Clever business

If we consider our manufacturing ERP software as the system of record, then we should think of business intelligence (BI) as "the system of engagement". So says Eric Duffaut, SAP's president of global ecosystem and channels. And while his terminology may sound grandiose, he's not wrong. Why? Because, contrary to popular opinion, BI is not just a reporting suite; it is, or should be, about intuitive data analytics. BI's value is in providing useful business insights to everyone that needs them on demand. So, in that sense, this software is the driver for better informed decision making and, in turn, 'engagement' with people and processes, both internal and external. That's essentially the proposition from the few remaining independent BI software vendors, and those long since swallowed up by the ERP vendors and, of course, IBM. They say that, by adding BI as a module integrated with your existing ERP system, you gain a flexible foundation capable of improving knowledge around almost anything, from production planning to product quality, supplier operations, market appeal, customer relations… You name it. Sceptical? You're in good company, and there are several reasons for that. First, many operations directors' concept of BI seems resolutely stuck in the 1990s, when the name of the game was indeed generating reports from ERP data. Second, even those that might be persuaded worry about ERP implementation creep, so tend to list BI as a phase two project – which then never happens. So, while there are examples of excellence out there, BI, in this modern sense, could hardly be billed as well proven, least of all among manufacturing SMEs and certainly not far beyond the financials. Malcolm Fox, director of product marketing at mid-market ERP developer Epicor, agrees that BI is still largely misunderstood, describing the software as "an unfulfilled promise". Getting past that, he says, requires would-be manufacturing users to get under the superlatives and see that today's BI provides a mechanism for "presenting users with abstracted information that they can navigate wherever they choose". What does he mean? "They might start with standard views of aspects of their operation, but, if they notice something interesting or unexpected in the data, they can immediately drill down and pursue lines of enquiry." When it comes to slicing and dicing data, that's far more than most mere reporting systems allow. And importantly, users no longer need to specify yet another new report every time they find an anomaly – which never works because, by the time the overstretched IT department delivers what you really wanted, you've forgotten your question, or the aberration has become irrelevant. "BI today is almost anarchic," explains Fox. "You can look at information any way you choose, using the navigation to drag and drop, and link views. You're effectively going on a journey of discovery." That might sound fanciful, except that, for him, this is entirely about uncovering otherwise hidden manufacturing and business issues in time to do something about them. He cites, for example, intermittent quality problems, variable machine set-up times and production bottlenecks – not just the usual sales and margins by region, by product group, etc. Steve Dunnigan, business analytics executive with IBM – which counts Cognos, SPSS and Clarity among its best BI developer acquisitions – agrees. Quickly understanding causal links, he asserts, enables users to break down hitherto intractable barriers to improvement – for example, between production cells, lines, shifts, departments, supply chains, even whole value chains. And he makes the point that extending the data sources beyond ERP and into your industry or even the wider world can also deliver surprisingly valuable results. So what about examples? "Some of our manufacturers are analysing information sourced via the web. They're using our software to determine what they should be doing in terms of, say, product development, or changing their marketing messages based on what BI shows them is being said on Facebook or Twitter streams," explains Dunnigan. "And they're doing that in near-real time." Others, he says, are analysing data coming off their production lines to determine patterns, with a view to optimising plant use. And others again are using BI as a tool to investigate repair and warranty statistics across their dealer networks to guide product design, quality and/or customer service improvements. He cites BMW, which started using IBM's SPSS data and text mining tools to structure and analyse combinations of so-called 'big data' on, in particular, vehicles and repairs, vehicle error memories and dealer feedback. Its initial project was about reducing error rates and cutting costs, but doing so by examining patterns and pointers hidden in all this information, to help prioritise continuous improvement, in terms of the auto maker's production methods, as well as the products and services themselves. As a BMW spokesperson explains, one important quality indicator is the number of faults recorded within any vehicle type's warranty period. Reducing this number, by using correlations and trends discovered in the data to drive root cause analyses, lowers the cost of rectifying vehicle problems and improves the products – resulting in better customer satisfaction stats. That's the role of SPSS, which, he says, "produces fast and efficient analyses… quickly handling several thousand queries… and combining the results". Now, around 1,000 BMW employees can access the analytics under the banner AVAQS (advanced quality system), which effectively embeds the SPSS tools within their familiar applications. As a result, users can see and act on the insights, whether or not they have direct access to SPSS or any knowledge of how to use that BI application. BMW gives the example of computer-assisted repeat repair analysis, which uses information on the kinds of repairs that cause customers to visit workshops most frequently, to drive engineering and production improvements. Similarly, the car giant cites improvement work at its Landshut, Germany, foundry, where it is using AVAQS to correlate process settings with casting quality, again to drive production improvements. All well and good, but, as with most such organisations, neither IBM nor BMW is keen to reveal too much in the way of project cost information. And, although Big Blue's Dunnigan states that its BI software and services are just as relevant to £20 million turnover companies as they are to £20 billion behemoths, we can be fairly confident it wasn't small change – even if the ROI is clear and evidently growing. So is cost a barrier? Much depends on how you buy your BI system and who you purchase it from. If BI is rolled into the cost of a new ERP system purchase, then the delta for the software licences is likely to be small – certainly compared with the likely benefits. That said, consultancy fees and training costs have traditionally been on the high side – meaning you still need to be sure of your direction and its potential. Meanwhile, if you're looking at a BI suite from one of the independents, you'll find pricing varies by a factor of nearly 10, depending largely on how you couch your requirements, the number of users and the industry preferences of the suppliers concerned. All that may be about to change, notably with Infor's release last month of its next generation Infor10 ION enterprise suite, with so-called context apps, which include ad-hoc, small-scale, user-configurable analytics readily accessible from the consumer style user interface. Thinking about the BMW project, we're not comparing apples with apples: this is more about enabling event-driven, collaborative working, based on instantaneous queries into Infor's Business Vault data warehouse. But it, too, could be a huge, and very pragmatic, leap forward for many manufacturers. Infor's president of product support Duncan Angove gives the example of a user in sales noting that his 'capable to promise' timeframes are slipping out, but then instantly being able to find the reason, via the relevant on-screen context app, as planned maintenance. If said salesperson urgently needs more capacity, one right click enables him to share the difficulty with everyone that needs to know, and sort it out. Think tweets and hash tags – you can follow customers, production runs, whatever – as well as 'presence', and you're about there. Sense and worth None of which provides direct answers to the cost conundrum, other than signalling that the horses for course aphorism is apposite. So, best to take a look at a couple of quite different BI projects at equally different manufactures, and then draw your own conclusions about what might make sense – and what it's worth. Take ultraviolet lamps manufacturer Hanovia, which provides equipment for disinfection and chemical reduction in industrial process water. The company's Terry McCarthy explains that it needed analytical functionality to deliver static reports on the one hand, but also ad-hoc analysis on the other – mostly to assess sales and stock information. Since Hanovia uses Pinnula's Uniplan ERP system, McCarthy turned to the company for a recommendation, and was pointed to Dimensional Insight's DI Diver BI tool, which is billed as powerful but low cost. McCarthy says that Diver extracts information seamlessly from Uniplan, creating data cubes for periodic and real-time analysis. "The improvement has made us far more effective when analysing sales trends and affectivity in market sectors," he states. "We used to report in fixed 2D formats, which were rigid. We now have user-defined fields and can quickly and easily use Driver to provide all the information we require." At an entirely different level, $70 million North American construction equipment manufacturer Blastrac effectively integrated business insights from six separate ERP systems' databases, improved reporting productivity by 95% – and also saved a fortune, by going with Tableau BI software. According to the then Blastrac CIO Dan Murray (now COO of Tableau reseller and system integrator Interworks), this project arose out of a need to improve visualisation of financials, purchasing and associated data worldwide. The apparent choice, he says, was a global roll-out of new ERP software (instantly rejected for its multi-million dollar price tag) or enterprise-wide BI, extracting data from the ERP systems to a new, central data warehouse – similarly dismissed, following quotes from $130,000 to $500,000. Murray takes up the story: "A colleague advised me to call Tableau, so I went to the website and was able to download trial software and install it in a couple of minutes. Within an hour I was fairly confident it would do everything we need – and without any consultants costs, too." A week later, he says five reports had been developed, which proved the system's ability to support interactive dashboards, taking data from the existing multiple ERP systems. Then, three weeks on, the dashboards had been refined and data extraction processes streamlined – and all with one licence of Tableau desktop and the free reader software. As an epilogue, Murray's development led to a four-month data cleansing project, which reduced intransigent errors in order entry that, in turn, prevented problems on the firm's shopfloor and improved margins, OTIF and customer satisfaction states. "I did the project for about $40,000 instead of at least $130,000. It was a fraction of the cost, and since the Blastrac implementation, the software has matured and the tools are even better. And another thing: if you have multiple data sources, most BI suites require you to build a metadata layer and ETL [extract, transform, load] from that. But Tableau connects without metadata and, if there are common records it gets them dynamically. So you can blend multiple data types and sources in a single dashboard." Those concerned that global business analytics will cost you the earth can clearly relax on that key point. There's a world of new opportunity out there, and getting at it simply requires some new thinking and then shopping around.