We work with top app developers, marketers, and publishers every day. Here are some critical principles we’ve learned in the process.
I. Data strategy before tools – While having the right data infrastructure and tools are important, even the best technologies are at the mercy of what data you choose to collect and how that data is tagged. Never forget this. Which leads to the second point…
II. Measure at least twice, cut once – When third party services are tacked-on hastily it creates substantial risks to your business. In mobile it’s not just “measure twice, cut once” — it’s measure a thousand times. That means planning ahead of time around questions like: What analytics and user engagement scenarios do you need to support in order to achieve your business objectives? How will you segment your users? What are the granular data requirements for all of this? What are the data attribute naming conventions? By thoroughly planning data collection you will make engineering resources more productive and avoid the need for costly and time-consuming rework later on.
III. Know thy audience – You can’t effectively understand and engage with customers unless you can recognize who they are. One of mobile data’s biggest assets is the persistent, logged-in user ID. When a user starts the app, you must be able to set that user’s identity and have all subsequent data collected in that user’s session be linked to the user. identification characteristics forms the basis of the relationship you build with your users and the foundation for effective analytics, segmentation, and personalization.
IV. Location, location, location – The three most important words in real estate are also extremely important in mobile marketing. Applying a location layer to your engagements can make them much more relevant and effective for users. By leveraging services like Radar and Foursquare, mParticle customers can integrate location-based event data alongside their other analytics, as well as associated attributes, including venue name, id, chain, category, confidence with these.
V. Push is not a strategy – Push notifications allow you to reach mobile users at critical moments, so it’s important to collect push tokens and be able to trigger notifications in real-time based on certain contexts (such as location, see above). Conversely, you should also endeavor to understand who has opted out of push notifications and ensure that you are engaging with them via other communication channels. Don’t think about having a push strategy, but rather instead on having an engagement strategy within which push is one among many tactics.
VI. Do no harm – Never ever let data collection kill user experience – UX must be preserved at all costs. Data collection should be implemented in a lightweight manner and fully removed from the user experience. Failure to do so results in performance sapping, app crashes, loss of data, and reduced battery life of consumer devices. Carefully consider which SDKs you invite into your app and whether you really need to embed that SDK in order to use the service (hint: probably not).
VII. Use an event attribute model in app tagging – Events are actions taken by the user (searches, purchases, logging in, or setting preferences as an example). Event attributes are name-value pairs that provide contextual details around the event. When the developer instruments an event, they call a log event method and include required and optional values. Example: Event Name = FlightSearch, Event Attributes=[Origin:JFK, Destination:SEA, DepartureDate:12/21/2014]. This approach allows you to use event names that align to your business and are easily shareable, so you don’t have to re-write code later on.
VIII. Don’t let convenient get in the way of good – You might think that point-to-point integrations are faster and cheaper than investing in a universal data layer, but you get what you pay for. With multiple point-to-point integrations, there is significant data overlap. Worst of all, your data doesn’t pass through your own infrastructure, so you’re giving up ownership. Fragmented, redundant, and disempowered is no way for data to go through life. Control your data, control your destiny.
IX. Take a user-centered approach to customer data – Far too often, companies let their organizational structures get in the way of a complete view of the customer journey. For example, crash data goes to the engineering team, advertising data goes to the acquisition team, and CRM data goes to the retention team. Yet, it’s all relevant to the same user or group of users. Break down these silos by deploying a universal data collection library. That way, you can segment and message users who have had multiple crashes in a specific app version just as easily as you can send a Facebook advertisement to someone that searched for a flight from New York to Seattle.
Have we missed something? Drop a note and tell us about your organization’s own commandments for Mobile Data Strategy.