Data Analytics for Product Management

GitHub repo: https://github.com/pepenunez/product-analytics

Time frame: 1-week

Introduction

For this project I wanted to validate that data can have a direct impact on product management and product decisions. To do so, I've chosen a digital product that I regularly use, know and that I could think of a potential challenge that they could have. Consequently, I've chosen Canva. Canva creates unique templates that users can use and adapt. After doing some research I've found that they have +100 sub-categories with a total of +60.000 templates. With this project, I wanted to perform an analysis that could help Canva's template team to priotitise, sport trends and improve user engagement.

File structure

• /data:

  • clean-data:
    • sub-categories: This folder contains a .csv per each sub-category that we will be analysing.
    • categories-post.csv: Category, Sub-category, Keyword, Search Vol (min), Search Vol (max), Number of templates
    • categories-post.xlsx: Same as categories-post.csv but in excel format.
    • priotities.csv: Sub-category, Search Vol, Recommended templates, Number of templates, Templates to create
    • trends-all.csv: Date, Trend, Category, Sub-category, Keyword,Search Vol (min), Search Vol (max), Number of templates, Search Vol (avge), Search Vol, isPartial, index
  • temp-data:
    • canva.txt: https://www.canva.com/create-a-design html in txt format
    • categories-pre.csv: Category, Sub-category, Keyword

• /presentation-resources: This folder contains some of the resources that I've used for the presentation.

• 01_data-gathering.ipynb: The goal of this notebook is to gather, clean and manipulate all the data that I will be using during the project.

• 02_data-analysis.ipynb: The goal of this notebook is to explore the data, identify interesting insights and perform the analysis.

• presentation.pdf: This is the deck used to present. It includes: approach, main takeaways and recommendations.

• tableau: There are some graphs that are not in any other place.

Initial hypothesis

1. Is there a correlation between Search Volume (Google) and the Number of templates that Canva is offering?
2. Are categories displayed in Canva's website sorted by Search Volume (Google)?

Assumptions

• I don't have access to Canva's data, so I will use Google's search trends data and assume that it can be transferred to Canva's users interests.
• The greater the number of available templates, the higher user engagement.

Data sources

1. www.canva.com
2. Google trends (API) | Google trends - pytrend | pyp| https://pypi.org/project/pytrends/#interest-over-time
3. Google Adwords | Google Adwors - googleads | conda | https://github.com/googleads/googleads-python-lib *Only available with a Google Adwords account

Main findings

There is a moderate positive correlation between the Search Volume (google) and the Number of templates. The model can explain 35% of the variance of the Number of templates. It is recommended to analyse which are the other factors that are influencing to the correlation.

Recommendations

• Prioritise and Analyse in-depth the subcategories in 'priorities.cvs'.

• A/B test displaying category by Search Volume (https://www.canva.com/create-a-design)
• Adapt priorities and displays to seasonality

Next steps | Further analysis

• Classify actual sub-categories by growth phase so we could visually detect which are the categories where we should be prioritising.
• Spot new trends by searching for similar keywords

Other links

• Tableau
• Canva