Cloud BI, Big Data, Mobile, Predictive – the IT Industry LOVES new trends. Just like with cars, clothes and consumer electronics, the titans of Silicon Valley are great at turning their latest back-room project into a mega-trend that we can’t live without. Us business intelligence companies are no different.
We all know that some of these trends will fall by the way side whilst others will become mainstream. Not all of them will be right for you or your company. The latest predictive analytics might be great for a large utility company or bank, but useless for a mid-size manufacturer or retail business. Cloud BI is anohter hot topic but is it right for your business?
So how do you know which to back and which to pass? Find out which are the 2013 trends from business intelligence companies to embrace and which to ignore:
What is it? Big Data is probably the biggest hyped trend from business intelligence companies. What it actually means is something quite simple: the ability to deal with and analyse the sort of massive data sets that 10 years ago we didn’t need to worry about. Back then we knew it was impossible to analyse petabyte datasets quickly unless we worked for the NSA and owned a couple of Kray supercomputers. Now, post the arrival of cheaper ways of handling big data, it becomes tempting to want to.
What can I use it for? As a mid-size business the applications of Big Data technology over traditional, structured data sets are limited. You’re just not going to create enough data to warrant needing Big Data horsepower under the hood. For example, if you’re doing around 1m sales transactions a year, each with 10 line items on it and with a value of about £100 then you’ve got a £100 million turnover business (nice!) but are still only generating 10 million rows of data. This sort of data set is barely going to make a free copy of MySQL sweat.
With big data you’re talking about handling billions of rows of data and petabytes of storage. You just don’t need this firepower for your traditional data. That won’t of course stop business intelligence companies selling you a product with a big data badge on.
There are Big Data workloads that could be useful to mid-size businesses. Web statistics analysis is one example. Another would be the analysis of unstructured data (e.g. social media posts) to find out how your brand is being talked about in cyberspace.
When to embrace? You’re heavily involved in research, modelling, or have large sensor-feedback datasets. You’re a massive company. You’re analysing brand conversations in social media or complex web statistics.
When to ignore? If you’re anything but the largest of companies, you’re not going to need “big data” to analyse your traditional, structured business data (e.g. sales, inventory, finance).
What is it? Being able to access reports and dashboards on your mobile device, smartphone or tablet.
What can I use it for? The applications are pretty limitless. You can issue it to your sales teams to allow them to be best prepared in client meetings. You can update executives whilst they travel on business performance. You can improve your brand perception by talking through figures on a slinky iPad, rather than coffee stained piece of paper. You can alert stakeholders as to changes in KPIs or thresholds so they can take action.
What you shouldn’t do is try to deliver the same reports you consume on your PC or as paper, onto your mobile device. You’ll just end up with a very small, very fiddly report. The trick with Mobile BI is to design reports suitable both the device and to the situation in which the consumer will be using them.
When to embrace? Every business can use some aspect of mobile BI, if they’re not doing already. 2 years ago, Mobile BI was a “nice to have”. Now, every customer we get though the door wants it, needs it and uses it.
When to ignore? Don’t.
What is it? It’s software that combines statistics features with data mining and some clever machine learning stuff to that analyze historical data sets and make predictions about future events. Phew, that was a mouthful.
What can I use it for? A classic example of using predictive analytics is in a mobile phone company.
All mobile phone companies face a challenge which is that we consumers aren’t very loyal. We swap and chop pretty regularly, which is usually referred to as “churn” in the industry. If they could predict we were likely to churn, then maybe they could send us a promo in the post for a cheap upgrade or new handset, we’d then stay on as a customer and bingo, they’ve saved a lot of money.
Predictive analytics can do this. It can, by looking at all the various data held in all their various billing, help-desk and CRM systems and comparing against historical customers that churned, assign each customer a “churn probability”. Mr Mobile Phone can then send a free cuddly toy to all those whose score is high, preventing them from churning.
Other examples include applying predictive techniques to border crossings, analyzing cars for how likely they are to be carrying drugs or contraband. CCTV and number plate recognition data is combined with data captured by border crossing officers and a predictive analytics based decision support system tells them which cars to search and which to waive through.
Cool stuff huh?
When to embrace? In our opinion, there are very few practical applications of predictive analytics in mid-sized or B2B orientated companies.
When to ignore? If you have less than a squillion customers.
What is it? In-memory analysis is the modern way of delivering analytics and “drill-down” capability in business intelligence solutions. The name refers to the fact that some or all of the magic happens in computer memory (RAM) rather than stored on disk as a file or cube.
What can I use it for? In-memory analysis delivers the same sort of drill-down, slice-and-dice, group, pivot and filter functionality that you would perhaps expect from any well implemented business intelligence solution. The advantages of doing so “in memory” however are as follows:
Firstly, in-memory analysis is usually super-fast i.e. performant. You should notice a perceptible different in response times from an in-memory solution over a traditional, for instance OLAP-based, solution.
Secondly, the fact that everything happens in-memory means there are no cubes to build, avoiding tedious and inconvenient overnight batch jobs. This can have a positive impact on your infrastructure requirements too. To build cubes, companies need big servers which sit idle all day, until they’re needed at night. In-memory avoids this.
Finally and most importantly but often overlooked, in-memory analysis gives the user a lot more flexibility and in doing so, reduces workload on IT. The reason for this is that with traditional, OLAP-based, business intelligence solutions, the BI implementer had to predict in advance how the user would like to drill-down through their data. They would then build this “hierarchy” into the cube. If the user wanted to drill-down a different way, they would have to request the cube or hierarchy be changed or a new one built for them. With in-memory analysis, the user can decide how they want to drill-down on-the-fly – allowing them to get the information they need more quickly and easily whilst reducing workload on IT.
It’s not all roses though. OLAP (the old-school alternative to in-memory analysis) can still pull some tricks that in-memory can’t. OLAP is much better at, for instance, temporal (time-based) comparisons e.g. Month vs Prior Month / Month in Prior Year. It’s also better at calculations and at joining information together from different places e.g. stock v.s. sales.
When to embrace? In-memory analysis is faster, more flexible and requires less administration than traditional alternatives. It should be part of your modern business intelligence solution.
When to ignore? OLAP can do some really important stuff that in-memory solution can’t. Consider using both or evaluate your requirements carefully.
What is it? The whole software industry is going Cloud crazy and it’s no different with business intelligence companies. Cloud BI is where some or all of your business intelligence solution is delivered on a Software-as-a-Service basis, over the internet – rather than you purchasing and hosting it yourself.
What can I use it for? There is a reason that everyone’s talking about (and doing) Cloud BI. It makes sense.
Ronald Coase said back in 1937 that organisations grow when they do activities more cheaply internally than they could externally – and more importantly, that the opposite was also true. He won a Nobel prize for that, so he must be right! What he’s saying then is it makes sense to have someone else run, for instance, your BI system, if they can do it as well or better than you could yourself and more cheaply. It’s the same reason we don’t run our own power stations or mill our own paper.
The Internet has made Cloud BI possible. And it works.
Aberdeen Group for instance reported last week that business managers are 51% more likely to have access to the analysis tools they need, 54% more likely to find those tools easy to use and a whopping 72% more likely to receive timely management information if you use a SaaS / Cloud BI solution, compared to a traditional one. They also reported in a prior report that Cloud BI solutions are 40% more cost effective than traditional alternatives.
There are, of course, Cloud BI solutions and Cloud BI solutions. Some are just reporting tools in the Cloud. Others, like Matillion BI for instance, are complete BI solutions including ETL, Data Warehouse, Self-Service Reporting, In-Memory Analysis and Mobile BI.
When to embrace? Faster to implement, cheaper and easier to use. It’s not bleeding edge anymore. Cloud has become mainstream and you should be considering a cloud business intelligence solution for your analysis and reporting requirements.
When to ignore? If you like servers, flashing lights and big, capital and time intensive projects.
For more information about how to select the right business intelligence company, software and run a successful project, download our free e-book, ‘Complete Guide – Evaluating and Implementing Business Intelligence’.