Hard-nosed advice hard-won experience

There’s more than one way to skin a cat, as the delightful saying goes: just so, but doubtless some work better than others. And that’s the point of this feature – to establish what works (best) (and what doesn’t). Not for skinning the poor creatures, but for implementing manufacturing enterprise systems (ERP) and going on to get the best out of them – in any organisation, no matter what its size, type of operations, industry or role. Why? Because it’s a seriously major undertaking, often hopelessly underestimated. And because the chances are you may not have done it before – or at the very least not for a very long time – and uncharted territory presents challenges: how you handle them can make or break your project. So you need all the objective and informed help you can get – in terms of understanding the scope of what you need to consider, the resources you’ll need, the scale of investment, procedures and infrastructure. And who better to turn to than those that have been there and done it? There’s nothing quite like hard nosed, hard won experience. The following then is an accumulated wealth of key system implementation success factors, as seen through the eyes of those who can put their hands on their hearts and say they know. I spoke to: Ian Martin, systems manager at electronics contractor JJS; Brian Metcalfe, systems manager at Leyland Trucks; Adrian Conway, operations director at car mat and baby products manufacturer Cannon Avent; aerospace components contractor Prematec’s QA manager Roy Thomas; Peter Young, company accountant at Jordan Grand Prix; Steve Davis, project manager at brush manufacturer GB Kent; and Ewan Tozer, IT and engineering manager at chair maker Giroflex. Establishing project scope First then is the issue of scoping your project. How do you do it? Cannon Avent is typical of the broad swathe of SMEs. It operates out of two locations – Tottenham making the car mats for GM and Ford, and Glemsford Suffolk making baby feeding bottles and silicon teats for Tesco, Sainsbury’s, Boots and Mothercare. Operations director Adrian Conway says he had to replace a 15 year old Hoskyns HP 3000 system which was beyond support. He says key drivers were the need to respond to a clear growth forcast and to bring accounting periods down from four weeks to five days. So his scoping process was simple, he says. The firm had to move up to modern client/server to get the operational and productivity benefits and it needed ERP to get the business integrated. Aerospace firm Prematec, which machines, for example, 60ft spars for Airbus wings from solid aluminium, was luckier than some: the firm is owned by a software engineering company, so much of the groundwork was done for it. But QA manager Roy Thomas says the new business system was needed primarily to control material movements throughout its large DNC and machine shops, also providing engineering project management and, most important, traceability. So scope of its Kewill ERP system had to include shop floor scheduling and data collection (with barcoding), as well as time and attendance, quote and order processing, material control, shipping control and reports – all linked to engineering planning and the machine controls. It was rather a different story at the world’s second largest truck manufacturer, Paccar-owned Leyland Trucks near Preston. Its project involved replacing some half dozen mainframe and HP 3000 mini legacy systems which collectively had supported the company to consultants Oliver Wight’s MRP II Class A status since 1988 – but had passed their sell-by date and “were far too expensive”. Leyland spent £5 million on its Oracle Applications ERP from 1995 to 1999 and built one of the most advanced build-to-order truck assembly facilities in Europe, with an output of 65 trucks per day across the Daf and Foden range. Systems manager Brian Metcalfe says scope had to include: MRPII and production scheduling; bills of materials (BoMs); engineering change control; assembly process planning; supply chain planning; purchasing; inventory control; order management; demand management; and financials, with .comprehensive manufacturing and engineering systems integration. But everyone is different: Sliverstone-based Formula 1 company Jordan Grand Prix sees itself as part manufacturer and part race team. Listening to accountant Peter Young, it’s clear that this very contradiction is still dictating project scope. On the car build side 90% of the manufacturing is outsourced, so there were supply chain concerns, but in the machine shop and assembly facility, the issue was managing stock in a company used to breaking all the internal business process rules to win. Its Scala ERP system, which remains not yet fully implemented, includes financials, stock control, manufacturing production control, purchase ordering, sales order processing and fixed assets. The goal remains full integration, but Young says there are still stock tracing problems to be resolved. Already a pattern is emerging, and it’s common sense. Good scoping is all about focusing on what the business needs to achieve first – not just the operational IT advantage – and then mapping elements of IT to that. £3.5m turnover, 50 employee brush maker GB Kent reinforces the view. Steve Davis, Kent’s project manager, says it had to replace its old AS/400 which was heavily bespoked, not Y2k compliant and no longer supportable. But the firm deliberately went for a strategy rethink, instead of simply taking advantage of modern IT, and found the software scope it needed pretty inclusive. “We wanted NT, SQL Server, Crystal Reports and ODBC links to development tools like Visual Basic, Access and Excel,” he says. But although there was no need to re-engineer basic production (“BoMs and routings were fine”), he knew the firm needed better visibility and accuracy. So scope was defined as improving manufacturing and production planning, and information to the business. And that meant implementing shop floor data capture, time and attendance, capacity resource planning and shop floor scheduling. And Ewan Tozer, IT and engineering manager at Welsh office chair manufacturer Giroflex (see feature on page ??), confirms the analysis. His firm underwent a management buyout in January 1998 and used the opportunity to revamp its software from top to bottom with the sole objective of being able to compete head on with bigger international competition. And for him that meant “looking for the business critical processes.” In Tozer’s terms, these were “on-line product configuration (there are thousands of permutations); price configuration (there are multiple channel deals); getting activity-based costing; and improving forecasting.” And there were materials management, assembly and order fulfilment. That translated into software for ‘available to promise’, advanced planning and scheduling (APS), fast rules-based product and price configurators and so on. Giroflex actually selected a best of breed integrated solution based on Fourth Shift ERP. JJS’ Ian Martin provides another angle: he believes his approach of keeping the initial scope wide open, but then locking down and delivering in phases was a good way to go. It meant opportunities weren’t missed, yet success, benefits and confidence were delivered early, thus keeping the momentum going over a protracted period. He accepts there is a danger of project creep, but insists that unless there are insurmountable deadlines, getting the best system for the company is always more important than meeting arbitrary ‘go live’ dates. What about an ITT? Views from the team differ widely on the usefulness of ITTs in finding the right IT supplier/partner. Leyland’s Metcalfe says his team produced a detailed ITT – and that it was very effective. “It was put together by experienced IT people,” he says and it meant very rapid progress to a realistic shortlist. But Cannon Avent’s Conway says ITTs are a waste of time. “Everyone will just tick all the boxes. You need to get understanding: get vendors to see how you work.” So, there’s the two extremes. Leyland Trucks used its ITT to do some weeding and then went for face-to-face visits; a day spent at each vendor “with a hit list of topics”. Discussions were already quite detailed: he cites, for example, assembly processes involving two parallel lines but handling a wide variety of volume and build mix – having to be able to respond to late changes, both of orders and engineering details that have to flush throughout the system. In the event, Leyland selected Oracle in favour of Avalon (then Dataworks, now Epicor) “since at the time it offered the best mix of functionality and the best opportunity to add our own to match our business processes”. He describes these as “our competitive strength”, and says in Leyland’s case the implementation was not a catalyst for change. Meanwhile, Adrian Conway made it an open competition. He brought all his seven or eight shortlist contenders together for a day at Cannon Avent’s Tottenham HQ to “show them what we were doing”. Those that felt they could make a sensible offer stayed on, he says, and he rapidly selected McGuffie Brunton on NT – not least “because they were financially sound and it was a good product.” McGuffie then ran a pilot: “they took our data and their system to the Suffolk site for two days, and all departments – sales, manufacturing, warehouse and distribution – were all invited to view.” Thereafter, Conway says he incentivised the firm by establishing payment on a per modules delivered basis – which, he says, worked well! In Kent’s case, Davis came up with another solution: he used MCS to shortlist four or five packaged software vendors, including Efacs, Visual Manufacturing, Kewill and Swan. He says it was relatively straightforward: Kent is an SME; he had decided on NT; strong manufacturing-centric IT was essential; and it had to be a sub-six moth installation. In the event he chose Efacs. ‘Vanilla’ applications or bespoke? Well, which is it to be: ‘vanilla’ or bespoke?. Davis is vehement: like almost everyone that’s suffered the consequences of bespoke systems, he insists you must go standard. “Bespoke is a big, big mistake,” he says. “Every time you want to upgrade it’s very expensive and time consuming.” And he claims that his bespoke systems were the reason for the firm “standing still for 10 years.” He urges anyone buying a packaged system to go vanilla, and if they must customise, don’t tamper with the code itself. He explains that where he needed extensions to his Lynx-implemented Sage Tetra CS/3 system, rather than insert code, programs were written alongside. “We tap straight into the data: we don’t change the code.” It’s a common theme. JJS’ Martin says that rather than change the software, his company changed many of its business processes to meet the ERP system: it meant not being left behind when new versions were released – and, he says, getting closer to perceived ‘best practice’. Leyland’s Metcalfe says Oracle consultants carried out a six month IT analysis “that delivered a good implementation plan,” suggesting around 60% standard and 40% bespoke. But all customising was done using Oracle’s own tools, again solving the upgrade problem. Metcalfe says his system has “gone through at least three new releases without any problems: it’s a sound change management system.” Cannon Avent’s Conway sums it up: if you go down the bespoke route, upgrades will cost you dear – in time and money. And bespoking tends to reinforce outdated and wasteful procedures. By and large, he says, flexible best practice has been captured in the main ERP vendors’ modules – so why not take advantage? Pitfalls and promises Many of us are looking for a quick fix, and we’re promised it from every quarter – not least the ERP software vendors. And by and large there is a consensus that their accelerated implementation methodologies and pre-configured and templated industry-specific modules do help. Giroflex, for example, used Fourth Shift’s implementation plan supported by Microsoft Project. Tozer says there were several hundred actions and they didn’t update as often as they should. But he says the method helped keep them “clearly focused on the seven or eight key aspects”. And Cannon Avent’s Conway says his experience of modern pre-configured software modules was also good, accelerating the time to the go-live. “It all worked,” he says. But there’s more to it: JJS’ Martin reminds us of the importance of testing and training. He took an unusual iterative approach, developing the system with testing, training and changing processes to match users’ ideas. “We were more concerned about getting it right than worrying too much about deadlines.” As Giroflex’s Tozer says. “You can’t afford problems when you go live: there’s no going back.” But sometimes you have to: Cannon Avent’s implementation, which was expected to take a year, actually took just nine months. But then it crashed. Conway says they had put in a new integrated wide area network (WAN) infrastructure to support telecomms, the database systems, CAD/CAM and all the IT. All systems were tested individually, but when the McGuffie system went live with 120 users across the sites (big for NT at the time) the network collapsed – and that cost Cannon three embarrassing months. As the ghost said to Edmund in Black Adder, “Somebody cut my head off at one point, but otherwise everything went swimmingly.” With refreshing candour, Conway says they installed what appeared to be an infrastructure sized for the job, but no-one had the expertise of large scale networked NT. “This was our big mistake,” he says. “We were leading edge for NT: we were up with the system for 12 hours and then we had to abandon it.” And there was more: in the rebuild phase Conway says staff forgot their training; then when live the NT system rapidly outgrew the new server; and no-one had anticipated NT’s reliability: “HP just never went down, but early NT was unreliable.” So what does a good team look like? Despite the above, Conway used what’s now widely acclaimed as best practice in putting together his implementation team. He took the best individual from each department, building a team of half a dozen and relieving them of their normal duties for six months. Not only did this ensure a good, productive team, but it developed the individuals concerned who afterwards went on to greater things within the company because of their learning opportunity. And this is the general pattern. Leyland Trucks was essentially the same but on a grander scale: Metcalfe was able to field a team of five business analysts from across the company and 20 technical staff. But he agrees that small is beautiful: in his case the steering group comprised the full Leyland Trucks management team, but the project was broken down into modules each led by a single management team member. And it worked: Leyland went for the ‘large chunk phased’ approach, with all core business processes, including BoMs, purchasing, engineering change, manufacturing management, material control, supply chain planning and cost management going live after 18 months. Front end processing, order management, demand management, production scheduling and accounts followed two years later in June 1999 – with parallel running to verify order and demand management. Then in September last year supply chain management and quality management went live. It worked for Kent too: Steve Davis says they achieved virtually everything on time and on budget. Five days of consultancy, based on training the trainers, paved the way. Financials went up in just six weeks – bringing instant confidence and credibility. Everything else went live three months later. If anything it was too fast, and Davis doesn’t recommend it: “We had to get it in very quickly, so there was no pilot – we just did electronic data downloads and went live.” The last word on this comes from Giroflex’s Tozer. Given his time again, he says he would go for a bigger implementation team. “It’s a hell of a project: we were working 100 hour weeks by the end. We should have recruited more IT people earlier.” The ERP is in: now what? JJS’ Martin insists that to maximise return on the ERP investment, no-one should treat it as a project with an end. “It’s an on-going process,” he says. Since ‘completion’ two years ago, his projects have included: developing the MRP II to generate integrated works instruction cards and new reports; and changing the Datastream SFDC mechanism to a fast intranet feed on the back of its on-line business management system. He’s also installed the CS/3 ‘fixed assets’ module, carried out version upgrades and is moving to upgrade production with Preactor finite capacity scheduling (FCS), which will also cut work in progress and inventory. His recommendation: send someone from IT to spend time (not hours; a few days) with each of the departments in turn – purchasing, manufacturing, sales – to see the shortcomings and scope and shape of improvement from the users’ perspective. Just one person can do this, he says. Meanwhile, Conway says Cannon Avent implemented a small data warehouse to keep queries and reports away from the transactional system: that’s now moving up to a relational database. He’s also installed EDI and payroll software, and is now looking at integrating time and attendance, warehouse management systems and automated material movement and management software. And the firm is developing e-commerce for the Avent range. For him, APS and FCS are not on the horizon: with its current growth path the firm’s only problem is throughput – building enough stock! And there are no engineering problems. Conway: “We’re not building oscilloscopes or fire engines!” Jordan’s Young says his next objectives will include reducing paper flow, by moving to email with workflow and archival. He’s also implemented Crystal Reports with direct database access, which he’s found “tricky to learn”. Now the firm is looking at SFDC following on stock control for more accurate real time information on stock booked out using barcodes. And as for Kent, Davis says he has prioritised trading with EDI, and is now working with Efacs to implement HR, also integrating SFDC with the payroll system. Next up will be a marketing-orientated Web site development, and Davis says he also wants to start using quoting and estimating software. Finally, Leyland Trucks’ post implementation projects have been driven by expediency, like handling product transfers from other Paccar plants – involving integrating alien BoMs with the Oracle system. For the future, Metcalfe believes that better optimisation of the supply chain will be key. He envisages improved integration and automated supplier processes over the next two or three years harnessing Web technology: Oracle vision of ‘e-synchronisation’. The big one: is it worth it? A resounding “yes”! Metcalfe says Leyland’s results are clear. He estimates savings of £1 million per annum on operating costs and the benefits that derive from a clear, accurate and up to date view of the supply chain. Leyland, he says, now runs MRP nightly (it was just at the weekend), delivering a new supply chain picture every day which has increased stock turns “from 12 two years ago, to 30 now” – and cut inventory from “£11 million to £6 million”. And he adds that decision-making is more informed with better quality and accessibility of information. And that’s pretty typical. Prematec’s Thomas says his system has changed the way the firm works –with work in progress now entirely visible, operations improved, and production, performance, quality and efficiencies constantly monitored, allowing the firm to move to accurate activity-based costing. Cannon Avent’s Conway is similarly pleased: not only has the firm achieved its financial reporting requirements, but stock turns have increased.

• If you do nothing else There’s nothing quite like hard-nosed advice from those with hard-won experience. The following is a collection of ‘if you do nothing else then at least….’ Ian Martin, systems manager at electronics contractor JJS, advises: “Go to a reference site – and don’t be steered by the software vendor or the IT manager: talk to the users themselves. They don’t have an axe to grind.” And he continues: “Don’t believe people that tell you support from the IT vendor is enough. Even on Priority One they only get back to you in half an hour: that’s not solving the problem that’s just returning your call! You need on-site IT resource: without it, you expose yourself too much.” Brian Metcalfe, systems manager at Leyland Trucks has another perspective. First and foremost, he says, you must have “the involvement, commitment and ownership of all the users”. Next he urges prospective ERP teams to draw up a “very solid implementation plan and scope” right from the start, and above all to ensure that “good people do it.” Car mat and baby products maker Cannon Avent’s operations director Adrian Conway has but one key message: don’t go ‘bleeding edge’. “Be a follower not a leader,” he says. Beyond this, he adds: “be very comfortable with your supplier – the company directors, not just the salesmen.” Success demands a long and lasting relationship, and you don’t want litigation or liquidation getting in the way. And he urges: “Get a good, committed team in house.” Aerospace components contractor Prematec’s QA manager Roy Thomas is pragmatic. “Plan very carefully for timescales and responsibilities; and define very clearly who is responsible for what – otherwise it’s a disaster, and you end up doing it yourself.” He also emphasises the importance of “a well balanced team” which he says must represent all key parts of the business – Prematec realised the value of this rather late in the day. Finally, he adds: “document the way you do things as you go along.” Memory is a fragile thing. And here are three more. Formula 1 firm Jordan Grand Prix’s accountant, Peter Young: “You can never do too much planning; don’t neglect training (we probably didn’t do enough); and lead by example.” Steve Davis, project manager at brush manufacturer GB Kent & Sons: “If you’re looking for fast implementation keep the steering committee small; target specific critical areas you want to get in quickly first; and go for small successful project chunks – otherwise you’ll be there for ever.” Finally, Ewan Tozer, IT and engineering manager at chair maker Giroflex, says: “Force a demonstration of the system. Focus on key processes where you think you’re different. And get the best people in the business on your implementation team.” And he adds: “Remember, your relationship with the IT vendor will be one of the most important you will ever have.”