Procter & Gamble does it. So does communications giant Motorola. Bayer, Siemens, HJ Heinz and Vodafone are also enthusiasts. The “it” in question? Optimised e-sourcing, a technique rapidly gaining favour as a tool for wringing even more value from e-auction or RFQ-based tendering. Just as industry portals and reverse auctions came to represent leading-edge procurement practice in the late 1990s, so optimised e-sourcing is being touted as the equivalent for the second half of the present decade.
But do the claims being made for the approach really stack up? Is optimisation, as its proponents claim, a surefire plug-and-play winner? And, critically, is optimised
e-sourcing a broadly based solution, deployable by a wide range of businesses for a wide range of spend categories, or merely a handy niche “point solution”? These are the questions that top the CPO’s optimised e-sourcing agenda. But the accepted answers – “yes”, “no” and “maybe” – aren’t as cut and dried as they are billed.
The basic idea behind optimisation is straightforward. When a company issues an RFQ, or sets up an auction-based “sourcing event”, it is essentially inviting bids on pre-determined “lots” of items, each of which is expressed in terms of parameters such as quantity, specification, delivery date, and delivery location. In other words, it’s the buyer’s view of what he or she wants; the supplier’s job is to indicate the best price that they can offer against each item.
But the buyer’s perspective isn’t necessarily the one that yields the lowest overall cost, and it’s this awkward caveat that leaves the door open to optimisation techniques. Some suppliers will be able to offer a better price on some items than others. Some will be able to offer even better prices, but through variations in the bid parameters: offering different specifications, different qualities, different delivery locations and so on. Optimised sourcing allows suppliers to bid “expressively”, as the jargon puts it, and so indicate any better bids they could submit if, for example, the quantity, specification or delivery requirements were varied.
The trouble is, for the buyer, making these comparisons is a time-consuming business. And complicated: if, say, a dozen or so suppliers are making expressive bids on several hundred items, how are they to be evaluated against each other so as to pick out precisely the best combination of offerings that yields the lowest overall cost? It’s certainly possible to make some sort of judgment, but equally possible that any human judgment will fall short of strict mathematically based optimisation. And with hundreds of thousands – or even millions – of pounds, euros or dollars at stake, the difference between an optimised selection and a judgment probably won’t be negligible.
Which is where technology comes to the rescue. Computer-based optimisation, observes Gregg Brandyberry, vice-president of global procurement systems and operations at GlaxoSmithKline in Philadelphia, “consumes half the resources, is twice as fast, and delivers twice the savings”. That’s on the basis of a straight like-for-like comparison between a human-based approach and a technology-based approach, he says, drawing on data that the company has gathered from almost 4,000 e-sourcing events carried out since 1999.
But optimisation also yields other benefits that would elude any human-based approach, adds Brandyberry. For the third year in a row, GSK is about to undertake its global hotel sourcing event. For the purposes of this exercise, the world is divided up into around 220 regions, and bids sought from almost 1,250 hotels through a combination of electronic sealed bids and reverse auctions.
Previously, he says, such a highly granular approach simply wouldn’t have been possible: the company would have either had to take a broad global look at hotels, or examine fewer locations. Like-for-like comparisons are thus impossible. “We just didn’t have the resource to do anything more in-depth,” says Brandyberry. “Now we get all that bidding and negotiation out of the way in just two or three weeks.”
Fast-food restaurant chain McDonald’s is another organisation that has used optimised e-sourcing techniques to good advantage, says Dr Thomas Schachner, Vienna-based senior director of logistics and distribution within the company’s worldwide supply chain management function. For the last two years, he explains, optimisation techniques have been used to analyse expressive bids and “what if?” scenarios on ocean freight cargoes (primarily of toys) sourced from the Far East and shipped to the US and Europe.
Owing to high service-level requirements and tight timetables, he adds, the optimisation analysis has not primarily been used to make the final decisions of freight allocations. “It’s an extremely good tool for collecting bids and running ‘what if?’ scenarios, and gives us the basis for the final sourcing decisions,” says Schachner. And while he can’t disclose cost savings, he’s happy to report that despite significant price increases in the ocean freight rate marketplace over the past two years, McDonald’s has managed to avoid paying any more.
Motorola is more explicit about the savings it has achieved through optimisation. Beginning in 2002, the company conducted 200 optimised e-sourcing events totaling nearly $7 billion, equivalent to 40 per cent of its total spend on direct and indirect materials and commodities. In 2003, a further 450 events saw it source nearly $10 billion of spend, equivalent to about 56 per cent of its total spending. By the end of 2003, says CPO Theresa Metty, Motorola had captured nearly $600 million of savings through optimised e-sourcing.
Food manufacturer HJ Heinz is another convert. As with many of the companies using optimising techniques, it’s reluctant to share precise details, which is usually an excellent indication in itself that a technology or tool is more than mere hype, and is capable of making a genuine competitive difference. Nevertheless, it emerged last year that Heinz has run over 1,000 auctions covering ingredients, packaging and indirect goods, and saved more than $60 million, and that of the 150 or so events that the company ran in Europe, around 40 per cent employed optimisation techniques. By 2006, reports Rob Hemsley, general manager, European purchasing, Heinz expects that optimised auctions will account for nearly all of the 60 per cent of its multi-million-dollar European spend that will be e-sourced.
With testimonials like these, the buzz surrounding optimisation is predictably hot. But, equally predictably, that leads to confusion in the marketplace about what exactly optimisation techniques are, and which vendors are best equipped to provide them.
Ariba, for example, acquired some optimisation technology when it bought FreeMarkets in early 2004. These were contained within the latter’s FullSource offering, says Daryl Rolley, a former FreeMarkets executive who is now senior vice-president of Ariba’s international operations. “Ariba still offers optimisation as part of FullSource, but we’re also saying that going forward, optimisation will be built into our other tools,” he says. The plan is to add an analysis layer into sourcing events that can be switched on or off by companies, depending on whether they decide to use optimisation techniques or not.
That said, Rolley is to some extent dismissive about the power of optimisation tools. “The algorithms are out there – you can get a linear programming algorithm for $100,” he says. “Linear programming and similar algorithms aren’t new: what is new is applying them to sourcing. Companies will argue that their algorithms will deliver a superior result, but the skill lies in applying a given algorithm to commercial procurement in the real world.”
In defence of mathematics
While few people would argue with the need to blend mathematics with solid purchasing good practice, not everyone is so dismissive of the complexity of the mathematical techniques involved. At Procter & Gamble, for example, optimised e-sourcing reportedly accounts for over 10 per cent of the company’s $28 billion annual spend, delivering hard savings of $120 million in the fiscal year ending June 2003, and over double that in the year ending June 2004.
The expressive bidding optimisation techniques involved are “a combination of linear programming and mixed-integer programming, coupled to a very efficient solver,” says Dennis Begg, P&G’s associate director of purchase innovation. Although auctions in themselves are a zero-sum game, he notes, “by allowing suppliers to bid to their own strengths, and squeeze out their own inefficiencies, expressive bidding actually adds value”.
But in seeking to emulate early adopters such as Procter & Gamble, Motorola and GlaxoSmithKline, companies face something of an ironic quandary. Intent on playing suppliers off against each other, they find themselves with a paucity of appropriate vendors of optimisation techniques. What’s more, while several companies claim optimisation capabilities of some sort, just a tiny handful offer in-depth expertise.
Andrew Bartels, a research analyst with Forrester Research, says: “Right now, there are really only two vendors with what you might call sophisticated bid optimisation techniques.” The two in question? CombineNet of Pittsburgh, Pennsylvania, and Emptoris of Burlington, Massachusetts. Despite the fact that both companies offer optimisation techniques, he explains, they offer very different capabilities, and will appeal to different organisations for different reasons. Nor are they an either-or choice: a number of companies use both.
Emptoris offers optimisation through a self-service software suite. This was a conscious decision, says Emptoris’s director of client service, Sean Devine – one that was reached after deciding that there was a gap in the market for plug-and-play optimisation techniques so straightforward that consultancy hand-holding wasn’t required. Optimisation “scenarios” are collected through auctions or RFQs and then analysed, says Devine. But the Emptoris software “doesn’t do the business thinking for you”, he stresses. “It describes the scenarios in terms of the costs and trade-offs, but doesn’t make the choice for you.”
CombineNet occupies a very different niche. “CombineNet has a strong faculty of operations research people skilled in developing and deploying tools for expressive bidding and advanced e-sourcing,” is how Joseph Raudabaugh, president of AT Kearney Procurement Solutions, describes the company.
“If Emptoris specialises in spend analysis and spend management in the context of sourcing events, then CombineNet is focused on the higher-level science and mathematics of the events themselves,” adds Lora Cecere, a research director with AMR Research. Typically, she says, businesses will use CombineNet technology in order to optimise the outcome of quite specialised sourcing events, such as transportation or labour services. Emptoris, on the other hand, is often deployed more widely.
“Our customers typically execute between 10 and 20 events each year,” says Tom Finn, vice-president of sales and marketing at CombineNet. “The average category spend size that our customers use CombineNet for is $75 million, but we’ve done several projects this year that were between $750 million and $1 billion. It’s Pareto’s law: 20 per cent of the events are equivalent to about 80 per cent of the spend – and we specialise in those 20 per cent of events.”
So far, the experiences of the Princeton, New Jersey-based Bristol-Myers Squibb Company are a little more modest. The company’s first CombineNet-optimised e-sourcing event was completed in mid-December, explains Rob Hannay, vice-president of global strategic sourcing. Covering secondary packaging, the category equates to about $2.5 million, but is used right across the business by six internal customers.
Unconstrained, optimisation yielded savings of 50 per cent, but would have involved individual manufacturing sites dealing with up to a dozen different suppliers. An aggressive but more realistic scenario resulted in savings of 25 per cent, whereas a conservative scenario produced savings of 15 per cent. Discussions are ongoing with the relevant business units, says Hannay, but the likely outcome is within that range of 15-25 per cent. “It’s about twice what we would have obtained otherwise,” he notes.
And although the category is small, it’s essentially been a dress-rehearsal for larger-scale events. Air and ocean freight is next, he explains: a $25 million category with around 2,500 “lanes” or routes, and potentially dozens of suppliers. “It’s both a nightmare for us to analyse, and a nightmare for suppliers to show how capable they are,” he says. “Optimisation suits both us and our suppliers.”
BRIEFING:
Optimisation explained
The key to optimisation is allowing suppliers to make “expressive bids”. These, as the name implies, are bids that allow suppliers to express preferences – in terms of parameters such as quantity, specification, delivery date and delivery location – that vary the prices contained in the bid.
The trouble is, this at a stroke loses one of the key advantages of an RFQ or e-auction: the common playing field upon which suppliers bid. With every bid potentially differing from every other bid, how are comparisons to be made?
Enter optimisation techniques. These allow expressive bids to be evaluated against each other so as to identify the combinations of offerings that yield the lowest overall prices. Such “scenarios”, as they are termed, are then presented for selection. Reducing the supplier base for a category from, say, a dozen to five may be inappropriate for strategic reasons, but at least the cost consequences of doing that have been identified.
The business makes its choice and selects the scenario that meets its overall requirements – a reduction to, say, eight suppliers – knowing in advance the prices that this will entail.
Optimised e-sourcing, say experts, works particularly well in three distinct situations: first, large spend categories; second, where a category comprises an extended list of items; and third, where usage is widely distributed, with each “use point” comprising its own micro-market – packaging, for example, or hotel accommodation.
Vodafone, which has transacted around ¤200 million worth of optimised e-sourcing since mid-2003, reckons that optimisation generates a further 5-10 per cent of savings over and above those obtainable from a conventional auction.
Commercial manager Ginny Warr says the technique is principally used to source business services, particularly those that are consumed across dispersed locations, such as desktop computer maintenance. “It’s a lot quicker, and it uses far fewer purchasing resources.”
Malcolm Wheatley is a freelance business and technology journalist who writes for a range of leading UK and US publications
