Automation in Retail – tightening corners, not cutting them

All the pressure of keeping up in retail is exhausting, why not reach for a robotic hand?

Retail margins are under threat across the board. And competition is increasing between existing brands with new threats – in the shape of grocery pickup and delivery start-ups who seek to disintermediate stores from the grocery purchase journey—adding to the pressure. That’s on top of more demands for higher wages, supply chain price inflation and COVID-19-driven investments to bolster both ecommerce and delivery capacity. So wringing every bit of efficiency out of day-to-day operations is becoming a major lever for achieving profitability.

Many, if not most retailers, however, have already bolted down traditional costs as tightly as they can, and are now exploring automation options to protect and improve margins.

Robotic automation—using advanced robots in place of staff to perform repetitive tasks to maximize productivity, quality, safety and efficiency, particularly within distribution centers—is garnering a lot of attention.

And there is another form of ‘robotic’ automation that is worth exploring.

Robotic Process Automation (or RPA) is the catch-all label given to a suite of business process automation technologies that emerged in the early 2000s and which use software tools (called ‘robots’) to automate tasks traditionally performed by humans.

Whereas traditional approaches to workflow automation required (often costly) custom software development to enable two systems to exchange data, RPA implementations are configured by having RPA software ‘watch’ (and record) how tasks are performed within an application’s graphical user interface (GUI).

Once the end-to-end task has been recorded, the RPA robot is able to repeat those tasks without human intervention or monitoring—only much faster and with no errors.

This “task recording” approach meant RPA robots could be configured by non-technical staff and could be applied to a much larger range of applications that haven’t otherwise provided APIs for automating data exchange.

RPA implementations are typically focused on procedural, repetitive back-office tasks, such as data entry, copying data from one paper form or software application to another, and extracting information for inclusion in reports and dashboards.

A typical RPA robot might be configured to perform the following workflow:

Workflows can be simple, like the one above, or quite complex, factoring a range of exceptions, decision tree logic and escalations. In short, if you can flowchart the various paths, steps and choice logic, then RPA robots can be configured to do the task.

As organizations started exploring more complex automation use cases, RPA software vendors have increased the ‘intelligence’ of their toolsets.

In addition to the more traditional ‘screen scraping’ approaches, many RPA tools now also incorporate simple ‘drag-and-drop wizards’ for designing workflows that take advantage of API services offered by most major enterprise platforms (such as ERP, finance, supply chain and payroll applications).

For example, where an organization is looking to automate data exchange between its finance and payroll systems, the RPA toolset would display a list of icons representing pre-defined “data objects” made available by each system’s APIs. These icons can be drag-and-dropped into a linear workflow sequence displaying the order of steps to be taken. Each step in the sequence can then have process actions or logic applied, by choosing the relevant option from drop-down menus.

This WYSIWYG (“what you see is what you get”) approach to visual workflow design places powerful tools into the hands of non-technical users, with the result that robotic process automations can be implemented in days and weeks, rather than months.

RPA solution vendors are starting to incorporate AI tools, such as computer vision, to empower robots to extract information from semi-structured and unstructured documents (such as PDFs, scanned documents and contact centre or chatbot transcripts) and machine learning, to allow robots to understand and classify documents or emails according to their contents.

As these AI capabilities mature, it opens up the possibility of designing automated workflows that include decision-making steps, such as ‘Is this an unusually high invoice for this vendor?’ or ‘Has this employee claimed an unusual number of hours?’ And they can be handled by robots, leaving humans to handle only critical exceptions.

Related: How technology-enabled innovation can help brick-and-mortar retailers compete with pure-play eCommerce

As organizations increasingly adopt ‘data driven’ strategies, this often leads to significant extra (and often unaccounted for) staff time and effort spent extracting, manipulating, compiling and publishing data across numerous systems of record.

Robotic Process Automation tools are designed to make common, well-defined and well-structured tasks more efficient, reliable and scalable.

Investment cases for these tools typically point to a range of business benefits:

• Cost savings – After the initial implementation costs, RPA can generate significant cost savings via reduced labour and staff training expenses, reduced data double-handling, reduced paperwork generation, and a reduction of costs associated with rectifying process or data errors.
• Time savings – Working 24/7 and at inhuman speeds, robots can significantly reduce the time required to process voluminous, repetitive tasks, giving decision makers access to data sooner.
• Improved accuracy and consistency – All data handling processes involve the risk of error and inconsistency – whether it’s accidental or the result of fatigue, inattention or worse. The more steps and sources of data input/extraction involved, the greater the risk that errors will be introduced. Automation eliminates these risks, ensuring processes are rigorously adhered to every time, ensuring both output and outcome consistency.
• Increased efficiency – Automation frees up staff from monotonous, recurring and low value (albeit necessary) tasks, allowing organizations to achieve more with less effort.
• Improved scalability – As organizations identify beneficial automations, these can be rolled out organization-wide, without the need to hire and train additional staff. For example, once a common distribution centre or store process is automated, and its unique logic captured (say, a change in supply process or EOD/EOM cash and credit reconciliation), it can be rapidly rolled out to all geographies with minimal cost or business impact.
• Extra security – Security and confidentiality issues associated with handling sensitive data and financial information can be minimised by using automation to limit the number of staff involved.
• Reduced employee turnover – Research suggests that as much as 90% of employees regularly miss key deadlines for time-sensitive projects and large initiatives because up to 60% of their time is spent on ‘work about work’ – replying to emailed queries, looking for files or information and chasing colleagues for input. Reducing this unnecessary productivity drain though smart automation allows staff to focus on skilled work and tasks that create higher value, which in turn can boost staff satisfaction and lower turnover.

Related: Future, interrupted: The new store of tomorrow and how retailers can adapt

3. Pre-automation.

Once the initial process(es) has been identified, the ‘heavy lifting’ begins.

Business analysts need to work closely with stakeholders to pull together details on how the process(es) works, the specific activities, steps and systems involved, known variances, alternate paths/escalations and known exceptions.

In many instances, it may require multiple iterations of forensic examination, documentation and review to land on consensus view of how (and why) a process works. In most organizations, the originators of processes are long gone, and staff often only understand their allocated step in any detail (and many cannot explain why those steps are required).

Tip: Anywhere from 50%-70% of project time and budget can be spent developing pre-automation documentation.

Once the candidate process(es) have been documented satisfactorily, it is worth investing the time to challenge every step and logic rule, using ‘zero-based thinking’ (ZBT) principles.

Tip: ZBT involves pushing back on ingrained thinking (‘we’ve always done it this way’) and adopting a clean sheet approach: If we were to design this process today, making the most of the technology-enabled and digital solutions available to us, how would we do it?

Reimagining and redesigning process using current know-how and tools ensures that existing inefficient processes are not ‘codified’ via RPA. Automating inefficient processes magnifies the inefficiencies – it doesn’t correct them.

4. Pilot.

Use a test-and-learn approach to rolling out the first RPA pilot, as this will help to identify potential defects in documentation, gaps in knowledge and other bottlenecks that may become much larger as you scale.

Pilots should have clearly defined measures of success, defined and agreed prior to launch, to keep the project team honest.

5. Outcome measurement.

Outcome data should be captured while the pilot is in-flight, with the results fully documented, including lessons learned. The final report should present a clear-eyed view of costs as well as analyse the gaps between the predicted performance and the actual performance.

It may be necessary to iterate the pilot design until positive outcomes are achieved, and the pilot is ready for wide-scale deployment.

Before the pilot is cleared for further deployment, a change management plan should be prepared, outlining the steps required to implement, manage and sustain the positive change the automation will create.

6. Expansion.

The expansion stage sees the next priority process enter the ‘Pre-automation’ step and the ‘Pilot’ step repeated, while the just completed pilot program is rolled out across the business.

RPA projects are never ‘one and done.’ Both back-office and front-of-store technologies—and all the applications in-between—are changing constantly, which means robotic process automation should be added to every organization’s continuous improvement plan, as well as their arsenal of productivity improvement tools.