3. ● Analytics 101
● The Data vs Gut Paradox
● Validating user journeys
● Retrospective event tracking vs planned event tracking
● Building your analytics infrastructure: Segment, Heap, Pendo, Totango, Kilometer.io, Full Story, Jaco
and more.
● Don’t be afraid to kill features
● Being data driven is not easy.
Main takeaways
15. ● Analytics 101
● The Data vs Gut Paradox
● Validating user journeys
● Retrospective event tracking vs planned event tracking
● Building your analytics infrastructure: Segment, Heap, Pendo, Totango, Kilometer.io and more.
● Don’t be afraid to kill features
● Being data driven is not easy.
Main takeaways
Focus on Clarizen’s platform capabilities.
Worked as a senior delivery consultant managing enterprise deliveries and gained his certification as a PMP.
Omri has worked with global companies such as Unilever, McKinsey, KPMG, Amdocs and more.
Omri began his career path in 8200, the intelligence unit of the Israeli army.
He holds a first degree in Information Systems Management, and an MBA from Tel Aviv University.
the systematic computational analysis of data or statistics.
“Let’s define Analytics as consisting of the technologies, applications, people, and processes that allow a firm to transform their product data into actionable insights.” Accidental Product Manager Blog
“While unsolicited customer feedback helps product managers understand only the vocal minority (the customers who voluntarily share their thoughts and ideas), data provides an objective look at every customer, making it an excellent way to learn more about who your customers actually are and how they’re using your product.” UserVoice Blog
Events - Events are user interactions with content that can be tracked independently from a web page or a screen load. Downloads, mobile ad clicks, gadgets, Flash elements, AJAX embedded elements, and video plays are all examples of actions you might want to track as Events.
Metrics – A metric is a standard of measurement. For example, when measuring the popularity of a website, we would need to use metrics such as # of visitors, # of traffic sources and etc. But here’s the catch. It is essential to differentiate between actionable metrics and vanity metrics. Actionable metrics are Accessible, Actionable, Comparable and are usually represented by Rates or Ratios.
Segmentation – A segment is a subset of your analytics data. For example, of your entire set of users, one segment might be users from a particular country or city. Another segment might be users who purchase a particular line of products or who visit a specific part of your site.
Funnels – Funnels are a way to visually measure how customers move through any series of events. Funnels can help answer questions like “Which A/B test got more visitors to convert to a paying customers through our free trial page?”
Cohorts – A cohort is a group of people who share a common characteristic over a certain period of time. We could group customers by how they were originally referred to your business and track how much money they spent over time. In the case of Saas software, cohorts can be used to track new user engagement by viewing how often users log in after signing up. You can rate your product features’ “stickiness” by how often customers use them and show customer loyalty in repeat purchases and renewals.
The first problem I encountered, was how I could get a clear picture of what our users are doing in the system. To better understand our customers, I needed to know what features of the products they used, how often they used them and even more important, what they features they weren’t using at all.
A clear picture of what your users actually do in your system. To better understand your customers, your need to know what features of the products they use, how often they use them and even more important, what features they do not use at all.
As product managers, we make many decisions on a daily basis.
The type of decisions may vary from strategic decisions such as feature prioritization to small decisions such as the size of the font in the new user registration form.
The problem with making multiple decisions is the mental effort entailed in the process. To ease the mental effort needed in making decision we can use Data or Gut. Or both.
We need a mix of both gut and data.
“What is needed is a combination of gut feel (a hypothesis based on deep knowledge of the customer) and validation through well-constructed research that is precisely designed to test that hypothesis”.
The tricky part is the validation process.
Customer module example
“In data science, intuition and analytics work together in tandem, each informing the other. First, intuition guides analytics. Second, analytics informs intuition.” Steve Hillion, co-founder of Alpine Data Labs
When encountering the conflict of data vs gut, I suggest to ask a few questions:
What is the problem we are trying to solve ?
Whose problem is it ? Customer related? If it is customer related - to which customers?
What are all of the possible actions you will take and what are their conditions?
I suggest using a decision tree for that. (If the feature funnel ends in less than 10% “conversion”, we need to try a different user journey because not enough people actually end up using the feature we designed”)
Working with data requires expertise and time. Data wrangling and ETL (Extract, Transform, Load) processes make around 70% of the time of a data scientist.
When you, as a product manager, invest your time working with data - you do not spend your time interacting with customers, prioritizing or developing a new idea.
Events - Events are user interactions with content that can be tracked independently from a web page or a screen load. Downloads, mobile ad clicks, gadgets, Flash elements, AJAX embedded elements, and video plays are all examples of actions you might want to track as Events.
It is hard planning ahead how to measure the user journey success. It is easier to do so in retrospect. Why?
Because throught the development process things change, your encounter new challenges/bugs/limitations, new change requests and the original design changes. What happens with version 2.0? Would we need to create user journeys again ?
Let the developers actually write the program, there are solutions today that you can use in retrospect without wasting the developers time.
I encountered 4 limitations/constraints in regards to analytics tools:
Specific integrations
Google Analytic sampling
Mobile analytics
R&D Load
Specific Integrations
Segment (Data Collection) - One integration
Analytics Sampling - GA’s free version is limited to up to 10 million hits(events) per month per property and sampling occurs automatically when more than 500,000 sessions occurred. Unless you want to pay 150,000$ for the GA Premium version
Don’t send event on every interaction
Mobile Analytics
Load on R&D.
Setting up the infrastructure
Connecting to Segment
two different workspaces (Prod and QA)
User/Group Traits
GA
Top of the funnel analytics with Google Analytics
Data precision in case of sampling
Heap Analytics
you do need to think in advance of all the events you want to track.
Heap will automatically calculate the number of times users clicked on the element in the last X days.
The flexibility that heap provides by selecting events on the fly, makes the life of the product manager easier and frees R&D resources to focus on development.
The way I suggest using Heap is to make an assumption as to what should be the user journey. After you have a clear user journey in mind, validate the assumption using a funnel chart. Let’s say the assumption is that a user lands on the homepage of the application, from there the user clicks on the navigation panel, then the user clicks on the project modules and selects a specific project. It would be interesting how many users actually go along that path and how many drop and where they drop. You can see the result in the screenshot below.
Customer Adoption
Totango you can see the number of active users per customer, % license utilization, most common activities, which users login and more.
Gainsight in addition
Two systems to triangulate the data > more confident
Next Steps
Data Warehouse using Amazon redshift.
Statistical Models
Why to remove a feature? Giving up on features can improve performance, reduce maintenance efforts, simplify the UI, and make room for new features that convert more users.The main takeaway is that you should not be afraid to give up on some functionality bells and whistles, if they don't make an impact on your business KPIs.
Giving up on a feature is similar to organizing your wardrobe – nobody does it. And when you have to do it – it’s a tedious process. Nobody wants to get rid of an old pair of jeans. It's the same with features. Features were previously defined, developed, tested and deployed... and now you want remove them? Wix had this issue to deal with, and they decided to check if some features could be removed. The strategy was to test the impact on conversion (new users->premium users) by conducting A/B testing. The reality showed that removing particular features made absolutely no negative impact on the business KPIs.
Giving up on a feature is similar to organizing your wardrobe – nobody does it. And when you have to do it – it’s a tedious process. Nobody wants to get rid of an old pair of jeans. It's the same with features. Features were previously defined, developed, tested and deployed... and now you want remove them? Wix had this issue to deal with, and they decided to check if some features could be removed. The strategy was to test the impact on conversion (new users->premium users) by conducting A/B testing. The reality showed that removing particular features made absolutely no negative impact on the business KPIs.
The main takeaway is that you should not be afraid to give up on some functionality bells and whistles, if they don't make an impact on your business KPIs. Giving up on features can improve performance, reduce maintenance efforts, simplify the UI, and make room for new features that convert more users.
Killing a feature that you designed, fought for, and launched is hard. But your product needs to maintain focus in order to maintain value.
the systematic computational analysis of data or statistics.
"content analytics is relevant in many industries"
information resulting from the systematic analysis of data or statistics.
"these analytics can help you decide if it's time to deliver content in different ways"