Driving Consumer Insights With Mobile Analytics

Increasingly, your audience is spending time interacting with your brand on mobile devices every year. And hopefully your company has at least started to build out a mobile strategy to provide a better experience for your users.

But what about tracking success? Did you know that measuring, analyzing and reporting insights for mobile can be quite different than what you currently do for desktop or laptop users?

The Driving Consumer Insights With Mobile Analytics at SES New York 2013 covered a lot of great information around mobile usage, creating a mobile experience and tracking for success.

Interesting Mobile Stats

For some background, these are the online platforms brands are using:

which-platforms-are-brands-using

Why this is all so important?

what-people-doing-facebook

Looking at Facebook usage specifically, it's staggeringly huge (20 percent of all pageviews are on Facebook):

  • 40 percent of users' time spent on News Feed
  • 250 million photos uploaded daily
  • 2.7 billion Likes every day
  • 37 million pages with 10+ Likes
  • 425 million mobile users
  • 43 percent male vs. 57 percent female
  • 20 minutes time spent per visit

When you start to plan your mobile strategy, you have to think about what content you want to leverage, and how you're going to leverage it. You can start to come up with really interesting and potentially engaging situations for your audience based on usage patterns and test them out over time to see what works for you.

On average Americans spend 2.7 hours per day socializing on their mobile device. By 2014, users will access the Internet on mobile devices more than on desktops and laptops. So if you haven't started thinking about how to execute on your mobile strategy, including how to measure performance, the clock is ticking.

Mobile Web vs. Native Mobile App

multiple-devicesThom Craver (@thomcraver), Digital Marketing Strategist and consultant, kicked off his information-packed session by recommending that you start your mobile strategy where you always should with any major project – with your core business objectives.

Once you have those figured out, then there are three main options for your mobile experience:

  • Mobile Web: Mobile‐themed website designed for smaller screens. With this option, according to Craver, a safe definition is that it's likely going to be an exact or close copy to what your website currently is, only meant for smaller screens. So when planning to build out the experience, screen size can make quite a difference ‐ compare a little 4-inch smartphone to a 10-inch or even a 7-inch tablet screen. And landscape vs. portrait, depending on how the user views the pages, can make a big difference as well.
  • Responsive Design: Your website designed to auto-size based on browser. With this option, your website will auto-size depending on what device you user has and what she does. For example, when someone goes from landscape to portrait, how will it move and shift the content around?
  • Mobile App (Native): Native to the operating system (e.g., iOS or Android). Finally, with this option an app is usually something that's not necessarily web-based (although it can be), and is native to the device that you're putting it on. So if you're building an app for an iOS, Android, Windows Phone or whatever, it won't usually be written with HTML and CSS, but rather Objective C, Java or some other higher-end coding languages.
    That said, there's a way to "cheat" this by using what's called a chromeless web browser, or blank web browser with no buttons, toolbars, menus, or anything else. So what you can do is create a custom UI and wrap it around the browser in the native language of Objective C, Java, etc, and all the rest of the app's content comes from the web. The main issue with going in this direction is that the amount of data usage can really rack up monthly costs if your users don't have an unlimited data plan. So you have to be careful about how this is executed.

You'll often find internally that every single department has an idea of what the digital experience should be (i.e., the too-many-cooks-in-the-kitchen syndrome). When you start talking anything mobile, social, etc. your marketing team will say you need to do one thing, and communications will say, no, let's do another thing, and legal will say, actually we need to do something else.

Bottom line: your efforts need to be informed by your business goals and objectives.

TKTS Case Study

tktsCraver then highlighted his experience with TKTS, which from their website, allows you to "get fast, accurate, real-time listings of all Broadway and off-Broadway shows available at the world-famous TKTS Discount Booths in New York City."

Craver basically found that TKTS had a mobile app experience that allowed him to avoid standing in line for potentially several hours wondering if it was going to be worth it. He instead used the app to find out more about each show, if tickets were still available in real-time, etc.

What he found interesting was that the app was very simple and straightforward and allowed users to navigate through to the information they needed very quickly. This was important because it was much different than the experience on their website where the pathways were designed for a desktop/laptop user and not necessarily what people needed that instant.

It was pretty clear to him that the experience was directly tied to their goals of generating ticket sales.

What do you Measure?

Once you have your clearly stated business goals as they relate to your mobile experience, then you can start to understand what you should measure. Some examples are:

  • Mobile Web: pages, events, goals often on a separate subdomain, etc.
  • Responsive Design: pages, events, goals the same as any other website, etc.
  • Mobile App (Native): pages (screens), events, goals, ID-specific information, app versions, usage, crashes, devices and operating systems (OS) used, etc.

We can't actually determine if we're succeeding against our goals unless we start measuring. And with Mobile Web and Responsive Design it's pretty straightforward. But with Mobile Apps, you might have some very ID-specific things tied to users or individual device types.

To add a layer of complexity, if you're developing for Android you have other things to consider as well. For example, there are so many different manufacturers of Android devices, at so many different price points it's almost like buying a computer. You have low-end and higher-end options.

You have to look at the versions of your app that people have. Are users upgrading it over time when you have new releases? What hardware do they have?

You have to take all of these things into consideration, because if your app starts crashing or giving the user a bad experience you're dead in the water. So if that happens and you have version and hardware info, you'll be able to respond and fix issues much easier and faster.

Measuring Mobile Websites

Craver went on to say that mobile websites come in two varieties, usually something like:  m.yourdomain.com or yourdomain.mobi. So here's what you do to start tracking in Google Analytics (GA):

  1. Turn on subdomain or multiple TLD tracking.
  2. Set up profiles for mobile, non-mobile. Note: Keep original profile unchanged!
  3. Add filters to the new profiles (not the original profile!)

According to Craver, the trick to setting up tracking is that you have your main GA profile with no filters, a second profile with an Include filter just for your desktop/laptop site, and a third profile with an Include filter just for the mobile site. The reason you do this is to avoid losing any data (i.e., once you inadvertently exclude traffic from a GA profile, it's gone forever).

The next step is to turn on subdomain/TLD tracking. To do that, go to your profile page and select the appropriate option. Once you've created the profile, you can then can go back in to the setting and add the Include filter and any other filters you need. So then you're done setting up tracking for your site.

Once you have your profile set up, the GA reports for mobile websites are pretty much the same as what you see for your desktop/laptop site. Remember your mobile website is just a differently styled site. That said, you can look at really interesting stats (e.g., traffic by mobile device, operating system). You can also look at Advanced Segments (e.g., organic traffic, visits with conversions).

Mobile App Development Strategy

Know more about your audience. This should come first before you decide which platform you'd like to build your app for (e.g., iOS, Android, etc.).

One great way to find that out is to look at your analytics data for your desktop/laptop site: if your website indicates that most of your visitors have a certain device-type that they're using (e.g., iPhone, iPad, Android, Window Phone 7) then obviously start there for your app.

Craver mentioned Chromeless browsers under the Mobile App Experience above, and likely the two most popular related platforms to use are: PhoneGap and Appcelerator. They both take your website code and kind of throw it all together and wrap it up in the Chromeless browser. The benefits of PhoneGap and Appcelerator:

  • They're cheap.
  • All you need is a web programmer that can get through the framework.
  • They provide support.

Keep in mind these are not true mobile apps, rather they're actual web pages that are wrapped around what look like mobile apps.

Measuring Mobile Apps

So now getting in to measuring mobile apps in Google Analytics (GA), Craver mentioned they have mobile app tracking that's still in beta, but you can sign up for it.

google-analytics-what-would-you-like-to-track

Once you set up tracking for an app and sign up for the beta program, you then download the software development kit (SDK) – they have one for iOS and one for Android. All you have to do from there is take the SDK code and plop it into your app when your developer starts making the rest of the code.

What are you measuring in your app? As Craver mentioned, it's different than measuring website performance. Companies usually look at:

  • How Often the App is used by Visitors: Visitor intent and visitor metrics are great and it can tell you a lot more about how people feel about your brand, about your app and about the interaction with you as a company. Did they download it and install it and then launch it once and never go back to it? If so, why?
  • Averages Screens/Session, Session Length
  • Goals Completed
  • Other things you can measure: Did users have any problems? Did they send feedback to you? Did they crash (which is usually found in the Play Store or Apple Store), etc?

And then specific to the Google Play store, here are even more things you can measure:

  • How somebody found your app
  • How many downloads
  • How many installs/uninstalls
  • Comments (If somebody leaves a negative comment you can see which device they had and you can reply to these requests; you couldn't do that before.)
  • Crashes (by version, segmented out)

In-App Measurement in Google Analytics

According to Craver, the set of reports you get for tracking apps in Google Analytics is different than what you get for websites. For apps, you'll have your typical overview, visitors data, etc. But the left-side navigation is different with the following options: App Overview, Acquisitions, Users, Engagement and Outcomes.

google-analytics-app-engagement

Some additional things you can track for apps in Google Analytics include:

  • Version Control: this is very important ‐ how many people are actually using your app and then upgrading it when there's an update? 
  • Crashes: also something you should be looking out for.

Speaking a little bit more about crashes, Craver mentioned that if you have good programmers, they can create what's called exceptions to handle errors. So if there's a problem that's going to happen before the app crashes you can say "let's throw a more friendly error and slowly direct users somewhere else." This can really help with user experience.

Segmenting in Google Analytics

Talking a bit into segmenting, you shouldn't be looking at your analytics in aggregate ‐ if you do, it keeps you from getting to actual insights. So Craver recommended a few segments to consider when tracking apps:

  • Mobile vs. Non-Mobile
  • Tablets vs. Phones
  • Service Providers (e.g., Verizon vs. AT&T vs. Sprint)
  • Referrals
  • Location

Why Mobile Apps?

mobile-marketing-maturity

Next up, Marcia Kadanoff (@openmk), CEO & Founder of Open Marketing, started her portion of the session by advising everyone to use the above graph when thinking about why mobile apps are important and how your own organization can use them.

Kadanoff said that winners use native apps to get into mobile; thought leaders are looking at contextual targeting, and full customer management on mobile apps to put the state of mobile apps into context.

Web Analytics vs. Mobile Analytics

Kadanoff's sense is that people are struggling today to understand mobile app analytics. Here are some significant differences you need to know about when it comes to web analytics and mobile app analytics:

  • On the web analytics side you typically track by sessions and it's mainly done through cookies and JavaScript; whereas, with mobile app analytics the general setup might be similar, but much more intricate implementation is required. That's because you basically have to pre-define the events you want to track, which is not always easy ‐ you're likely going to have a much more intimate relationship with your coder or developer. Also, session tracking on a mobile app is primarily done through Unique Device ID (UDID) on Android and sessions on iOS.
  • On the web analytics side the human interface is obviously a keyboard and a mouse; whereas, on the mobile side the human interface is gestural and touch-based and there's just a lot more that can go wrong.
  • On the web analytics side the web measurement model is really centered around things like pageviews referrals, search, visits, etc.; whereas, on the mobile side the measurement model is less about referrals and more about engagement and loyalty.
  • On the web analytics side, unique visitors are tied to individual or server IP addresses; whereas, on the mobile side you can talk about sessions, but it's almost impossible to get to visitors because there's a lot of people that start on mobile but end up on desktop or laptop and vice versa. So it's really hard to get to unique visitors on mobile.

So because of these major differences between web analytics and mobile app analytics about 40 companies sprang up to do mobile analytics according to Kadanoff, which is a lot. That said, many of them also seem to do a great job of letting people look at vanity metrics, which often include:

  • Number of app downloads
  • Total number of sessions
  • Total number of first-time users

Unfortunately there isn't a lot of substance to these vanity metrics. They don't tell you why, how come, or what to do. As marketing professionals we're all be about customer insight and understanding customer intent when somebody's buying from you and why.

So with mobile applications, downloads are not enough. It's actually relatively trivial to buy your way into a lot of downloads for your mobile app, but that doesn't actually translate into using it.

how-often-apps-are-downloaded-and-used-once

An astounding 25 percent of people who download a mobile app only use it one time – and that number is going up over time, according to Localytics. And the number of people who download a mobile app and never use it? We don't know because the industry won't talk about that, Kadanoff said.

So what the mobile app industry focuses on is driving loyalty, which is defined as 3x uses in the mobile app industry.

Kadanoff thinks this is a function of two things:

  • More bad apps rushing in the market.
  • Just more apps.

There are a few studies that say people are spending a tremendous amount of time on their iPhones, tablets, Androids, etc. but they're spreading it out against the same number of applications – so new app comes in, old apps goes out, and they stop using them.

Kadanoff emphasized that good analytics starts by asking the right questions. She highlighted the AARRR model from Dave McClure, a VC in Silicon Valley, with his five questions:

  • Acquisition: How do users find you?
  • Activation: Do users have a great first experience?
  • Retention: Do users use it subsequently?
  • Revenue: How do we make money?
  • Refer: Do users tell others?

So Kadanoff applied this model to mobile using three different types of analyses.

1. Trend Analysis

Trend analysis is just what you would think ‐ it's one step up from vanity metrics. On Activation, we obviously want to know, "for the number of people that actually activated our products using it 1x - is that going up or down over time?"

For Retention, Trend analysis would cover the change in the number of loyal users over time.

For Revenue, she talk about changes in average revenue per user (ARPU) in Trend analysis, assuming you're monetezing your mobile app.

And then for Referrals, Trend analysis would cover changes in sentiment over time. For example, are your users leaving product reviews, referring the application to others, etc.? Referrals are an important part of discovery on mobile devices

2. Funnel Analysis

The next type of analysis is Funnel analysis, which basically allows you to see who, at Activation, downloaded the product, used it 1x and filled out a profile. And so you can say, "I'm getting a lot of downloads for the app" or "I'm not getting a lot of 1x users, and therefore I'm not getting a lot of people filling out a profile" etc.

3. Cohort Analysis

The highest life form in terms of analysis is Cohort analysis, said Kadanoff. Cohort analysis (which isn't that easy in Google Analytics) is like a survey you get from college, where it says "you were the class of 1996 and here's how many went on to grad school, here's how many went on to a job in consulting and here's what they're making." That's Cohort analysis.

Getting to Actionable Customer Insight

Many people call Flurry the granddaddy of the mobile analytics business. Apsalar is one step up from Flurry, said Kadanoff, and has a lot of great functionality.

If you're building an app today and want to find your instrument, always look at two products: Apsalar and Localytics. Localytics isn't free, but might as well be at $95 per month.

Mixpanel and Kontagent are two enterprise products and they're priced accordingly. One of the main reasons to step up to an enterprise analytics solution would be to integrate back in to your database (e.g., if you sell magazine subscriptions and you have a mobile app for you magazine).

When you compare all of these products, you'll notice that they're pretty close together in terms of what they're able to do. It really comes down to being a matter of personal preference as to which one you choose, Kadanoff said.

What you'll see with the free products (Google Analytics and Flurry) is that they don't particularly support drilldown into customer segments (e.g., women, age 18-34).

Kadanoff then talked a little bit more about Flurry. In terms of types of data available in Flurry, almost all the applications are modeled after GA. That said, Flurry excels in application engagement – you can define as many custom events as you need.

What's lacking in Flurry: the ability to do meaningful segmentation (e.g., meaningful meaning women, age 18-34). It's very difficult to get to customer segments. There isn't support for cohort analysis. Also, you don't have customer-centric funnels in Flurry.

TheFind ‐ A Case Study

Kadanoff then went into a case study about TheFind, which has an objective to allow you to find every product from every store.

How does TheFind use mobile analytics products? They currently have GA, Flurry, and a proprietary product that, all together, get the mobile analytics they need. And about 30 percent of their traffic comes from mobile.

From their insights, TheFind stopped spending almost anything on paid media and shifted more towards demand creation through search engine optimization (SEO) and email marketing.

The folks at TheFind really know their business. They can judge the impact of a particular promotion on downloads and subsequent usage. That said, they do find cohort analysis particularly painful and challenging.

Kadanoff then went through examples of Funnel analysis from Apsalar, an engagement report from Localytics and also Cohort analyses in Mixpanel and Apsalar. Here's what she highlighted:

There's a big link between engagement and revenue that you need to know about ‐ if apps can't generate engagement (i.e., if the app isn't being used, you're not going to be able to monetize). This sounds obvious but actually this is really important to note, because in the app itself it's very difficult, if not impossible, to measure all the way through to conversion.

Conversion oftentimes happens if a user currently has a free product and they're upgrading to a paid product – that conversion happens inside the Apple Store or inside Google Play. And while you'll get offline reporting you're not going to be able to measure conversion in the way that you're used with optimizing landing pages, etc.

Here's what you need to consider when evaluating products for your business:

  • Dashboard view: make sure the dashboard of the product can be customized the way you want.
  • Support for specific analyses you need (or anticipate you will need).
  • Ability to go beyond vanity metrics with more emphasis on engagement and ultimately revenue.
  • Referral codes for campaign tracking.
  • Integration – how easy or hard it is, particularly with the other data sources that matter to you.
  • Pricing model – free generally means free, but enterprise products are priced differently, on purpose.
  • Worry more than a little about the cross device problem – look for products that help you.

Some Final Thoughts on Mobile Analytics

  • Almost all tracking is done with anonymous device fingerprint tracking, which is about 95 percent accurate – more so for Android, less so for Apple iOS.
  • Apple no longer allows tracking by Device ID (UDID) and is expected to disallow tracking by Mac ID.
  • The leader in cross device tracking & analytics for mobile is a company called Drawbrid.ge ‐ worth checking out.