Case Study · Healthcare Payer

From 43 scattered pilots to one platform that paid for itself.

A US national health plan had more AI activity than any company should. The problem was that none of it was producing results anyone could measure. We came in and changed that.

$7.5M+

Annual savings unlocked

43 → 14

Pilots consolidated into a governed portfolio

37%

Measured contribution to FY24 business objectives

US National Health Plan · 14 months · 2023–2024 · AI Strategy and Portfolio Governance

The Situation

Forty-three teams, forty-three different ideas about what AI should do.

When we first spoke to the leadership team at this health plan, they were not short of AI. They had it everywhere. Different teams had spun up pilots using different tools and different vendors. Some were machine learning projects from 2021 that nobody had looked at in over a year. Others were brand-new generative AI experiments that two people in a business unit had started without telling anyone in IT.

The number that stayed with us from that first conversation was 43. That was how many distinct AI efforts were running across the organization at the same time. Not 43 use cases, 43 separate, unconnected projects, each consuming budget, time, and people, most without a single shared metric between them.

Key insight

The core problem was not a lack of ambition. It was that nobody owned the whole picture. No one could answer the question every leadership conversation eventually comes back to: what is all of this actually worth?

What We Found

The real cost of fragmentation is not what you spend. It is what you never measure.

When we mapped the full landscape of what was running, three things became clear almost immediately.

Before

43 pilots running on different frameworks, tools, and definitions of success

No central view of AI spend, budget scattered across 11 cost centers

Most pilots had no measurable business KPI attached at all

Multiple teams working on the same problems without knowing it

No process for deciding what to build next or what to stop

After

One governed portfolio with a shared methodology across all active use cases

Full spend visibility with ROI tied directly to business outcomes

Every active use case tied to a KPI with a baseline and a target

Shared infrastructure cutting duplicate effort across teams

A clear process for prioritization, oversight, and decision-making

How We Did It

We did not start by building anything. We started by understanding what already existed.

Our first instinct when we walk into a large organization is never to introduce new technology. It is to understand what is already there. In this case that meant spending the first month doing something simple: talking to people and writing down what we found.

Month 1

Full landscape audit

We mapped every AI initiative across the organization. For each one, we asked three questions: what business problem is this solving, how is success defined, and who owns it? Most pilots could not answer all three.

Month 2

Prioritization and triage

Of the 43 pilots, we recommended stopping 19 outright, merging 14 into 6 consolidated efforts, and accelerating 10. Telling teams that their project was being stopped is never easy. But without it, nothing changes.

Months 3–6

Building the governance layer

We built a central AI portfolio function with three things it did not have before: a shared KPI framework, a prioritization committee with real authority, and a reporting cadence that connected AI activity to business outcomes.

Months 7–14

Production and measurement

With the portfolio rationalized, we moved the highest-priority use cases to production. Each had a baseline, a target, and a clear owner. The $7.5M came from running fewer things better, not from any single project.

The Outcomes

Numbers that moved because we made sure they could be measured.

$7.5M+

Annual savings across the rationalized portfolio

37%

Measured contribution to FY24 business objectives

43 → 14

Pilots reduced to 14 active, governed, production-ready use cases

Cost centers consolidated into one AI portfolio budget

14 mo

From first conversation to a fully governed, producing portfolio

100%

Active use cases now have a KPI, a baseline, and a named owner

What We Learned

Four things we would tell any organization in the same position.

Stop before you build

The fastest path to AI outcomes is almost never starting a new project. It is stopping the ones that are not working. Organizations resist this because stopping feels like failure. It is not. It is discipline.

Measurement is not a reporting task

You cannot measure outcomes after the fact. The baseline has to exist before the work starts. Every use case we accelerated had a before-and-after comparison built in from day one. That is why we could show $7.5M with confidence.

Governance is not bureaucracy

Governance means clarity: who decides, how they decide, and what happens when a use case stops delivering. With that in place, things move faster, not slower.

The portfolio view is the unlock

No single use case produced $7.5M. The savings came from running the right ten things well instead of forty-three things badly. That shift only happens when someone is looking at the whole picture at once.