Generic filters
Exact matches only

Two of the hottest enterprise markets are about to collide. What does that look like and more importantly, what should it look like?

In our last post we looked at how historians and Big Data are aligned (quite well) and we have discussed all over this blog how to best mix historians and mobility. Now let’s look at Big Data as a larger concept and talk about what is possible, what is likely and what we feel is the right approach (for now).

Spoiler Alert! Mobile applications should only show results and analytics from Big Data that meet the characteristics of mobile user behavior in general.

This cannot be overstated and applies to all data, not just Big Data. Real humans use their phones and tablets in very specific ways and at very specific times, which is why you see common characteristics show up in almost all of the top mobile apps (business or consumer). Here are a few:

  • Small chunks – break data into small, actionable parts I can work with on a small screen or with limited time. Analyzing wildly complex pivots of historical data on the train or waiting for the DMV is not the key use case here.
  • Timely – show me what is happening right now or just happened in case I need to take action. Mobile is usually not the case to do deep analysis into the past or the future
  • Context – give me more than numbers and pretty charts. Tell me if they are good or bad without me doing all of the work (KPIs are good for this)
  • Location-aware – where something is happening may be just as important as its severity. Am I close to the problem and can I fix it?
  • Alerts – the application should tell me when I should look at it, just like SMS, email or Facebook badges do.

We’ve probably all seen videos and software demos from the big BI vendors (SAP comes to mind) that show the CEO clicking a few buttons on their iPad to crunch some huge big data numbers from multiple angles to solve a wild problem right in front of the board of directors, but let’s be realistic. The data guys will always be called upon to solve these issues and they will most likely do this on the desktop. Not because of technical limitations, but because it makes sense.

Mobile BI in general often reminds me of a quote from comedian Patton Oswald that goes, “We’re science: all about coulda, not about shoulda.” We consistently see what people can do on mobile devices, but after talking to hundreds (probably thousands) of customers we’ve seen a distinct pattern emerge for what people should, or want to do. Just like with email, Facebook, Twitter, Google Maps, banking apps and more, mobile users want to know what is happening right now and what they can do about it. They are constantly checking feeds and stock prices, getting directions to their next appointment, etc. This behavior must be applied to mobile BI apps to be truly useful.

We must learn from Facebook, Twitter, email and other mobile success stories to really see what subset of Big Data will be useful to mobile users.

Hopefully you are seeing the answer already. Some of the best Big Data tools like Splunk, Cloudera, Pivotal and many others are doing a great job helping you get answers from huge, diverse data sets. Like BI vendors, they also may have mobile versions of their desktop applications, but as we’ve discussed this isn’t where success will be found because the behavior doesn’t match the feature set.

The ultimate first task is to refine those answers down to the ones that make sense to show mobile users and let them drill in only when necessary. For the moment, this lends itself most to operations and other fast-moving data, where people need to be alerted on a pending crisis, or know when inventories are low, or see what they should work on next based on their location. To get this right for a large number of users, and preferably those without extensive training, you need to surface only those items that match mobile user behavior.

As always, we’ll have more on this topic soon. Let us know what you think!