Skip to main content
Shameek ConyersData ScienceResumeContact

Dashboards for a Hospital's Physician Marketing

The hardest data to measure was typed by hand into spreadsheets. That's where the work started.

Client
A large hospital with a multi-million-dollar marketing program aimed at doctors, while I was at Ringer Sciences.
My role
Data scientist. I brought the hand-typed records into BigQuery, wrote the entry checks, built the dashboards, and presented the findings.
Result
The hospital used the dashboards to decide where to move its marketing budget, and kept paying for them.
How we know
The client's own budget decisions and its continued paid use. No percentage was measured.

Result: the dashboards became a recurring paid product for the client and informed how it reallocated its marketing budget.

The problem

While I was at Ringer Sciences, a large hospital client was running a multi-million-dollar marketing program aimed at doctors. Its goals were hard to put a number on: recruiting physicians, keeping them, and building the hospital's standing in the profession.

The campaigns ran across research journal ads, targeted Twitter and LinkedIn, PubMed, surveys, and the hospital's website, with Google Analytics late in the funnel. The ad platforms reported on themselves. The journal ads didn't, because that data wasn't available through an API, so the client's staff typed those placements into Excel and Google Sheets by hand. Nobody could see the whole program in one place.

What I built

Getting the hand-typed data in. I pulled the spreadsheets through the Google Sheets API into BigQuery on a schedule, so the team kept working in the sheets they already used. Hand entry brings typos, inconsistent names, and shifted columns, so I wrote checks that caught entry errors before they reached a report.

Joining it with the platforms. I joined the journal records with the ad platform data, so journal placements sat beside the digital campaigns in one set of tables instead of in a separate spreadsheet.

Dashboards for two audiences. The dashboards refreshed daily to weekly. They served the marketers running the campaigns, and physicians and clinical decision-makers who needed clear takeaways rather than the mechanics. I presented the findings back to the client.

How the client used it

It's still my favorite project because it ran end to end: messy spreadsheets in, data checks, modeling, and a readout the doctors could use. The client used the dashboards to decide where to move its marketing budget, and kept paying for them as a recurring product.

Related work