Pandemic Proto-Personas

Created using Google Analytics and Hotjar Data

5 min readOct 7, 2020

A short write up of how I used quantitative data from Google Analytics and qualitative data from Hotjar to create some Proto-Persona for our Sportsbook product. Note that I used data from between January and July 2020, during the pandemic. As such I expected to see some data anomalies, especially from the quantitative data because a lot of the top sports where not active during this time.

Quantitative data

There is a huge amount of data captured in this tool, the key is knowing/learning where the useful stuff is hidden. My analysis started by looking at some basic attributes: Location, Age and Gender.

The first finding here was that 3 quarters of users were male and 1 quarter were female, from this I decided that I would create 4 personas, hoping it would give an overview of all our users.

I was also able to identify popular locations and age groups, knowing these allowed me to add secondary criteria to the data to make further discoveries. For example, cross-checking the age and location data showed me how old our users were in each of our most popular countries.

So now I had the Location, Age and Gender data for the 4 persona. I would use this later as the secondary criteria when looking into other types of data.

An example of this “other type of data” is what google calls ‘Interests (in-market)’, this data is showing the potential things that these users may be looking to buy in the near future based on their current data/browsing habits.

Looking at the ‘Interests (in-market)’ isn’t helpful on its own, so what I did was to add a secondary criteria of Age, I then filter out to focus on each age group. Now I know what our 18–24 year old users are currently shopping for, which is slightly helpful. I can add another level of analysis by adding Location, to do this I have to extract the data and use another tool. In this case I was using MS Excel, which allowed me to put data side-by-side to spot trends.

This is now giving me some useful data that I can use in my persona, for example I have identified what our users in Brazil who are aged 18–24 are shopping for.

I continue this cross-analysis, focussed on the same 4 profiles identified in the first step. Looking at any data in the tool that may be suitable for personas, such as, language, personal interests, social platforms, device types.

I now have 4 profiles with lots of general data about each type of user, from here I start to look into more specifics of how each of these types of profile use our site. Including, but not limited to: which pages are they looking at, which pages do they complete a transaction on, and are they using any of our advanced features.

All the data points are pulled together for each persona and use it to build out each persona. Hopefully you’re still following, and find that helpful. That is the basics of how I analysed our site’s qualitative data, now slightly more interesting…

Qualitative data

I’m a fan of Hotjar, I find their heatmap tool insightful and love the ‘rage click’ feature that lets you see screen recordings of “unhappy” users who click repeatedly on your site.

For this project I used the Polls tool on Hotjar. Users were first asked to rate their experience out of 10 (I’m not using that data for this project) with a follow up open-ended question asking why they chose this score.

I extracted all the open-ended responses from the first 6 months of this year into excel. Looking only at the 4 locations of each personas. And carried out some data cleansing, like removing blank answer or incomprehensible ones. As well as translating some of them into English.

As I am working remotely and I wanted to get some help analysing it from my UX team I decided to upload all the responses into Miro post-it-notes (this is actually very simple to do, definitely quicker than writing all the notes out).

Affinity map analysis of Hotjar Qualitative data using Miro.

Positive comments on the left and negative on the right, the results were broken down into 3 main categories:
• Betting: which included anything about the betting experience or betting preferences of the user.
• Service: is things the users specifically liked or disliked about our product.
• Money: any financial related issues, good and bad.

The main categories are broken down into sub-catagories and the different colour notes represent each of the 4 different locations.

The trends from this analysis were summarised and could then be used to help build the personas.

Results

This was a worthwhile exercise and allowed the creation of 4 personas, however the data may be slightly different because of pandemic conditions and the results may be very different if the same analysis is run again during “normal sporting” conditions, I look forward to finding that out when we get through this.

Here are the four “pandemic personas” I created…