As almost everyone can agree, workflow automation can deliver organizations extreme business value with reduced operational costs, improved client services and competitive advantages. So, with the promise of such great efficiencies, especially with amazing innovation such as cloud services, artificial intelligence, robotic process automation and others, why isn’t every organization implementing workflow automation solutions? The answer is very simple: overcomplication.
Don’t get overcomplicated with your initial digitization projects. Often, projects try to solve the entire problem all in one huge chunk rather than small, tasty nuggets. Resist the urge to do too many different things all at once.
My advice is to start by documenting your existing business processes that involve any sort of manual or human intervention. Then consider how you can digitize each one of these processes from the very beginning of a transaction.
Ask yourself, “Is this a process that can be born digital?” Next, if the work process can utilize some sort of capture device, determine which type of device is most appropriate for capturing the information.
If you are capturing documents, then is it best to use a smartphone camera, a digital copy machine or a single-purpose document scanner? If you are capturing video, is a smartphone sufficient or does it need a fixed camera? Or if you are capturing voice, will a home voice assistant-type device work, does it need to be an embedded microphone, or is the best approach using headphones like many of us use on our video conference calls?
The answers will vary based on the specific workflow, so the only point is that if you start by digitizing everything you can start your workflow process.
Index everything (that you can)
I’ll admit that indexing your digital content is often the most difficult part of an effective workflow automation strategy, but it’s critically important to the rest of your downstream workflow process.
Indexing, or creating metadata, is the crucial point where your workflow system is really going to be smart by delivering intelligent content into your downstream process, or, as the old saying goes, “garbage-in, garbage-out.” Your downstream workflow process can suffer immensely from lack of metadata, having to correct too many errors in recognition or, the worst of all, wrong metadata.
You often only get one chance to capture a bit of information and digitize it; therefore I always recommend collecting as much metadata you can about whatever information you are capturing. You never know when you might want to use data such as GPS location, timestamp of the transaction, or who created a piece of digital content. You just never know when some piece of metadata you didn’t think was important for your process might be very useful to use in some analytics AI application where you can find cures for cancer, find a pattern to trade stocks more efficiently, or unlock other valuable business insights stored deep in your corporate data silos.
It’s often not the technology itself that’s the difficult part but rather the taxonomy, or logical organization, of your content that must be a decision before implementing a technology. In other words, using an accounts payable workflow as an example, a decision must be made on what indexing information is required from each document type – such as an invoice, purchase order or bill of lading. Only then, after a decision has been made, should the technology be applied to automating this process.
When it comes to your indexing strategy for scanned documents, do yourself and your colleagues a big favor and at the very least use full-text OCR to index your content on a per-document basis. This means that instead of storing the OCR index information into a database, the index (aka metadata) information is stored as a hidden text layer that is part of the PDF file itself.
You never know when this hidden text layer of index/metadata will prove useful by being able to import it into a centralized search database at some point in the future. Believe me, you will have done yourself a great service by applying full-text OCR as a logical first step to your overall indexing strategy, especially when using modern innovation such as artificial intelligence and machine learning for indexing. These technologies can take this full-text OCR, and through an algorithm be able to perform even more advanced indexing, such as unstructured document processing or even semantic indexing.
Think globally, act locally
While digitizing your content and indexing the metadata within your business processes is a highly recommended strategy for efficient workflow automation, this does not mean you should get overly ambitious and do things recklessly. It is better to take a measured approach and implement change in small increments.
Do not think that all this digitalization needs to be neatly connected with a perfect workflow process on day one. It is much better to not try and boil the ocean; rather, take a small process and make it better. Start with one department and fix a particular pain. After fixing this pain, then create another “win” with interdepartmental collaboration.Far too often when considering workflow automation solutions, we get caught up with hitting a grand slam instead of something like a solid double that fills the gap, to use a baseball analogy.Click To Tweet
Model workflows before and after scenarios to prove clear ROI
Far too often when considering workflow automation solutions, we get caught up with hitting a grand slam instead of something like a solid double that fills the gaps, to use a baseball analogy. One way to do that in your business processes is to model before and after scenarios. There are specialized software tools for business process management (BPM) modeling, or you can use something as simple as an Excel spreadsheet.
One of the best practices when planning to deploy new types of innovative workflow automation is to observe human behavior and be honest in the assessment of the time spent doing finite tasks. Carefully document every individual step it takes a human to complete a task from a time-consumption standpoint. For example, in a document scanning workflow use case scenario, you would document in the number of minutes each small task takes, such as removing staples, time spent on manual key entry and moving paper from one person’s desk to another person’s inbox.
Only after collecting all the data that you can about the current workflow process can you start modeling a before-and-after scenario. Resist the temptation to offer suggestions or improve the process in your data collection process, because the intention of this work is to establish a baseline of a “current time consumed” calculation. Establishing a baseline of existing business processes, upon which everyone can agree, is an extremely valuable (often required) part of the process in order for you to justify the expense of purchasing and implementing new types of innovative workflow automation technologies and solutions.
Remove points of failure with a centralized workflow process
Just as I suggested “digitize everything” previously, the same applies to “centralize everything.” Once again, my advice is to keep things simple when it comes to a centralized approach for your workflow processes.
As it relates to the document scanning process, a major point of failure, believe it or not, is the local scanner driver itself. How many of us have gone to use a USB-attached device such as a printer, a webcam or another external peripheral and been frustrated when, for whatever reason, the USB device wasn’t recognized? Then after unplugging and reattaching the USB cable five times and something like three computer reboots, the device comes back online. It’s usually not a bad design on behalf of the scanner manufacturer or a fatal flaw that cannot be recovered; it’s just the way that operating systems work, and sometimes there are unexplained errors.
With modern scanner driver innovation, using a web services scanner driver architecture with PC-less document scanning is preferred to remove points of failure in a workflow process. Why? Because with a web service scanner driver so many traditional technical maintenance problems are eliminated, such as operating system compatibility, operating system updates or USB conflicts. Additionally, web services mean that one server can host many scanners instead of a one-to-one relationship between one USB scanner and one attached computer.
Measure and monitor everything
In summary, an effective workflow automation system is a never-ending process of iterating for improvement. Therefore, it’s really important to collect, monitor and measure as much data and metrics as possible so that you can easily detect and improve your processes.
Make certain to have very granular key performance metrics that measure items such as overall throughput, or volume, of the workstation or operator. In other words, how many documents are scanned or how many calls are made per person in the process?
However, be careful not to fall into the trap of judging the efficiency of your workflow automation system simply by the assumption that a higher volume means better work. Sometimes a high volume does not indicate good work, so also measure the quality of clean data coming through in your workflow process. Basically, make sure that errors in the workflow process aren’t causing you to have to go back and fix things later, because that can be very complicated, costly and expensive.
While all this might sound overwhelming, my best advice is to keep things simple and remove complications. Just remember to act locally by modeling before-and-after workflow automation scenarios, and to digitize, index and measure everything in a centralized technology architecture.
If you apply this type of thinking when implementing new innovation to your workflow automation, then you can increment your way to success.