“Trust your gut,” the stalwart advice of many a successful leader feted for their intuition. But a new generation of bots is proving far better at using data to make crucial choices – which has implications for procurement leaders
There’s no doubt that managing a supply network requires skill and experience. While it takes time to earn such traits, the mark of a true leader is often considered to be intuition; an innate sense for business which can’t be learned. But as the Harvard Business Review says, “anyone who thinks that intuition is a substitute for reason is indulging in a risky delusion”.
There’ll always be a need for business acumen, and some people’s strengths will naturally sit in strategy, but in the midst of a global hunger for technology and the incredible quantities of data it produces, not taking full advantage of this information seems stubborn at best.
For procurement and supply, the web of interactions is compounded by squeezed margins, increasing volatility and the need to mitigate ever-emerging risks – and that’s before any mention of social issues. In a time of increasing complexity and a persistent pandemic, why would any leader choose blind instinct over information when making decisions? For many, the reality is that we simply don’t understand how to use data.
Visionaries sold a dream of using technology to spend more intelligently, more efficiently, and of a level of network visibility that supports razor-sharp decisions in an instant. If this dream has come true for some, it’s certainly not the case for the majority. Instead, there’s little clarity over how to manage information, maintain and interpret it, meanwhile data is flooding into organisations that don’t know how to grasp the opportunities it presents.
As a result, it can be hard to gain value from such an investment and almost impossible to secure the additional costs needed to fix it. And therein lies the problem, or at least a significant part of it. Data is only useful if you can make sense of it. Today, in many organisations, it’s common for information to be poorly managed, inaccurate or stagnating in silos. Quite often all three.
Supply must move with the times
“From a procurement perspective there’s a lot of ways that manifests,” Susan Walsh, a consultant on data quality in supply chain tells Supply Management. “You see multiple versions of the same supplier, old or outdated companies, or information that’s not classified or is partially missing – either way it’s not that useful.” From this vantage point, the much-vaunted clarity, insight and transparency that data can offer are a long way off. This is the norm, says Walsh, who often finds clients’ critical information, such as descriptions, POs or invoices, are vague or even nonexistent. Frequently, she says, there is no taxonomy to the way the information is classified either.
“It’s a situation that hasn’t changed much in 15 years,” says Brian Glick, founder and CEO of logistics SaaS firm Chain.io. “What we see is companies collecting data without a thoughtful and well-organised strategy for what they’re going to do with it. Companies will jump into data exchange without understanding business process first or what they want to accomplish – and when you start projects from a data or IT perspective without a business objective, you fail almost all the time.”
Walsh agrees: “If you’ve classified your data properly and accurately, then you can be confident in it. If you don’t know what you want, start at the end. What do you want to report on? What do you want to achieve? Then classify and clean it so you have that information.” This may seem like common sense but it’s certainly not common working practice, according to Walsh. It’s a convoluted problem; 15 years ago, few would have predicted the extent to which data would govern the way organisations work today. And even fewer would have had the infrastructure or inclination to act upon that knowledge. As such, retroactively sorting data into a resource that is accessible, visible and useful has been something of a major cultural and operational shift.
“More often than not, companies will need a new taxonomy for their data, so we literally have to build it from scratch,” Walsh says. “In the organisations where that is the case, they are aware their data is valuable but there is so much to fix that it seems an impossible task, so it gets left – but that’s the worst thing you can do. Clean data drives value and profitability. We need to start thinking of it as an investment and not a cost.”
Another sticking point is tools. Microsoft Excel is the largest transportation management system in the world by number of companies using it, according to Glick, which is fine if you make around 1,000 deliveries a year but could be unwieldy when dealing with higher volumes. He says systems that rely on manually populated data sets are limited in a variety of ways, from the quality of the data itself down to the lack of real-time visibility required to effectively mitigate risk. So while Excel is good, persisting with it puts a business in an untenable position.
“We live in a world where consumers expect real-time. I can see my pizza being assembled in an app so, if I’m purchasing six containers of parts from China, I should have the same capability. That’s the expectation. Freight companies, for example, are now optimising for quotes from days down to hours, and they should be aiming for seconds.” Progress, he says, is a question of organisational structures and circumventing the status quo.
The value of cross-referencing
Once you have high-quality, clean data there’s almost no limit to what you can do with it. Where conversations relate to price, performance metrics and contractual obligations, third-party data is helpful because it can shed light on what’s really going on; when discussing delivery times, for example, data from the port recording when a container gated can be compared with the supplier’s information to assess both accuracy and level of trust.
And when analysing quality, check independent quality control data and correlate that with returns and rejected products at the consumer level.
“You want to collect raw data points as granularly as possible and build them into a story,” Glick says. “When you are able to collect high volumes of small pieces of data, you can use them to answer tons of questions so you’re getting more value from it. It’s worth way more than a single, purpose-built data set from a vendor that reports I had a 13% failure rate on the line. I can’t do a lot with that.”
The message is simple: collect facts, which, in essence, shouldn’t be hard. Glick advises viewing the supply network as you would a factory floor – there are inputs and outputs, and each component can tell you about one piece of the whole picture. Such a process may be valuable when, say, near-shoring supplies.
“You might be moving suppliers from China to Europe,” Glick says. “The European suppliers might not have the experience of working with companies like yours, or the data the Chinese company has. But if you’re collecting simple information, you can be very clear on what you want because if you’re clear, most companies can provide that.”
Towards real-time resilience
Gaining that all-important visibility has real-world implications. Against a backdrop of global volatility, supplier dependency and an expanding remit, the need for procurement to make smarter and better decisions has never been more important – but that’s only possible if you have an accurate picture of where you are.
“What we’ve focused on isn’t just having the data, but an ability to visualise it” CEO of Vodafone Procurement, Ninian Wilson, explains to SM. “Some people can look at a spreadsheet and know what is wrong, but not everyone can. So we’ve visualised our data into a control tower. If you go in and view our compliance dashboard, for example, you can see the different metrics we use by country and by category. On the efficiency side we have those metrics, and on the performance side we have those metrics, all dashboarded and in real-time, or near real-time, with analytics capability.”
That minutiae of insight provides a boost to resilience. Real-time data capability deeply affects the way supply networks are managed, Wilson explains, adding that Vodafone plans to implement a platform that maps risk globally. “If there was a terrible incident, such as a tsunami in Japan or an earthquake somewhere else, we’d be able to map our physical supply chains on to that and see if there’s a potential for disruption,” he says. “We’re not there yet, but we’re working on that now to map for better resilience.”
It is one of a growing number of use cases for external data that finds parallels in other areas of the supply chain that have, hitherto, been invisible. These uses are enabling companies to chart metaphoric maps of the territory with exacting degrees of accuracy. BT Sourced, for example, has established a six-strong team of data scientists to spearhead the organisation’s negotiations analytics – the aim of which is to connect internal and external data to gauge market sentiment.
“We’ve developed a bot that uses natural language programming to gather information on what suppliers are saying and what’s on their agenda,” says BT Sourced CPO Cyril Pourrat. “In so doing, we can gain a better understanding of what’s happening with suppliers and decide what negotiation strategy to take.” It’s an innovative and profound use of both data and technology, and one that forms part of a wider ecosystem of tools that are making sense of the complexities that underpin the company’s supply network.
“We are trying to understand every element of our position,” says Pourrat. “We want to know what we are spending, what our position is in the market, what our suppliers and the market think of us and who our stakeholders prefer to work with. These elements aren’t usually taken into account but fall under the umbrella of data that we will blend into the procurement process.”
One company has gone further than blending – its confidence in data is so high that shipping giant Maersk has recruited specialists in mathematics and data science to work alongside its procurement team. The intake are supporting the company in developing its own procurement-specific software and algorithm expertise. The current processes are “too manual”, says Lars Johan Andersson, procurement platform owner at Maersk, who says he wants to use data to free procurement from “bread and butter” tasks.
According to Andersson, Maersk has three goals: digitising at least 90% of supplier interactions, producing better quality data, and improving the link between procurement and the rest of the business. “When we send an RFQ to a supplier, even if they don’t win, that’s a data point we want to capture. It’s really important that we capture all data from all supplier interactions and start using it.”
A network of virtual workers
Maersk uses three tools to maximise its use of data, which it extracts from all internal operational and financial processes and combines with external data from markets and suppliers. To “harmonise” the information, and ease the burden of tasks for employees, Maersk developed several algorithms and “virtual colleague” bots called Holger, which Andersson says can manage more than 100 FTEs worth of tasks every day.
“This is completely redrawing the playing field for procurement,” he says. “Virtual colleagues such as RPA and API-based workflows can now execute a number of complex actions independently. We have more data, better solutions to analyse it and intelligent automation solutions that can execute the tasks.
“We consider each activity in the end-to-end source-to-pay process as an opportunity to produce data that we can use to improve our decision-making. Our process mining tools and algorithms oversee this activity and can trigger recommendations to our human colleagues, or even trigger automated actions by Holger. The key is to transform data into insight, and with our virtual workforce we can automatically transform insight into action. With this approach you can immediately generate value and improve decisions from data produced.”
The company’s second automation tool is called Hilda, the Human In the Loop DAshboard. When data is missing or incorrect in an automated taskflow, Hilda alerts a human worker to fill in the missing piece; once corrected, it returns to the automation flow. “We try to have humans involved only by exception. This will be implemented across all of our spend categories and this is where we want to change the procurement function from being involved in those tasks to working with better data,” says Andersson.
Emphasis on project outcomes has enabled Maersk to transform the function, says Andersson, to be more strategic, generate more value and do more advanced work. It has also helped procurement play a bigger role within the company with the third tool, the algorithm-powered Procurement Mind software.
This advanced software aims to improve the data produced by the procurement team, how it’s used by other parts of the business, and how Maersk connects with suppliers. Andersson says so far the algorithms they have developed independently manage PO creation for around half a billion dollars worth of spend, which they’re aiming to increase to 97% of spend in core spend categories (roughly US$15bn) through the tool.
“Procurement Mind will be interjected in the procure-to-pay process and capture all the data that is produced in this process. It will ensure compliance to our contracts and a structured and automated spot process. With improved data we can forecast our cost better which will improve the offering to our customers. It will also have an advisory module where all the data that is produced will be analysed by the algorithms and they will start making recommendations to the procurement colleagues on actions they can find based on analysing data patterns and trends.”
Ensure better internal relations
While these projects have technology in common, they also have a similar end goal – to regain control of complex supply networks where risk remains a constant threat and certain fields, such as ESG, are something of an information black hole. But Andersson believes onboarding tier-one suppliers to powerful data systems will help an organisation improve visibility and put procurement in the driving seat.
“Procurement needs to develop our business offering and that’s what we want to achieve with our platform,” he says. “It will link the procurement function up in a very good way to the rest of the functions in Maersk and ensure the way we work with our suppliers supports the value we create to our customers. It will also enable us to incorporate ESG data deeply into our operational processes.” Pourrat agrees and runs a similar process at BT.
“We have a dynamic risk management platform, into which you can feed any source of information,” he explains. “What we’re trying to do is create an ecosystem around that because risk is relatively open. For example, you can see the localisation and geolocalisation of our suppliers; you can check for geopolitical problems, issues with staff; you can find your tier-one suppliers and check their ESG compliance – and check the compliance of the rest of their supply chain. By feeding that kind of information into the platform we can constantly view and monitor risk and compliance.”
Procurement has the capacity to be a strategic powerhouse, to lead on external relations, ESG and the bottom line, among other facets – and data management will support this. “Our modus operandi is to share the tools and information we have,” explains Wilson. “We give access to the data we have to other parts of the business. They need to see and work with the same data we have, because it shows the scenarios that we all need to be discussing.”
According to Wilson, sharing data has made for better relations with internal stakeholders and has positioned procurement as the catalyst for projects within the organisation, a stance Andersson and Pourrat share. BT Sourced has what Pourrat calls a Rubik’s Cube of data, one that draws from numerous sources and platforms and that can be rendered in different ways to suit the scenario. With such a powerful tool, he says it makes little sense for procurement to hoard that, so they are opening the risk platform up to their customers.
“I can see my customer’s tier one and tier two, and the same for them, which is key for all of us.” He adds that his internal procurement department has done the same with its spend data, opening it up to the entire organisation. It allows the data to empower projects within BT and reflexively to enhance the way procurement does its job.
“What’s interesting for me,” Pourrat says, “is that these platforms enable everyone to do their jobs in a better way. Procurement has traditionally been guarded about its information. [With this] we’ve been able to exit our sandbox, go to other departments and give them a visibility over their customers they haven’t previously had. That’s enabling me too because I’m able to have better dialogue with my stakeholders and with my customers as well. Essentially, it’s connecting the business and coming to the business with solutions that it has been looking for.”
Algorithms trained to think like humans
In the early 2000s, a series of adverts ran on UK terrestrial television implying that businesses without a website wouldn’t be able to compete in future markets. At the time laughable, it proved to be correct. Now, the same could be said of companies that don’t fully utilise data. The explosion of AI and machine-learning has yet to reach full maturity, especially in supply chain, and this journey could take a decade. However, not embracing the inevitable is a risky strategy when there’s so much to gain.
“There’s quite a lot of runway in these new tools and new capabilities,” Wilson says, adding that Vodafone has just launched a platform that autonomously manages end-to-end sourcing without the need for human involvement. “It chooses and creates the tender, sends the tender out and evaluates it when it comes back in. It can negotiate, sign the contract, upload it to SAP or our ERP system. There’s no human intervention.”
While Vodafone doesn’t use that tool at every level of the supply chain, there’s an inevitability about how this plays out. But far from the foreboding TV ads, hectoring deniers into setting up websites, Wilson suggests there’s a land of opportunity to explore once data is properly harnessed.
“At some point, it will be our AI versus the supplier’s AI, doing the negotiation to get the best possible outcome for both parties. I would see that being ubiquitous in procurement in the next five years or so.” And this vision is shared by Andersson, who sees the same journey for the negotiation bots being piloted at Maersk.
“At the core it’s about the algorithms. If we can train them to think like category managers, or like 1,000 category managers, based on the right policies combined with good control and operational processes, we can really evolve the way we manage our categories,” he says, and adds that suppliers and smaller businesses will need to keep up. “For us, it’s been a way to generate more value for the company. I feel 100% confident this is the best journey for us. Maybe five years ago IT and technology was a hindrance but there has been a big change and today technology helps drive transformation of procurement tasks.”
“I think it’s tremendously exciting,” says Wilson. “This is a step change in the way we approach performance in procurement, in the way we approach supply chains and in how we do business. My advice is to take advantage of the technology, because opportunities like this don’t come around very often.”