Making field learning measurable

The situation

AstraZeneca's US biopharmaceuticals business invests heavily in what its field force knows: launch readiness, clinical fluency, coaching quality. But the evidence of whether that investment works lived in fragments — survey tools, learning platforms, coaching forms, program trackers — each with its own scale, cadence, and owner. By the time anyone assembled a picture, the moment to act on it had passed.

In 2024, the client's learning strategy team set out to change that. I joined as analytics lead within the program's agency partnership, responsible for turning "make learning measurable" into systems leaders would actually use.

The problem underneath

Learning data is not sales data. There is no ledger of record, no agreed definition of a "field experience," and no natural unit of comparison across brands and teams. The real work wasn't visualization — it was measurement design: deciding what a learning signal means before deciding how to chart it.

The question leadership needed answered was simple to say and hard to build: is the field ready, and where do we coach next?

Decisions that shaped the work

Standardize the experience model first. Before building anything, we defined a shared semantic model for learning experiences that every program would inherit. This made the first release slower and every release after it faster. The trade-off paid for itself within the year.

One dashboard per program on a common core — not one mega-dashboard. Each sales program kept its identity and its own release, while inheriting shared definitions, patterns, and infrastructure. Small, signable releases kept stakeholders confident and rollbacks trivial.

Treat releases like regulated software. Signed release documents, full technical documentation, CI/CD coordination with the client's engineering team, and deliberate IAM and access-group design. In pharma, that discipline is what makes speed possible: once a release is signed, nothing gets relitigated.

What shipped

Seven experience dashboards moved into production across 2024–2025, built on Power BI over the client's data estate with a documented SQL semantic layer. The largest release consolidated thousands of data points into an experience the client described, unprompted, as "so fast and seamless."

Outstanding … Check another one — and the largest one — off the list! Thousands of data points and the experience from the user perspective with the dashboard is so fast and seamless — what an incredible job well done!

— Client program leader, on the August 2025 production release

In 2026 the program crossed a threshold: from a suite of dashboards to a single learning intelligence hub. After an on-site working session, I translated the discussion into a tangible hub mock-up — a centralized experience unifying coaching, performance, and learning signals — which the client endorsed as the program's next chapter.

Leading it

I led the client relationship directly — weekly working sessions with learning-strategy leadership, release announcements, access verification, and the unglamorous work of absorbing pivots gracefully. Programs changed goals, scopes, and structures repeatedly across two years; the architecture (and the relationship) were built to make that cheap.

The clearest evidence came in April 2025, when the client's program leader sent an unsolicited note to agency leadership:

Their work … has been second to none. Each and every time there is a pivot, a new ask, or a conceptual change to the design, [they] display the utmost degree of professionalism, added value within the work, and a collaboration that lets us all know we have true partners.

What moved

Every release scheduled for 2025 shipped on time — four of four, each with a signed client sign-off. The partnership was formally recognized twice, with program leadership reporting that the client views the work as "groundbreaking." Most tellingly, the engagement grew: from one dashboard, to a program suite, to a client-endorsed vision for a unified hub.

Lessons

The semantic model is the product. Charts are interchangeable; agreed-upon definitions are not. The standardization work nobody sees is what made everything visible possible.

Cadence builds more trust than scope. Four modest, on-time, signed releases beat one ambitious late one — especially in a regulated environment.

Make the client the hero. The strongest consulting outcome isn't a deck — it's the client explaining the vision in their own words, because they own it now too.

Sources: client emails 2024–2026, formal recognition may 2026, public linkedin post jun 2026