How Procurement Can Deploy Bots Effectively
Bots are gaining adoption for front end business process such as customer service. Finance organizations are also looking at bots as a way to automate repetitive, rules-based tasks, and they could serve the same purpose in procurement organizations. The name “bots” is somewhat deceiving, as what we are really talking about for the enterprise are not mechanical robots, but snippets of code that are relatively easy to deploy. However, in order to use them effectively, you have to develop a strategy for how and they will share the workload with humans, how they will impact policies and procedures, and how you’ll monitor their activity on an ongoing basis. In this article, guest columnist Kristin Ruehle of Accenture shares some best practices.
Companies everywhere are getting more excited about bots, and it’s easy to see why.
Bots can automate repetitive, routine tasks to reduce costs, improve efficiency and accuracy, and accelerate execution. In the process, they free up humans for other higher-value activities that require judgment and intuition. They also easily scale: Bots can handle a far greater workload than a human employee and never get sick or take breaks.
And they’re highly versatile. Bots can be applied anywhere that involves rules-based processing, standard data input and output, minimal exceptions, and high volumes. This makes them especially suitable for back-office functions such as finance and accounting, human resources, and customer service/help desk. But even some front-office activities such as order management and customer management are also good candidates.
To date, bots have yet to penetrate the procurement organization to a large degree—they’ve been limited mostly to desktop deployments for discrete process steps. But that will change in the near future, as enterprise bots—which are scalable to accommodate all individuals involved in a specific task—are tapped to become a key procurement workhorse. They will increasingly assume responsibility for any procurement activity or process that involves a high degree of both structured information (such as a supplier name, category, commodity code, discrete item description, or SKU number) and rules-based processing (“if X, then do Y”). A simple example is using a bot to automatically, with no human intervention, create a purchase order based on requisition data when all required fields are complete and accurate.
In fact, bots—paired with analytics and artificial intelligence—can be critical to the future procurement organization that relies more on data, and less on processes, to make better business decisions, potentially reduce operating costs, and improve overall efficiency.
Three Operational Considerations
But there’s more to benefiting from bots than simply inserting them into a process or activity. To deploy bots effectively, you first need to develop an overall Intelligent Automation Agenda to ensure you’ve thoroughly determined how you’ll use bots across the enterprise, and to what purposes, rather than pursue ad hoc, one-off applications. But as important is taking the next step: Thinking through what it takes to make bots work at the operations level and identifying upfront the things to watch out for. For procurement organizations, three such considerations are especially important.
Bots can be used in two basic ways: assisted, in which a person must be involved in the activity in some way; and unassisted, in which the bot autonomously executes an activity with no human intervention. You should consider carefully the type of activity involved to determine which makes sense.
Take, for instance, editing vendor master data. Today, a person in procurement will receive a request from a vendor to make some change to its master data—say, the address or bank account details. That person first must review and validate that the documentation supporting the request is legitimate—it’s accurate, on the company letterhead, and signed by right person. Then, he or she goes into the master record and keys in the new information. It’s a very manual, time-intensive activity that’s ripe for automation because the data fields don’t change.
So the question is, could you enlist an unassisted bot to perform all the data entry activity once a person validates the request? Or would the bot only be allowed to autonomously change certain less-sensitive fields—such as email address—while a person would be required to change other data, such as bank account details? Perhaps an assisted bot could even do a first pass at validating the initial request, enlist a person to review it and give the final approval, then take over again to handle the data entry. In this case, human intervention may be needed because the company requires an audit trail that indicates the appropriate person reviewed and signed off on the request.
Most companies have a set of internally defined policies and procedures designed to provide a level of control over procurement activities. For instance, only a vice president can approve a purchase over $100,000, or the procurement organization must get involved for purchases above $250,000, or all non-PO invoices have to be reviewed and approved before being paid. Adding a bot to an activity will require a change in the policies and procedures to which that activity is subject.
Think back to the activity just mentioned, validating a vendor’s request to change its master account data before making the change. If a policy’s in place requiring such validation, you’ll need to amend it to allow for the use of an unassisted bot in validation, and call out the specific conditions for the bot’s use (i.e., which activities the bot can and can’t handle).
Deploying a bot without first making relevant changes in policies and procedures can result in a variety of challenges. Audit controls could be compromised, introducing new risks to the organization. More effort could be required to double check bots’ work, which would eliminate the productivity improvements that were bots’ biggest selling point. And you could miss out on opportunities to effectively use bots to improve processes.
One of bots’ big advantages is the ease with which resources in the business can establish them. Because bots simply mimic existing human actions—e.g., keystrokes on a computer—they can be deployed quickly with little to no change in the activity or process itself. But that comes with a potential downside: Because bots are so easy to implement, a business team—such as procurement—can deploy them without fully thinking through how the bots should be monitored and who should be doing it. The team may think the IT organization is keeping tabs on them, but if the bot was a business-led initiative, the bot may not even be on IT’s radar. And that can spell trouble.
Take, for example, a bot that extracts certain information from an ERP system and transfers it to another tool for use by the business. What happens if the ERP system undergoes a major upgrade? Does the tool still work as designed after the new version is live? Without a “bot doctor” performing regular health checks, a company will never know. The “bot doctor” needs to confirm the bot’s operating at the same level of efficiency as before the change, as well as determine if certain transactions or requests are being missed because the data requirements are no longer being met. Failure to monitor bots can have a big impact on the overall organization. FTEs who were released to focus on other activities must now rededicate their time to handling the manual activities again. Worse, the company doesn’t recognize the bots are no longer working effectively and are compromising outcomes.
The key to monitoring is twofold: first, establishing accountability for monitoring; and second, actually doing it. In our experience, IT should be responsible for the regular monitoring of bots just as they are for the company’s ERP and other systems. And IT should engage the business owner to investigate and resolve any issues identified. Monitoring should be done daily to ensure the bots are running, and a deeper performance assessment should be completed monthly and during any system changes.
- Human/Bot Division of Labor
- Policy Impact
- Ongoing Monitoring
More Bots, More Considerations
Bots are nothing new—they’ve been executing tasks on desktops for years, and have become a core tool companies use to handle interactions with customers, answer FAQs, and process payments. They’ve even begun to make their way into forward-thinking procurement organizations to handle such things as vendor and catalog management, end user support, purchase order creation, invoicing, and requisition and payment processing.
But this is only the tip of the iceberg. In the next several years, bots will become even more pervasive as organizations increasingly look to deploy them across the enterprise. That makes it even more critical to think through the operational details and impacts well before the bots hit the ground.
Kristin Ruehle is Procurement Business Process Services Offering Lead at Accenture, a Coupa partner. This article previously appeared on the Accenture Intelligent Operations blog.