On Using OCR for Invoice Management, Horses, and Buggies: A Counterargument from Pierre
In last Friday’s guest post from Coupa, we heard an argument exploring accounts payable automation and why, in the words of the contributor, OCR (optical character recognition) is really like finding a better buggy to strap to your horse rather than just getting a car (i.e., integrated P2P and e-invoicing).
First off, I’d like to thank Coupa for the guest post and sharing their point of view. I certainly get the premise of it. OCR is obviously not the ultimate technology to enable the end-state of P2P nirvana where such horses don’t exist. However, until that time, OCR is a key technology to help companies transition from horses to cars. It just needs to be deployed in a new way.
OCR is merely like the magic wand in Cinderella, except instead of turning the mouse into a horse, it makes the horse (supplier invoice document) look like a car (supplier invoice data). As the world moves from horses to cars, P2P providers can help accelerate the modernization rather than complain about all the horses and tell the horse owners to buy cars. Sorry, but in P2P, and with smaller suppliers, we’re in Amish territory. “Let them buy cars” makes me think of “Let them eat cake!”
The attitude contained in those famous words, whether Marie Antoinette spoke them or not, did not bode well for her.
So, rather than complain about the older manifestations of OCR, why not improve it? An innovative supplier network should accommodate the small suppliers with horses and allow them to do self-service training of an embedded OCR capability as part of an intelligent supplier network. Thus, the supplier on-boarding process lets the supplier use its existing horses until it upgrades to a car. The supplier can optimize its existing processes, keep its costs low, and reduce on-off process variations that may introduce error.
In fact, the supplier might even be so thrilled with this seamless level of integration into their horse farm that they jump at the chance to recommend and introduce Coupa as the magic wand service to its other customers. Both sides benefit. This is the network effect.
P2P is all about supplier adoption that includes spend adoption and transactional adoption, and an OCR capability is merely just a tool in the toolbox. However, it's an important tool. That said, this is not necessarily good news for traditional OCR vendors who deploy single enterprise versions (e.g., on premise versions) of the product where each instance has to re-learn a supplier format.
Smarter supplier networks will accommodate current reality and objectives while also supporting future objectives. Supplier networks must get smarter, and they are getting smarter. The key here is that the “smarts” get built on a one-to-many basis, not a one-to-one basis. We’re seeing an increasing trend towards supplier networks using self-service workflow and rules-based, or – preferably – inference-engine based learning to let trading partners train the network for their use (i.e., to train the network on how to properly OCR their invoice document – as a re-usable network service).
Currently, Nipendo and Tradeshift both offer this type of supplier self-service training for the network, and the good news for Coupa actually is that if they can similarly embed this capability, their sweet spot of intuitive functionality for P2P will be made “spot on.”
My feeling is that it’s only a matter of time before others get this as well, philosophically and commercially. The need is real. But whatever happens, please, don’t put the cart before the …