## Framing Good Questions > Reframe your analysis as a question to make it easier to comprehend. *Insight:* And if we can't frame our goal as a straight forward question, we are probably not yet ready to do the analysis. ### DAU: Like for eg. Daily Active Users (DAU): How many active users do we have everyday? Now we need to define this to validate, as in, what is an active user? - A user is active on a particular day if they have done anything on that day. - So any user coming on our platform and performing a task is flagged as an active user. - It also means, that if a user does 0 behaviour in a particular day, they were inactive. **All Events:** is used for DAU as we don't really need to know what behaviour the user did, if we select events based on a particular behaviour, then we will be measuring **engagement**. #### Sessions: > Number of users performing tasks distributed in the # events performed. - Sessions utilises the math operators to distribute user events in batches like: - opened the app and performed one event (order sandwich) - opened the app and performed two events (ordered sandwich and upgraded to premium) - and so on... #### Engagement: These ones are tricky as there no standard definition to what engagement means across all products and services. - Be creative and find meaningful inputs for an engagement metric. - It can include events which people wound't do accidentally, like ordering a sandwich, updating payment details, and so on. - Each of these behaviours would clearly represent an intent. - Tag these events as "engagement". - Make a report and add all the engagements as unique users. - Add formula: A+B+C+D - add a contextual benchmark for this metric - The formula should be a ratio, a ration of engagement divided by - numerator = number of engaged users - denominator = total population of users active on that day - Which is the DAU metric - So the formula would become: (A+B+C+D)/E - Here, E is the DAU metric and A,B,C,D are different meaningful engagements. ### Creating Clear Visualizations: >Charts helps us visualise the data so that we can see trends that would otherwise be difficult to find. **1. Bar Charts:** these are really good at comparing the quantity of things. - Can be used whenever we want to know the most or least of a particular thing and see the magnitude of difference between them. ![[PMing_Analytics_Bar Linear Graph.png]] **2. Pie Charts:** Pie charts are helpful when measuring quantity relative to a whole. - Useful when there is a clear and obvious maximum value. - Very Useful when looking at *financial metrics*, or *resource consumption*. ![[PMing_Analytics_Marketing Channel.png]] **3. Data Tables:** Useful when we need to stack ranks or sort different attributes within our table. - Useful specifically while viewing *ranking data*. ![[PMing_Analytics_Table VIew.png]] **4. Line Charts:** Special because they have two axis. - The horizontal axis is almost always time. (helpful in change over time || Growth Trajectory of a metric) ![[PMing_Analytics_Linear Line.png]] - Mixpanel offers allows us to customise the window and unit of time both, the days selector adjusts the range of the horizontal axis and the unit selector determines how far the points in the chart will be plotted. ![[PMing_Analytics_Last Five Days.png]] **5. Linear Graphs:** Shows the upward and downward movement between each data point. ![[PMing_Analytics_Line Chart.png]] **6. Cumulative Graphs:** Only shows the growth of a metric. ![[PMing_Analytics_Cumulative Line Chart.png]] ## Creating Dashboards - Collect and present all the gathered information into a neat, colorful and easy to understand package for people who are not data analysts or product managers. #### Rule of Thumb for Effective Dashboard: - Every visualisation should have a clear title and description. - The choice of chart type should be intentional to make it clear what the person who's looking at the data should care about. - The dashboard should be setup to tell the story of our metrics. **Interesting feature:** Email digest for stake holders and Product Team.