Analytics for Everyone (Growth Design Monthly) RESEND

July 2019Dear Readers,Sound data is the key to sound decisions. I’ve found data to be an invaluable way to inform what to build and to reflect back how well each product release met its goals. But good data is not automatic. You have to work to have strong, useful data.As an industry, it seems like we talk a lot about data and rarely about how you actually work with it. In this issue, I give a detailed overview of how to set up and work with product analytics. If some of the terms are new to you, I included a short list of definitions towards the end.Lucy in the Sky with Data,Lex

Choosing ToolsWhen looking for analytics tools, you want to choose something that: Your team finds easy to set up and maintain Your team finds easy to access (no complex permissions logic) Your team finds intuitive to view data and pull reports Seems like a stable tool that will continue to exist and that other companies trust Your company is able to afford for the amount of data you’ll be sendingI like to use Segment’s integration catalog to find the latest and most popular tools.

For small teamsIf you’re building a brand new product or if your company doesn’t have a data team, I recommend a self-serve product analytics tool. You’ll want a tool that doesn’t require special data expertise to set up or to generate reports.My favorite analytics tool is Amplitude because it’s super intuitive, they ship updates often and they actually talk to their customers.Other options include: Mixpanel HeapI don’t recommend Google Analytics for product teams unless you have no money to spend because it’s very difficult to use in any remotely helpful way other than “oh cool, 50 of our users are from Norway.”For big teamsIf you’re working at a larger company, survey your team to see how well the current system is working. Do they have access to the data they need? Do they know who to go to for help? When was the last time they pulled a report?I find that design teams are often the last to get access to the information they need. If this is the case at your company, talk to your data team about getting access to tools and training. Usually, they’re excited to help. If that’s not the case, talk with your partners in product and engineering about implementing a self-serve tool. I’ve had success with this even when localized to just one slice of a product in a large product organization. See the section above on self-serve tools.

Instrumenting EventsInstrumentation is the code your team writes that specifically calls out what information should be tracked about users and their activities.As you think about instrumentation, three important considerations: Think about what you want to learn. Your analytics should stem from your goals. Don’t track everything you build. Only measure what you’re planning to act on. Be especially mindful of what you track about people. Personal data is abused (see: Facebook and Amazon) and is generally not that valuable in an analytics context.For example, let’s say you work for Airbnb and you want to learn how many people are booking an experience. In order to understand that, we need to track each part of the critical path to book an experience. We’ll start when a user submits a search on the experiences page of Airbnb.com. We could call that event searched_experiences.

Maybe you also want to know what they searched for in which case you might fire a property on that event like search_term. You can have more than one event property per event.Your event would then look like this:Event Name: searched_experiencesEvent Property: search_term (where value is a string)Fires When: User submits a search in the search bar on an experience pageIt's helpful to think through all of this instrumentation as you're writing up user stories or experiment plans. Sometimes, I see Designers and even Product Managers leave the instrumentation to engineers. If you do that, make sure you understand what they named the events and when they fire so you can properly read your reports.

Creating ReportsThere are all kinds of ways to look at data but I use two chart types on a daily basis.Segmentation ViewA segmentation view or unique event count shows you how many users performed an event over a period of time. It’s useful for seeing stability of your app, overall growth in activity or overall declines.To create this kind of report, search for the relevant events in your analytics tool, decide on the cadence (daily, weekly, month) and the time period.

Source: Amplitude Demo

Funnel ViewA funnel starts with how many users performed a single event and then counts how many of them continued to the next step and so on for however many steps you like. They can be somewhat limiting but are great starting places for understanding how your product is doing. And luckily, tools like Amplitude let you dive in further and create reports and user cohorts from any step.To create this type of report, add each event along an important path in your product. Consider how long it should take to complete that path. This will be your conversion window which indicates how long someone should take from first to last step (for example, 30 minutes or 30 days). Each tool has their own way to decide whether or not to include a user based on the steps and conversion window. Read more about how your specific analytics tool handles funnel attribution in their documentation.

Source: Amplitude Demo

Business Intelligence Tool (aka “BI Tool”)A tool that visualizes data from another source. For example, from a data warehouse or simply from a spreadsheet. Usually considered more reliable than product analytics when they are drawing from the "source of truth" aka your app's underlying database. These tools are often used to reflect higher level business metrics like revenue, monthly active users or new accounts created. Examples: Looker, Mode, Tableau, Chartio, PeriscopeData TaxonomyA logical list of what’s getting tracked into the analytics system. This is the list of all the events that fire and when.EventAn action taken by a user that is being fired into an analytics system.InstrumentationThe code your team writes that specifically calls out what information should be tracked about users and their activities. For example, track every time someone hits this button (“event”) and track this information about that person (“user properties”).Product Analytics (aka “behavioral analytics” or “analytics”)A system that captures and reflects information about product usage and its users. Usually used to track pages, interactions and user demographic information. Can be considered as reliable as data warehouse information when using solely backend data and properly quality checked. Examples: Amplitude, Mixpanel, Heap, Google AnalyticsUser Properties (aka “People Properties” or “Profiles” or “Audience”)Information that describes a user. For example, what kind of subscription they have, where they are located or when they created an account.

• Find Tools using Segment’s Integration Catalog

• Segment’s Analytics Academy

• Amplitude Analytics Library

• A 5 Step Guide to Sustainable Analytics Instrumentation

Yesterday, All Our Data Seemed So Far Away

Data can truly be very approachable and easy to work with. With the information above, you can get more involved in analytics and better inform your design work. If you can see where customers are bailing and understand the impact of every release, you will create stronger products. Don't be shy about asking for what you need. Get in there and make it count!

Lex is a Product Designer focused on growth at The Black Tux. She's the primary organizer behind GrowthDesigners.co. Interests include product analytics, AB testing and making Spotify playlists for her friend's cats.

EventsThe Growth Design SeriesJuly 29-Aug 1 (San Francisco)This event is (maybe) the first ever Growth Design conference! It's an after work conference and you can get a day pass or stay the whole time. 

JobsProduct Designer, Growth (Contract)Bleacher Report (San Francisco)Product Designer, GrowthTwilio (San Francisco)Senior Growth DesignerVimeo (New York City)

This is the Growth Design Newsletter. We produce monthly issues on topics relevant to growth design. Want to write or edit an issue? Hit reply!In the August issue

Customer Development