probable actions to construct a gamified offering rewarding their current actions, along with incentive to try a new behavior
  3. Reinforce new behavior using less and less reward until it becomes a habit
  4. Move to the next new behavior

The Fog Behavior Model was central to our product thesis.

Product Philosophy

We quickly ran into a challenging problem: humans. In particular, the operational users, business strategists, and data scientist that were responsible for customer engagement strategies. While we could show that automation delivers better outcomes, people want credit for the results. A bridge was needed to move people from what they understood (manual experimentation) to the promise land — an AI-driven system creating a variant cohort of one.

For the operational UX, we build the product to serve three primary operations:

  1. Strategy and creative translation
  2. Configuration and experimentation
  3. Measurement

An end-to-end solution for loyalty execution was a game changer for our customers.

Dashboard to manage experimentations and the allocation of total population to experimental RL vs. hypothesis-driven optimization.

Effective offer constructs were saved in an offer library for future use and evolution.

Manual segmentation criteria could be configured. Giving operators control, even when automated optimization was delivering far better results, was important for the teams running loyalty programs.

Internal UX walkthrough of offer experiment configuration.

The Art of Sales

A brand like Starbucks as a first customer was both a blessing and a curse. What we built for Starbucks was over-engineered for their specific needs, scale and budget. No other company in the world spends as much on customer retention. The marketing side of my design team continuously iterated on ways to tell our story outside Starbucks, while also trying to capitalize on the success of the program. In the last couple of years, we developed an effective sales strategy using design research as a way to target loyalty managers at fortune 100 companies. This translated into a number of sales.

Our Starbucks offer constructs created a new category of personalization dubbed "Hyper-Personalization."

United Airlines emulated our Starbucks offering and realized similar results of 3x lift in incremental spend.

The Formation brand was not your typical enterprise SAAS offering. We created the brand to convey the premium offering and personal nature of loyalty.

Future Focus - Motivation Alignment Platform (MAP)

The next phase of innovation we identified centered on the alignment of offers to individuals’ intrinsic motivations. We built working prototypes to show the added value of a system that understands the underlying motivations that activate customers. The exploration began by interviewing customers, behavioral psychologists, and academics studying human motivation. The functional prototype showed that we could realize up to a 50% savings in reward spend, and a 25% engagement lift. We did not implement the capability prior to acquisition.

Our goal was to provide support for motivation cohorts — allowing brands to tune rewards to individual intrinsic motivations.

We mapped existing motivation frameworks to reward behaviors.

An overview of cohort motivation gives businesses a clear view of how to optimize reward spend.


In the 5 years prior to acquisition, Formation delivered more than $1b of incremental value to each Starbucks and United Airlines. We also proved the model with other business verticals including grocery, retail and hospitality.

The Formation team and product is now a part of BCG's loyalty and personalization offering.

I wrote about the growth journey of Formation and the lessons I learned as a venture leader at BCG, startup founder, and operational leader of an enterprise SAAS company.