Analytics why, when, what, where, how & gotchas @niallroche [email protected] Get in touch Niall Roche
DECIDE
why use analytics?find out who users are
find out how they are using your product (or not)
find out what they are/are not interested in
saves wasted effort or can try differently
either way you have learned something!
can help with innovation accounting
can help improve product with A/B testing
DECIDE
provable learnings
compare the data to measure effectiveness new features
cohort analysis
identify new ideas from what people want to do, but can't
pretty dashboards and spreadsheets
DECIDE
know your userswhere geographically (may not be accurate)
how did they get there (referrals & campaigns)
gender & age (may be implied)
what devices
if from mobile, how does the website look on those devices
what languages are they using
when are they using it (any interesting trends?)
DECIDE
creates lots of data creates data points for further analysis
a data point to make predictions against
can dangerous!
can apply Machine Learning
can be a source of value and revenue
could be used for targeted advertising
DECIDE
without analytics
no idea how users are using your site/app/device
hard to prove to investors and prove metrics
how to measure against KPIs?
how to know if something new or different is working
DECIDE
when?start tracking as early as possible (no really!)
missed data can be lost value
especially when it shows something is wrong
learnings as important as the technical aspect
maybe more!
figures don't lie (ish)
figures are not everything (converse with users!)
DECIDE
what?decide what to track
and how to interpret
what are the user journeys to focus on
what goal performance to measure
look crash reports exceptions and bad things
don't forget about some feedback!
DECIDE
Industry Baselinescompare with other apps in same category
user baselines sourced from the industry
• app annie
• visionmobile
• Flurry
• apptopia
DECIDE
Examine Sessionslook at session length
compare length with journeys
is page wait time an issue?
where do users get bored move on?
user behaviour may be in and out vs linger
supermarket vs convenience store
use at bus stop vs bus
compare behaviour of different types of user groups
Examine Funnels
DECIDE
where?
identify milestones along a journey
different decision points for user
all major events
should be mapped to KPIs and metrics
if in doubt - track as much as possible!
DECIDE
Metrics• Acquisition
• Activation
• Retention
• Referral
• Revenue
• many many more…..
DECIDE
cohorts• cohort analysis
• match campaigns to cohorts
• ignore vanity metrics!
• no of page impressions + signups not as important as engaged users
• can always buy users to visit your site
• what campaigns convert into paying customers for lowest acquisition cost
• useful when calculating lifetime value of customers
DECIDE
how? weblogs
limited but useful
event trackers
javascript event tracking
cookies to track repeat visitors
session capture & replay with heatmaps
DECIDE
toolsGoogle Analytics
MixPanel
Liquid
Flurry
Crashlytics
AppSee
UXCam
AppAnalytics
UsabilityTools
Localytics
Segment
Kissmetrics
customer.io
mparticle
many more
DECIDE
A/B Testingimplementation options
AB Tasty
Convert
Optimizely
Unbounce
many analytics tools have simple implementations built in too with reports on variance
DECIDE
potential pitfalls
DECIDE
App pitfalls
DECIDE
app launch eventsapp icon
spotlight
push notification
from other app / device
url scheme
intent (Android)
Document Handler
proximity trigger
geofence
beacon/iBeacon/eddistone/NFC
system event (low battery)
background mode may bypass initial start logic
DECIDE
multiple pathsA -> B -> C is not always straightforward
may be multiple routes to same point in app/site
how did the user get there?
Tab / Swipe / Press / Gesture
from Hamburger menu, context and other menus
tracking context of how an event was reached
DECIDE
• tracking signed in web users mush easier than using cookies as they may not span different devices
• identify hits from bots / screen scrapers / web search tools / developers & testers / uptime checkers
• page load time measurement (especially in areas where there are slow connections such as emerging markets, and poor connections)
• look for different behaviours and trends
• compare session length vs onboarding
• trying to determine reasons for dropoff
ga('send', { hitType: 'event', eventCategory: 'Video', eventAction: 'play', eventLabel: 'cats.mp4'});
//for Single Page Appsga('send', 'pageview', '/new-page');
Category -> Action -> Label + extra if needed
DECIDE
technical problems• some technical issues for websites
• leaving page on form submit
• careful with placement of code
• missing images, load events may not fire
• add timeouts for error handling
DECIDE
actioning with dataplatforms like Liquid (onliquid.com) can adapt an app experience based on data
identify variables to change inside the app
trigger an actions on a funnel of users (e.g push notification)
set rules to automatically intervene based on trends
personalised experiences