All of Us

GRAS HOPE Products

by MakeUsWell

MakeUsWell has developed a suite of software and analytics driven products for the COVID-19 vaccination market. The products were developed with:

  1. Data from ~100+ sources of a wide variety
  2. Millions of data points
  3. Structured and Unstructured data
  4. Surveys and conversations with customers over web, sms, email, phone, and in-person channels
  5. In the right time period, ~ December 14, 2020, the start of US vaccinations to ~ April, 2021 and ongoing

These products fit within a GRAS framework.

This framework considers the major forces and dynamics within the market.

GRAS = Acronym for Geography, Race, Autonomy, and Social


Person level Vaccination Score.

Applied to the United States Market.

Measures the probability that with the right marketing message—through the HOPE framework, to the right segment, at the right time—we can cause them to choose to get a Moderna or Pfizer vaccine. Within the next ~7-60 days.

How it works:

  • Each person gets a score of 1-100, or 100-1,000.
  • Say Bob's score = 70
  • And Diane's score = 80
  • For a non-vaccinated population a score of 70 means that the person has a 70% chance of getting vaccinated.
  • But a score of 80, may mean the person has a 10X higher chance of getting vaccinated. So it's not (80-70)/70 = 10/70 ~14.23% higher but 10X higher.
  • The scale may be either a natural log or a base 10 log scale.
  • ~25 - 85% of the US population can be scored with some degree of significant accuracy. And receive incremental positive lifts over today's ways of doing nothing or doing random things.
  • Typically, for populations in the big metros, we can score 65 - 85%.
  • Places like Alpine County California, or midway between Des Moines and Minneapolis, is where we can currently only score ~25-30% of the population.

Decision Trees

In certain cases we'll deliver decision trees to be used along with the score.

2 Factors improve:

  1. Whether MUW can score the population.
  2. How accurately can we score and predict.

How many variables we get from the customer on their existing customer database, such as age, income, behaviour, gender, political affiliation.

So what's our fallback, when we can't score an individual?


We can segment into 4 non-overlapping groups

  1. Strong Yes to Vaccines
  2. Strong No to Vaccines
  3. Undecided or maybes likely to transition to a yes
  4. Undecided or maybes likely to transition to a no

HOPE Framework

  • HOPE is an acronym for Hemingway, Orwellian, Paternalism, and Elide Partisanship
  • Use Natural Language Processing (NLP) algorithms—both horizontal from big players like Google and our customized ultra-specific ones for the vaccination vertical—and match them with different GRAS segments.
  • The Gras segments are highly dynamic and volatile given the macro environment and micro human behaviour in 2021 America.
  • This is all about communicating with the right language.
  • Every word matters, when "Biden showers people with stimulus causes controversy".
  • The wrong language will cause people to not get vaccinated themselves and embolden the anti-vaxxer forces.