User data analysis to identify a data-driven persona

• Data Science Project: Understanding our users to better optimize user acquisition + conversion

• Pricing Models: Exploring user and brand partner monetization opportunities to identify opportunities for revenue optimization

PROJECT #1:

Objectives:

Leverage the current Presently data set of customers in order to:

• Identify our most valuable users (organizers) through data-driven personas so that we might be able to target new users more effectively
  • Including identifying signals for success vs churn
• Identify patterns in gifts on our platform so we might build better brand partnerships
• Understand our contributors profiles and what signals to look out for for their conversion to becoming organizers so that we can maximize our viral coefficient

Methodology 

• I plan on conducting a data analysis, creating data visualizations, and potentially conducting linear and logistic regressions all with the express purpose of attempting to leverage our robust dataset to extract insights to better move forward 
• I also anticipate a large portion of my work encompassing scrubbing and cleaning the data so that any sort of analysis can be done in the first place 
• Another potential portion of my work could be the exploration and merger of other datasets that could potentially be a valuable add to the project as a whole 
• I want to add value to this project and to do that I will mostly be leveraging existing skills that I have developed including python (pandas, numpy, matplotlib, etc.). There are also a vary of other packages and libraries that I should easily be able to leverage. Some areas that I could also explore include leveraging R, excel, and tableau and powerbi. I also need to refresh and learn more about what to do with incomplete and non-perfect datasets (best practices for a row that is missing 1-2 columns). 

PART 1: GOALS & DATA CLEANING

Questions to answer

• Personas & Use cases
  • Who is actually using Presently?
    • At a high level – friends, family, coworkers, etc.
    • Geographic concentration
    • Company size + organizer job titles
  • What are people actually using Presently for? (Most common occasions, gift vs card, etc.)
    • For which occasions did people use Presently the most for? Birthdays will probably be the most popular, but after that is workplace use, graduations, or any other event particularly popular?
    • I think some data visualization will be helpful here, potentially even for marketing purposes.
  • Which times of the year are particularly popular? This could also be variable to other circumstances like Covid, when we are pushing marketing, and other factors, but trying to see if there is any sort of pattern in usage when it comes to time.
• Feature Value
  • What percent of users used each feature we offer? We have a lot of optional ones but how often are they actually being used? This could present some opportunities for us both in terms of prioritization of future feature development, but also in terms of how to know what to market.
• Gifts
  • What is the average amount raised per gift and what does the distribution of those amounts look like?
    • For repeat contributors did their contributions change? Let’s say for example I was invited to two different Presently gifts. Did I give the same amount both times, or did the occasion change how much I was willing to give?
  • What is the average distance between when someone contributes to a gift and when the actual deadline of the gift is? (will not be able to control when the invitation was potentially?)

Patterns observed

• Spike in events + funds in Jan, Feb, March, and then down in April
  • What was happening in the world: maybe covid vaccines?
• Potentially increasing number of coworker gifts organized (unverified)
• …

Clean up 

• Ignore individual e-cards
• Put it missing Occasion / Gift details from gifts pre-march ish 
• Ignore obvious ‘test’ gifts or spam accounts
• BUT don’t delete rows; just empty them. If it happens though, it’s fine, we just won’t merge data sets
• ….

Data to add (new columns)

• Gender to ‘Sign up form data’
  • Why: To see if our primary organizer has any pattern so we can better target them
• Metropolitan area for each group gift contribution in ‘Contributions’ and cross-check with the organizer ‘Sign up form data’ to see metropolitan area for organizers (based on their payment)
  • Why: to see if our primary organizers tend to cluster anywhere so we can better target those cities / regions
  • Note: group cards and organizer sign up have no zip code data; only group gift contributions 
  • Also filling in states based off of the zip codes that were provided
• Group category (coworker, friends, students) in ‘Sign up form data’
  • Why: To see if our primary use case is coworker gifts so we can better serve our ideal user and influence pricing strategies
  • How: Some self-labeled; others did not. If most contributors for that gift used a company email address, it’s likely a coworker gift. 
• ….
• Excluding: 
  • Kids vs adult – not important right now

Results

7/13/21

Added in assumed genders for Sign Up Form data, found 140 Male recipients and 358 Female recipients. Also was telling that there were a small number of gifts that were organized not with just one person in mind which is why I infer we rolled out the group gifts.

7/14/21

For Zip Codes we are not requiring a 5 digit number, not a required entry, non-US entries, can they put anything?

Local Area: 

• Plano, TX
  • 19 days between first and last payment

Expanding

• Wesleys
  • Locations
    • Connecticut x2
    • Long Island x3
    • Minnesota
    • NYC