Vaccine access: where Pfizer should act next

A team project for the MBA Global Wellness Visual Analytics assessment (February 2026). The brief put four of us in the role of Pfizer's new analytics team during the pandemic. We had to use real COVID-19 data to show whether vaccines worked, how distribution was changing, and where demand would come next, then pitch it as a data story in Tableau.

Our answer was the title of the talk: what if the difference between life and death during a global pandemic was not science, but access? Watch our presentation on YouTube.

1. What is the problem statement?

The vaccines were ready, but they didn't reach everyone at the same time. Pfizer needed to understand the current landscape and where to act next:

• Uneven toll. 7,034,280 deaths were reported worldwide. The Americas (AMRO) recorded the highest regional toll at 1,174,676. Africa's low reported count reflects gaps in surveillance, not protection.
• Uneven coverage. Vaccine coverage reached 46% in Europe but only 8% in Africa, even though willingness to be vaccinated was comparable. It was a map of purchasing power, not health.
• A gap that grew. Through 2021 the Western Pacific surged while Africa's line stayed flat. By late 2021 the gap between regions was wider than at the start of the year.
• The question for Pfizer: do vaccines prevent deaths, where is distribution stuck, which countries will need doses most in the coming months, and what should Pfizer do about it?

The distance between Chile's outcome and Argentina's was never chemistry. It was contracts, logistics and proximity.

2. What are the tradeoffs considered?

• Purchasing speed vs need. Doses went to the countries that signed first. Chile signed early and became the world's fastest vaccinator, at 1.08 doses per 100 people a day by March 2021. Argentina hosted Pfizer's Phase 3 trials and then couldn't buy the product. We argued allocation should follow need, not who buys fastest.
• How to weigh "need". We built a Vaccine Need Score: coverage gap (50%), death burden (30%) and vaccination delay (20%). Putting coverage first favours prevention for the future; weighting deaths more would favour where the damage already happened.
• Reported data vs reality. Under-reporting in parts of Africa means real deaths and delays were probably higher than the data shows. We kept the reported figures but flagged where they understate the problem, rather than adjusting them ourselves.
• A forecast vs a prediction. Our Tableau forecast (exponential smoothing, 95% confidence interval) projects up to 200,000 more deaths in six months without intervention. We presented it as directional, not as a precise prediction.
• Central supply vs local manufacturing. Pfizer's central production is proven, but the regions with the highest forecast need are the farthest from it. Meanwhile AstraZeneca (through Serum Institute licensing), Moderna and Sinovac were building local partnerships and supply chains, and with them long-term loyalty from health ministries.

3. What is the solution proposed?

We built the story in Tableau from WHO and Johns Hopkins CSSE data, WHO's vaccines-used data and the course vaccination dataset, then followed it to three recommendations.

1. Show where people died: reported deaths by region, and why Africa's low count is a data gap.
2. Show who got vaccinated: coverage by region, and how the gap widened through 2021.
3. Prove vaccines worked: we clustered countries into three groups (high coverage with suppressed deaths, transitional, and unprotected). Every regional trend line points the same way: more doses, fewer deaths per 100,000.
4. Measure the delay: days between global vaccine approval and each country's first dose. The longest delays overlap with the highest death burden.
5. Map the competition: where Pfizer was absent, a competitor was building procurement relationships.
6. Forecast and score: the six-month forecast, and the Vaccine Need Score against a 70% coverage target. 23 African countries showed a gap of more than 60 points, and the critical cluster sat entirely in Africa and the Eastern Mediterranean.

Our three recommendations for Pfizer:

• Decentralise manufacturing. Replicate the Pfizer–Eurofarma model in Brazil (100 million doses a year) across sub-Saharan Africa and Southeast Asia.
• Allocate by need, not purchasing speed. Use the Vaccine Need Score as a live allocation tool in quarterly supply planning.
• Protect the most vulnerable. Invest in booster programmes (doses 3, 4 and 5) paired with clinical-support protocols for cancer patients, older people and immunocompromised people.

4. Any other improvements I could have done?

• Stress-test the Need Score. I'd run the ranking with different weights to show which countries stay at the top whatever the weighting, which would make the allocation case much harder to argue with.
• Correct for under-reporting. Using excess-mortality estimates instead of reported deaths would give a truer picture of Africa's burden.
• Separate cause from correlation. Countries that vaccinated early also differ in age, wealth and testing. Controlling for those would make "more doses, fewer deaths" stronger.
• Back-test the forecast. Checking the model against a period we already had data for would show how far to trust the six-month projection.
• Build the business case. The recommendations are strong on impact but light on cost. I'd add what local manufacturing and need-based allocation would cost Pfizer, and what they'd earn in long-term government relationships.