Smart business

Business intelligence (BI) software may have been around a few years, but it's still causing confusion in manufacturing circles. Operations directors, managers and their ilk are unclear what value it brings, given that most already have systems perfectly capable of generating spreadsheets and reports. If the goal is business improvement, what intelligence do we want from our businesses – and how might a BI system help to turn that into action, where others have failed? All fair points. Which is why Works Management, with sister title Manufacturing Computer Solutions, staged the first of its 2009 IT Forum roundtable events last month at Gaydon's Heritage Motor Centre to examine BI: its value, where it works and how to get the most out of a system. Sponsored by Infor, SAP and Zebra Technologies, the event brought together invited manufacturing managers, who joined independent business, manufacturing and IT consultants and two presenters – Darren Dowding, senior business analyst for Cosworth Racing (now in phase two of its global SAP implementation) and Nigel Pendse, BI guru and author of 'The BI Survey'. First, what is a BI system? For most purposes, it's a system that provides historical, current and predictive views of business operations, with analysis, reporting and drill-downs, using data collected (usually from multiple internal and external systems) into a data warehouse, or directly in operational systems. What do manufacturers think? Our own online research among your peers paints a mixed picture. We find, for example, that about one third of manufacturers are currently using BI, with another 10% planning implementations, but half with no plans. Among users, significant numbers are not only using BI for financial analysis and business/capacity planning, but also to underpin S&OP (sales and operations planning), forecasting and even production planning and purchasing management. Indeed, well over half describe BI as very or fairly important across these central functions. Perhaps most surprising to non-users, 52% say BI is as mission critical as ERP, while no fewer than 72% believe that "BI is about closing the loop between information and action" – indicating that the phrase is not merely marketing speak. On the other hand, non-users' top reasons for shunning BI include what they see as perfectly adequate spreadsheets, followed by existing reporting software or ERP system reports – although budget is another issue. Interestingly, very few of this group see BI as complicated, but a far greater number of existing users believe companies aren't spreading adoption of BI because managers think employees would find BI too complex. That said, among users and non-users alike, one third still believe BI is more hype than reality. So is BI hype? Views were less polarised between manufacturers and IT vendors than might be expected. Cosworth's Dowding, for example, while agreeing that hype is an issue, insisted: "BI is about getting the right data to the right people, in line with what your business is trying to do." And Richard Neale of Business Objects (now part of SAP), agreed – but with the caveat that, for many purposes, that rules out spreadsheets. Why? "Because you don't know where the data came from, or how out-of-date it is." And Chris Field, consulting manager with Infor, while conceding that Excel has perfectly good pivot tables, added that as soon as you need additional dimensions, such as customer or currency, you're in the wrong tool. He also insisted that ERP reports, similarly, have their limits. "ERP was designed around the task of getting data in fast, and spreading it around organisations. But if you want to slice and dice information by product type, you need something else." Manufacturers are pragmatic souls and Mike James, production director of rail fastenings manufacturer Pandrol UK, spoke for many when he asked: "Why not go talk to people in the business and get the information that way?" He also expressed the widely held concern that releasing information without context to others could open a can of worms. Fair points. All agreed that BI is no substitute for face-to-face contact, hence the value of S&OP meetings. However, it's plain that, where companies are geographically distributed and/or where complexity and time get in the way of informed discussion, BI tools are invaluable for providing unambiguous insight. And here are some other pointers. Nigel Pendse: "BI is good for confronting new challenges and determining best directions for companies entering difficult markets." Cathie Metcalfe of Gradient Consulting: "Sometimes BI tools can give management fresh perspectives." Cosworth's Dowding: "Part of BI's value is communicating key information and helping to guide behaviour. It's also about shocking people with information." John Hammond of SAP: "BI can provide an early indicator of impending problems for senior management." And SAP's Neale: "BI replaces the manual processes of collecting data." Pace of performance Already, it's clear that BI has multiple roles, so it's time for Pendse's observations, based on eight years of BI surveys, the latest among 2,150 respondents. One of his most interesting statistics: "86% of those with BI would like to deploy it more widely – so they think it's worthwhile. However, what most complain about is slow query performance." Odd though it may seem, while Google returns results searched against most of the world's data in a fraction of a second, it transpires that many BI systems take more than 10 times as long to do their work – which turns off users and brings systems into disrepute. Pendse suggests looking at in-memory BI systems, such as QlikView and TM1, which, although small, are fast. Other useful guidance notes: manufacturers that choose BI systems based on multi-product competitive evaluations gain more benefits and record fewer failures. Pendse suggests talking to business users, not IT sponsors or IT vendors – and certainly not relying on project specifications "that request the bleeding obvious". What matters most, he says, is getting the right shortlist ("so it doesn't matter which one you choose") to serve middle management – "not which dashboard looks prettiest". Not that anyone decries dashboards (screens with summarised and personalised metric views, usually with drill-downs and querying). As SAP's Neale says: "A dashboard is a perfectly valid way of providing information to business users, but so are reports and spreadsheets that sit on top of your information supply chain. It's about getting views that are fit for purpose." Just so, but as Pendse cautions, don't patronise your managers – you'll limit their views of the KPIs they need. And two other gems: if you have to train people to use the system, you're likely to fail; and get your initial system in quickly. "Slash your requirements to the minimum, and get the thing live within three months. If you don't, the project will lose funding, or you'll end up delivering something the business no longer needs," warns Pendse. So how is Cosworth tackling its BI project? Dowding explained that his company took a phased approach – getting its SAP ERP system implemented across all four businesses and two continents as quickly as possible, before revisiting it in terms of data, metrics and analytics. Speaking of this second phase, he said: "Our objectives were to gain a better understanding of past performance and use that to predict the future. But we also wanted a system that would be as easy to understand for the guys on the shopfloor as for the managing director. And we wanted integrated measures driven by data from our ERP system." Dowding believes that reconciling lagging indicators (such as supply chain performance or waste) with leading indicators (sales or deliveries) is critical, and warns that there are several stages to achieving that. First is identifying the measurements required, and second is "understanding that data is only as good as your least competent user". Hence his emphasis on the need for business analysis to determine the right KPIs and thus what data needs to be collected and analysed. Hence also his focus on education and training. Beyond these, he advises users to plan for three further aspects: capturing data (ideally automatically), publishing exceptions and enabling decision-making. But there's more to this than meets the eye. So much depends on whether the goal is incremental and essentially inward-looking manufacturing and business process improvements (against parameters such as quality, cost and delivery performance), or strategic and outward-looking – aimed at transforming business direction – in which case you're after information-driven analysis of capacity, markets and the like. John Tripp, director of Goldratt TOC and a seasoned consultant in applying Theory of Constraints, urged users to consider the greater, strategic goal. "What questions are we going to ask of this IT? Where's the next market? What is going to give us rapid growth? What do I need to put in place to make that happen?" If so, he said, we must also use BI to understand the real difference between costs and selling price, true profitability of products, and which products have to go across which constraints – whether that's whole departments or manufacturing resources. "Very few systems are configured to answer those questions," he observed. Good behaviour Good point, but whatever your reason for installing BI, getting the metrics right isn't even just about considering business performance by function – it's also about the impact any metrics might have on behaviour. As Paul Williams, IT manager of a well known engineering consultancy, eloquently put it: "If you think of metrics as a vector diagram, individuals tend to move in a direction which is the sum of those vector forces." So getting the metrics right is very important – and you'll also need a mechanism for tweaking them as business needs change. So, which metrics? Infor's Field suggested starting by deciding what the business wants. "Define the critical success factors – what you have to do to succeed – and then which KPIs will influence them, otherwise you'll be collecting the wrong data." Pendse confirms that this is one of the greatest pitfalls: metrics in general are not well chosen and the result is invariably suboptimal performance. His suggestion: look for KPIs that your business would find useful year after year. Equally, Gradient's Metcalf considers temporary KPIs essential to enable specific business changes – as long as they're reviewed in line with development and removed on completion. And another useful pointer from Tripp: "Remember to focus on the outcomes you want, not the activities." How many KPIs you need will depend on the business, its objectives at the time, and the employee's role. Pendse believes half a dozen KPIs is probably right at the strategic level, but more as BI systems seek to drive operations for middle management. He cited one specialist steel manufacturer that used a BI system to discover its niche, define pricing, drive sales behaviour and optimise production operations, using historic and real-time plant data. At the other extreme, Andrew Knights, managing director of intruder alarm manufacturer Castle Care-Tech, made the case for just one overarching measure – for use as a cross-company success index to publish to employees. Tripp's view is characteristically controversial. He advocates focusing on the simple, top-level things that matter for taking your company to the next level, instead of getting bogged down in day-to-day complexity. "The essence is to understand the strategic constraints – the few things you can adjust to make a difference – and measure them," he says. Meanwhile, Dowding's experience suggests four or five top-level metrics, with three or four more underpinning each. And finally, data accuracy, latency and making the BI system work for your people. There's a strategy versus operations divide here. Tripp insisted that, if it's business transformation you're after, data accuracy isn't critical. Indeed, he believes managers get far too hung up on it. On the other hand, Dowding made the point that, at the operational level, the last thing you need is production managers arguing with sales people or purchasing people over data. For him, accuracy at the point of use is key – and achieving that is about education. As for data latency, SAP's Neale suggested that it's four times more expensive to get real-time data than information that's a few hours or days old. His advice: "Not all data is created equal: you might want defects data in real time, but other stuff might wait overnight, or for a week. So spend as much as it takes to fix the data required to fix the key processes." It's fitness for purpose again. He believes that investing in data quality stewards, who own the data accuracy in their domain, makes a lot of sense. Finally, what about your people? There are several aspects to this. First, it's essential to build a business culture that expects its people to use the data coming out of the BI system. That means investing in business analysts that understand how departments interact. Second, senior sponsorship for any project of this nature is key. Third, business users must be the drivers, because they know what it's about, can change the project scope and direction, and will deliver most benefit. Said Pendse: "If business users don't want to get involved, and won't help with definitions or correcting data, don't do your BI project. It will be a waste of money."