Forecasting tools are beginning to creep onto project managers' desktops for a good reason: accuracy.
BY MALCOLM WHEATLEY
the absence of precognitive abilities, project managers—who must regularly figure out what resources are required, when, where and for how long, have it tough. “Without a doubt, accurate resourcing is hard to achieve,” says Adrian Tillin, business development director of Project Management Professional Learning, a chartered PMI Registered Education Provider based in High Wycombe, U.K. “People come to us and say, ‘We always have difficulties with resourcing.’ And I reply, ‘Yes—that's because it's difficult.’”
ILLUSTRATION BY JOYCE HESSELBERTH
Probabilistic techniques create resource ranges, but they aren't the whole answer.
Interpreting forecasting reports requires a sophisticated management team that can use a variety of metrics to make tough decisions when resourcing conflicts arise.
Important distinctions exist between forecasting the workload on people versus forecasting it on fixed items such as machine tools.
Forecasting is easier when project teams create project plans as early as possible—ideally at the bidding stage—to paint the fullest possible picture of resource requirements.
In recent years, best practice in making the difficult doable usually has revolved around understanding and quantifying the amount of risk and variability in a resource prediction. Probabilistic techniques such as Delphi turn pinpoint numbers into ranges on the project timeline: The resource might be required as early as here, or as late as there.
“My preferred method is to use the Delphi technique, supplemented by any existing reliable corporate data, apply the [program evaluation review technique] formula, and let senior management select the percentile ‘comfort level’ that they prefer,” says Paul Giammalvo, a Jakarta, Indonesia-based project manager who also has worked in Vietnam and Israel. “I prefer this to using buffers or contingencies. Experience has shown that work often expands to fill the allotted time, or that if there's money to be spent, someone will find a way of spending it.”
I Want It Yesterday
Trouble is, this approach relies on a mature, informed understanding of what those probabilities imply. “When there is uncertainty, there is no single path [on the network diagram] that is certain—and that is important,” says Arnold Ruskin, PMP, a partner in Claremont Consulting Group, La Cañada, Calif., USA. In such circumstances, whether management can meaningfully interpret a forecast range depends on how sophisticated that management group is, he asserts. “Some management is simply willful, and in denial,” Ruskin says. “They want it when they want it, and are not interested in what others are telling them.”
Under pressure, people will accelerate their work rate and [condense] tasks together. Machine tools don't do that.
President, B2B Analysts Inc.,
Cambridge, Mass., USA
But if best practice today is to put a “buffer zone” around a forecast to minimize its inherent error, best practice tomorrow might be more concerned with improving the accuracy of the forecast itself before wrapping it in a layer of contingency factors or probability bounds.
“Forecasting has had its ups and downs in corporate America over the last few decades, and forecasting certainly doesn't play a major role in project management today,” says Paul Robinson, PMP, vice president, The Eastern Management Group, Bedminster, N.J., USA. However, he thinks change is afoot. The approaches presently reshaping forecasting techniques within supply chain management also probably have an application within project management.
The advantages are obvious. Despite corporations' on-and-off love affair with forecasting, a huge body of knowledge has accumulated, especially in the area of consumer product demand. Even so, warns Dwight Smith-Daniels, an associate professor at Arizona State University in Tempe, Ariz., USA, caution is advisable. “The statistically based forecasting that is generally used within supply chain management is all about aggregating individual small units of demand over a relatively short time period—six months or so—to build up a picture of, say, the number of cans of baked beans that consumers will buy. That's not necessarily what resource forecasting in a project environment is about.”
Information technology can clearly help. Throw in, too, the overlay of the different management techniques and theories underpinning these approaches, and powerful synergies can result. Admittedly, there's a downside: When the IT industry applies a bandage, the approaches to automation come from very different directions. But although the resulting duplication might look wasteful, the result—as the dot-com era proved—is a Darwinian hothouse where only the hardiest of solutions will survive.
People come to us and say, “We always have difficulties with resourcing.” And I reply, “Yes—that's because it's difficult.”
Business Development Director of Project
Management Professional Learning,
High Wycombe, U.K.
Reko International Group of Oldcastle, Ontario, Canada, for example, manufactures one-off injection molding tools for the automotive industry. Each tool is a giant assembly of carefully machined parts that represents the culmination of 6,500 hours of machining and assembly work over as many as 18 weeks, says Process Control Manager Lido Zuccato.
“As soon as we receive an order, we create a timeline in Microsoft Project that shows every work center that the order is going to hit, and roughly in which time window it is going to hit them,” Zuccato says. This information then gets loaded into a project management-oriented enterprise planning system from Cincinnati-based Encompix Inc. Critically, this not only takes account of the resource forecast side of the business, but also the resource consumption side too—“work-in-progress,” to borrow a term from the manufacturing and supply chain industries.
Encompix, the first of a new breed of project-based enterprise resource planning (ERP) systems, uses a project plan from Microsoft Project as its starting point instead of a master production schedule or forecast, according to Chuck Stewart, Encompix vice president of operations. With a high degree of granularity, labor hours can be booked against the project, subassemblies built and finished, and costs accrued. The top-level view of the project status is consistently updated—not just within the Encompix scheduling system, but within Microsoft Project as well.
At CODASciSys, Chippenham, U.K., a contract software subcontractor closely involved in the European Space Program, a different approach to forecasting is being taken. Having long been frustrated with the inadequacies of its spreadsheet-based approach to forecasting and resource consumption, the company turned to project and resources manager Caroline Coard.
THE IMMOVABLE OBJECT
When applying forecasting techniques from industry to project management problems, don't forget the important distinctions between forecasting the workload on people versus forecasting it on fixed items such as machine tools, warns David Dobrin, president of B2B Analysts Inc., Cambridge, Mass., USA. “People resources are compressible and expandable in a way that machine tools often aren't,” he says. “Under pressure, people will accelerate their work rate and [condense] tasks together. Machine tools don't do that.”
Machine tools also are more directly comparable with each other. One lathe or machining center usually will churn out widgets at much the same rate as its twin on the other side of the factory aisle, he warns. “A person hour isn't the same across a pool of individuals; different people have different work rates.”
“Create a project plan as early as possible, ideally at the bidding stage,” adds Kenny Ingram, consulting services manager with enterprise planning system vendor IFS of High Wycombe, U.K. “Lots of companies don't do this, but the advantage is that you'll have a project plan for every serious bid you've got out there and every live job—which gives you the fullest possible picture of your resource requirements.”
IFS enterprise software has become popular in the project environment, he explains, because the Scandinavian countries in which the company originated, and which still represent a core market, are light on traditional batch manufacturing, but big on heavy projects such as shipbuilding and oil exploration.
“If necessary, use templates to make it easier, push it out to the planning tool and resource it,” he urges. “Or copy over a previous project plan. It's like ‘rough-cut capacity planning' in the supply chain world—but in a project environment.”
Market conditions can change very rapidly, and the forecast ROI may be out of date.
“The ability to forecast and report right across the company is important to us, and it's difficult to do if you're spreadsheet-based,” Coard says. Given the nature of CODASciSys' work, the need is to identify resource needs in man-weeks or months, not hours or days. “Project granularity has a bearing on it: We rarely need to identify a gap in a diary tomorrow,” she says.
While CODAScisys has chosen to replace its spreadsheet systems with a project planning tool from British firm SharpOWL, the company so far has implemented only the Time and Expense Manager module, not the Forecasting module. That's because, explains Coard, the Forecasting module relies on Microsoft Exchange and Outlook functionality, but CODAScisys is a Lotus Notes-powered company. Even so, the allure of better project forecasting may well be enough to tip the balance, she asserts.
At insurance company The Insurance House, Marietta, Ga., USA, still another approach is underway. It falls to Director of Strategic Business Services Robert Golden to figure out the resources that will be required to complete the company's projects. “We have a robust project management methodology, based around A Guide to the Project Management Body of Knowledge (PMBOK® Guide), that provides us with estimates of what resources are required and when they are required,” he says.
These estimates then are fed into a resource planning and portfolio management system from Niku Corp., Redwood City, Calif., USA. “If we need a Web developer with particular language skills on a particular date, we'll be presented with a choice of appropriate candidates,” he says. “If we had two or three individuals with the right skills profile, but only one was available, we'd only see that one person.”
If no one were available—perhaps fully booked onto another, but less important project—this fundamental portfolio resource conflict heralds an element of unification. Better forecasting can help avoid such conflicts, but when they happen, tough decisions must be made.
A quick and dirty solution is to look at ROI and select the project with the highest net present value. Or, if cash flow has a bearing on the project return, look at the Internal Rate of Return. As Giammalvo points out, “In practice, decisions are rarely made on the basis of a single metric.”
At The Insurance House, for example, Golden agrees that ROI scoring offers a way forward, but one that must be chosen carefully. “Market conditions can change very rapidly, and the forecast ROI may be out of date. At the very least, check that the ROI you thought you'd get from a project is still likely,” he says.
At CODASciSys, Coard is more blunt. “Conflicts are handled by negotiations, but the ground rules are clear: The most important project always wins,” she says. But “most important” means different things, depending on whom you ask. David Hurwitz, vice president, marketing at Niku Corp., offers a thought: “Look at metrics, sure, but also measure how well a project is aligned with your strategic goals. If it isn't, then why are you doing it?” PM
Malcolm Wheatley is a U.K.-based business writer who has appeared in magazines such as CIO, CSO and MSI.
To participate in an online discussion on forecasting tools with your colleagues, visit communities.pmi.org and go to the PMI Member Community.
PM NETWORK | APRIL 2004 | WWW.PMI.ORG
APRIL 2004 | PM NETWORK