Generic filters
Exact matches only

I just finished reading a great post by Hannah Smalltree at called Mobile Business Intelligence – Will it take off (for real, this time)?  Again, a good read, but it highlights many of the issues that still cripple mobile BI (for now).  Even the success story she highlights is riddled with flaws which have to be overcome before Mobile BI can really hit the mainstream.

Here are a few examples:

  1. Why do BI vendors keep assuming that software needs to be installed on the device?  Somehow most companies in the industry have ignored how powerful the browsers are on these new devices, and I’m not just talking about the iPhone.  Blackberry, Windows Mobile and Windows Phone, Android, Palm, etc. all have really powerful browsers that can accomplish nearly everything that an app can without putting software on the device (and this gets even better with HTML 5 on the horizon).  Do we have to remind software companies how hard it is (for the vendor and the customer) to create a new version of your app for every device that gets released?  There are hundreds.  It is impossible, frankly.  Oh, and if you solve for the phone’s browser you also get the desktop solution for free (and new devices like the iPad).  Some might say security is an issue, but it isn’t and here’s why.
  2. Why does everyone assume there needs to be a big project to accomplish all of this?  I expect it is because most of these vendors make the majority of their money on services (the custom project that was referenced was $30,000, which was quite small but still involved low-value services work which was unnecessary).  This is unfortunate because when done right from the beginning, most key metrics can be viewed, in context and from all of the right angles, in just a few key screens (which should be generated by default), especially when those KPIs are operational in nature.  Key “views” always include:
    • What is the state of my x (business, factory, profit, product, supply chain, etc.) at a glance? (in our case, a % rollup view of all KPIs in a group)
    • What is the current value of a key individual metric? (found on almost every screen)
    • How does it compare with its expected value (thresholds, limits, etc)? (trend, KPI map, bar chart)
    • What does this value look like compared to other similar KPIs? (scorecard)
    • What led up to this point? (the history, trend)  This leads you to… (trend)
    • Are things likely to get better or worse given the trend? (trend)
    • Am I alerted when something is particularly bad or good? (monitoring, alerts via email, SMS; see an example on the right)
    • Optional: Who is responsible for this KPI and who do I alert?  (alerts)
  3. Mobile BI should focus on operations data, not future planning data.  Traditional BI (OLAP cubes, data mining, hard-core analytics) often looks at planning data and also focuses on huge problems that need deep analysis (e.g. how many people might buy an iPad if we build it).  This is hardly the right application for the mobile platform.  Think about what people care about on their phones, regardless of what is possible:  email, Twitter, getting directions, text messaging, sports scores, breaking news – all of these are operational, or “here and now” metrics.  If you are going to analyze hundreds of variables around future planning scenarios, you are going to want to be in front of a big monitor with plenty of time and a pile of empty Mountain Dew cans – this does not make an ideal fit for the mobile scenario.
  4. Mobile BI need not be rocket science.  Most vendors we run into think mobile BI applications need to be wildly customizable.  We respectfully disagree.  My second bullet above highlights this.  Transpara’s Visual KPI application has been running with customers’ data in industries as diverse as utilities and power generation, oil and gas refining, data centers and biotech and we talk with those customers all the time.  The funny thing is, we almost never get comments like “if only I could make a new screen that had x in it” which for a long time even worried us.  Are they even using it, we thought?  Turns out they were, and the screens we created showed them all of the key metrics in the right way.  Now, this doesn’t mean we don’t have a significant list of improvements to make (we do) but it does highlight how straightforward mobile BI software must be in order to deliver value.
  5. Mobile BI software is not just a front-end client to traditional BI.  Again, read bullet 3 above about operational data.  Traditional BI is a great source of data for a mobile business intelligence solution, but they are most often not a 1:1 match and certainly not exclusive to each other.  The ideal mobile solution often includes marrying multiple data sources to create that real-time solution.  For example, get some thresholds or limits from deep analysis in your data warehouse but combine that with some real-time data from a streaming database, historian, or even a traditional database that is refreshed faster for a true “here and now” mobile application.  With the right solution, this will be easier than you think.

I could go on for weeks on this topic, but customers await.  If you have comments we would love to hear them, especially if you disagree.  Oh, and you can read a potentially more intriguing operational intelligence case study here (See the intro section on Mohawk Paper and Information Everywhere on a Blackberry on page 3 – I will write up more about this one in another post soon).  Thanks as always for reading.

Michael Saucier