Before you dive into the live blogs of Apple’s event this morning to see if the rumored new versions of the iPhone are true, it’s worth understanding the state of the Apple & iOS world today. This Data Snapshot looks at activity on Apple devices worldwide on Monday, September 9, 2013.
Today’s Apple announcement will likely add new devices to their portfolio, a huge potential opportunity to reach more customers globally. Today, all their mobile & tablet activity is spread across just three — iPad at 60% of total activity, iPhone at total activity 33%, and iPod Touch at just 7% of activity.
While we are still more than a week away from the planned release of iOS 7, it’s been in developers hands for several months now and we are starting to see measurable activity. Yesterday (Sept. 9) 1.5% of all iOS activity is on iOS 7.0. It’s worth noting that iOS 7.0.1 has popped up, the level of activity is too low to be measurable–but it’s there.
To get more detail on the iOS version share yesterday visit Mixpanel Trends.
iOS vs Android
While there may be more Android devices in market, iOS devices continue to generate more activity. As of yesterday, 58% of all activity on mobile & tablet devices was generated by iOS devices compared to 41% from Android devices. And these percentages have remained fairly constant over the past six months.
To dive into more detail visit the Android vs iOS report on Mixpanel Trends.
Opportunity for Developers
Along with new devices, Apple will undoubtedly add new features and functionality. This always creates new opportunities for application developers. As of yesterday the five verticals that drive the most activity on iPhones are (in order) Media, social, photo/video, finance, and ecommerce. It’s slightly different on iPads with the top five verticals as of yesterday being Photo/video, media, travel, finance, ecommerce.
If you are interested in more Apple data, it’s also worth taking a quick read of John Koetsier’s story in VentureBeat yesterday on the worldwide distribution of smartphones. Now onto the live blogs!
Our friends over at Appstem have invited Mixpanel’s own Aliisa Rosenthal to join their Webinar on App Analytics and Diagnostics on September, 26, 2013 at 11am PT. If you are just getting started with mobile apps and don’t have experience with analytics, consider listening in. You can get more information about the webinar and register here.
This post is short and sweet, just like the event and property names you’ll be sending to Mixpanel after reading it!
If you’ve joined a Mixpanel webinar or read our best practices blog post from last year, you know you should use clear, human-readable event and property names. Why? Because it makes it easier for everyone at your company to understand what the events and properties actually mean.
But there’s another reason you should avoid long, hard to understand event and property names: Mixpanel’s UI will reward you for it. We go to great lengths to make Mixpanel’s reporting UI as elegant and powerful as possible. The tradeoff for having such a clean interface is that we have to adhere to certain character limits when displaying event and property names in the UI to avoid truncation. So while the technical limit on event and property names is 255 characters, in the UI the event and property names are cut off after as few as 22 characters (see below).
It’s easy to avoid this issue by keeping a simple set of best practices top of mind while you are naming your events and properties in Mixpanel:
Keep it short. Whenever possible, stick to ~20 characters when naming properties. There is considerably more leeway when it comes to event naming (though keep in mind that event names will be truncated at around 20 characters when they appear within chart legends). Note that we are referring here to property names and not property values which, like event names, can be considerably longer without being cut off.
Pick a name and stick with it. Changing property and event names after you’ve started tracking data in Mixpanel requires considerable work and is not recommended. Even if existing event and property names are being cut short, you should stick with what you have and revise the naming protocol in your next Mixpanel project. If you absolutely must adjust event and property names after the fact, shoot us a note at firstname.lastname@example.org for guidance.
Put distinctive information near the front of the name. While we always recommend that you keep event names general and store details in your properties, make sure to put distinctive information near the front of the name regardless. For example, if you’re not into the whole brevity thing and were tracking an event called “Big Lebowski Quote Selected”, you’d want to name your properties “Character Name” and “Quote Length” instead of “The Big Lebowski Quote Character” and “The Big Lebowski Quote Length” which would look exactly the same in o
ur interface if truncated at 20 characters.
In People properties, always use the spacebar. Take extra care when naming people properties. Within our “Explore” report property names will be truncated after as few as 12 characters unless you incorporate spaces into your property name (which will expand the row height and wrap the text). So instead of ‘number_of_quotes_read,’ go with ‘Number of Quotes Read’ (see example below).
Choosing the right event and property names in the beginning will help you get the most out of Mixpanel and keep your reporting as clean as a whistle.
If your app is using an old version of the Mixpanel library your reporting could become inaccurate at the release of iOS 7. To make sure this does not happen, please update to the latest version of our iOS library before Apple releases iOS 7 to the general public in September.
Starting iOS 7, Apple has removed the ability for apps to access the MAC addresses of the device. Specifically, every device will begin reporting a generic MAC address of “02:00:00:00:00:00.” Because Mixpanel iOS libraries 1.1.1 and earlier relied on the MAC address to generate the default unique identifiers (distinct_id) for the device, every new customer who installs your app on a device running iOS 7 will be assigned the same distinct_id.
The latest version of our iOS library, which has been available since April, does not rely on the MAC address to generate a unique identifier. You can access the latest version of the library at GitHub.
How was distinct_id determined in iOS library versions prior to 2.0.0?
In older versions of our iOS library we concatenated every MAC address available to the phone (including Wifi and Cellular interfaces) along with the app bundle name. We took this resulting string and hashed it. The result was the distinct_id.
If an existing user of my App updates to iOS 7, but has an old version of my app that uses an old Mixpanel library, will they be tracked correctly?
Yes. Because the distinct_id that we generate is saved in persistent storage on the phone, your existing customers identifier will not change and remain correct.
If I do not update my app with the new version of the library, and a new user installs the app on iOS 7, what will happen to my Mixpanel reports?
In this case, all of your new users will be assigned the exact same distinct_id. Every new user will be considered the same user in Mixpanel reports. This will break any reports that rely on unique user calculations including Funnels, Retention and all of People Analytics.
Is 2.0.0 a drop in replacement for 1.x versions of the lib?
For the most part, yes. There is one exception. In the latest version of our library, we eliminated the distinction between [mixpanel identify:@"user123"] and [mixpanel.people identify:@"user123"]. Now there is only [mixpanel identify:@"user123"] which must be called to enable People Analytics. If you’d like to enable People Analytics using the default identifier, you should use the one liner: [mixpanel identify:mixpanel.distinctId]
What if I always assigned my own identifier using the identify() method?
If you always assign your own identifier to every user before they fire any events, and your own identifier does not rely on the MAC address, you will be unaffected by these changes in iOS 7.
How can I check the library version we are using in our App?
The version constant is defined in line 36 of the Mixpanel.m file. Check the copy in your X Code project.
Retention is the underdog of Mixpanel reports, often taking backseat to more popular reports such as Segmentation and Funnels. These are are intuitive and amazing, and they’re great places to start. But don’t overlook Retention reports. They are also incredibly powerful, and a fantastic way to measure the frequency with which people return to your app or site to take valuable, meaningful actions.
To give you an idea of how to drive the most value from retention, I’m going to use the example of a fictional dating site. Because I am a sucker for both puns and biannual cleanings, we’ll call our site OkCuspid, which is a dating service that facilitates romantic connections between dental professionals. OkCuspid’s revenue model is much like that of LinkedIn: people are charged a small fee per message sent to a potential love interest. So while their philosophical goal is to find love for people who love teeth, OkCuspid’s business goal is to get users to send multiple messages.
A retention report that shows cohorts that “Create Profile” and then return to “Send a Message” is valuable to every team at OkCuspid. They can filter the retention report by any property sent with the event(s) in question. That lets OkCuspid’s team focus in on retention rates for different genders, age groups, types of accounts, or geographical regions. This lets them answer questions ranging from “Do men send more messages than women?” to “Are users between the ages of 25 and 32 who live in Dubai more engaged than users of the same age in San Francisco, and if so, which ad campaign acquired them in the first place?”.
These insights allow their team to act swiftly to develop the product and reach out to their audience with clarity, specificity, and assurance.
The OKCuspid product team can use this retention report to correlate changes in retention rates with different features they’ve added, or changes they’ve made to the user experience. Perhaps rising retention rates correspond to the addition of emoji in messages; alternatively, a dip in retention might follow the rollout of a new mailbox interface.
Their business development team can use this retention report to see that people who create an account come back to send messages in the first week or two, but their activity trails off in subsequent months. They can use this data to test different pricing schemes (perhaps a free first week of messaging, or cheaper fees per message sent) and measure if the retention rates rose in accordance with these changes.
A retention report is especially powerful for the OKCuspid marketing team. While they do get customers from organic, word-of-mouth referrals, they also do online ad campaigns. Let’s say these campaigns are on Facebook, Twitter, the New York Times website, and WebMD.com, and they all employ utm-embedded links to OkCuspid’s homepage. (Whenever a user comes to your site from a utm-embedded URL, Mixpanel will automatically grab the utm-params and set them as super properties for that visitor, which means they’ll be tied to every event fired by that individual. This is great for seeing later-stage events and tracing them back to their origin.)
The OKCuspid marketing team can filter the retention report by any utm-related property to see exactly how effective their various campaigns are in drawing loyal and engaged customers. This allows them to target their efforts and focus on channels that yield the most recurrent and consistent use of their product.
Retention reports are not only valuable for those in the dental dating industry. They give any company powerful and actionable insights into how their customers engage with their product over time. If you’re a Mixpanel customer and want help getting started, drop our solutions team a line at email@example.com or check out our video tutorial.
When does it make sense to build vs buy your analytics solution? What’s the best fit for your business? Data driven companies should think through the pros and cons of each approach.
Many companies have the expertise and experience to build analytics in-house and for eng teams it’s an interesting challenge. But with that challenge comes costs–both in building infrastructure costs and engineering time. Once it’s built the system also needs ongoing eng support to make sure it scales as the company grows–not an easy task– as well as to make sure data is being recorded correctly as your team builds new products and features.
In-house analytics tools require an in-house expert or building out a BI/Data team. These people will become the gatekeepers that extract the data, use SQL to run queries, and export to Excel. That process usually runs something like this:
Figure out which data is stored in which database. Most companies use multiple different databases.
Write the appropriate query–usually SQL–to download the data you’re after. Note that this query will need to be constantly updated to ensure you are collecting the right data points.
Download the data to Excel and build a custom, comprehensive analysis.
Share the insights from data via email or presentations.
This means the data lives off-line, siloed, unshareable, and out of date.
These elements are all part of the cost of ownership that most organizations tend to underestimate and why building often ends up being more expensive and much more time consuming than using a third-party tool.
With a tool like Mixpanel, anyone can run reports and everyone is referencing the same real-time data. Creating your own data store means marketing probably can’t use it, and any new hires will have to spend time learning the system. Mixpanel is easy to pick up in ten minutes, even for your social media intern. More widespread access to accurate data = better business decisions = growth.
When you build your own analytics you are much more likely to have data inconsistencies and make errors in your reporting that take longer to cover. Mixpanel has thousands of customers constantly vetting our calculations while you’ll only have a couple.
At the end of the day, we have seen people invest in building analytics tools only to end up utilizing a service like Mixpanel. We’ve even had customers come to us wishing they had gone with Mixpanel upfront, rather than wasting time and resources building in-house.
Companies look at the cost of building analytics in-house as simply the price of one engineer. In reality, there is a far more significant ‘total cost of ownership’ associated when you factor in database maintenance, database upgrades, the variable time of a BI rep running queries and slicing data in Excel on an on-going basis. Not to mention increased risk of making decisions based on faulty and stale data, and data bottlenecks in an organization.
Trying to prioritize which prioritize which product features to invest in? Read this post from Lee Munroe on how he developed a Minimal Viable Product (MVP) and used Mixpanel to measure how valuable his customers would find it. Data driven product development.
The Data Driven Conference kicks off today at 2pm. We are excited to welcome our amazing lineup of speakers to Mixpanel HQ. They will be sharing the ways that data is shaping everything–from investments to product development to customer engagement–with our sold-out audience. If you don’t have a ticket, don’t worry–you can still learn from the best.
Your data is a valuable resource that can tell you how people are using your product so you can grow your business. But to get the most value out of your data, you need to change how you think about analytics, actionable metrics and properties. That’s where Mixpanel comes in.
Mixpanel is easy to integrate. But if you need help figuring out the ideal implementation or don’t have the technical resources, you may want an expert to help you out. Finding the right expert who really gets it can be hard, so we went out and found them for you.
We’re pleased to announce our inaugural set of Consultants That Get It: a group of world-class analytics consulting companies who have been trained by our Solutions Architects in the Mixpanel Way. These companies have a thorough understanding of the best ways to plan and integrate Mixpanel into your web site or mobile app.
These consultants have worked with giant Fortune 500s, innovative startups, and passionate non-profits to make sure they are tracking the right data points to help them grow their businesses. We’re confident that if you are looking for an outside consultant to help your company move forward with analytics, one of these consultants that get it will get you what you need.