Daily Archives: July 29, 2019

How to Get Past Applied Indirection

As per our recent series here on SI, when most vendor sales rep start to claim they have AI, they are really just telling you to your face that they are trying to mislead you into thinking their trivial automation, simple fixed ruled-based workflow, and/or classic statistical projection capabilities are much more advanced than they really are, hoping you won’t ask what AI really stands for when they use the acronym.

Given that your number one priority is to get more spend under management (SUM) and that this priority is only realized with the help of modern platforms, you’re going to be dealing with a lot of sales reps for years to come, especially since, at best, you’re on a generation 2 platform (and, to be honest, if you have anything, odds are it’s really generation 1), and that just doesn’t cut it anymore. So you’re going to have to find the right platform for you.

Now, the good news is that you have help narrowing down that shortlist with the help of Spend Matters Solution Map, co-designed and, in core areas of platform technology and Strategic Procurement Technology, scored by the doctor, and that as part of this narrowing down, we can help you identify vendors with the foundations for real AI, as well as, if we’re lucky, select capabilities that fall in the domain of assisted intelligence.

But just because we can give you a partially pre-qualified short-list (which can be tailored to your specific organizational needs by way of the Customer Map offering), that doesn’t mean that the vendor sale reps still won’t try to stretch the truth or, in some cases, even lead you astray on aspects of the solution we don’t score. So you will still have to deal with some level of applied indirection even if you’re proactive enough to take our advice and start with the right short-list. (Which can also be based on unbiased customer scores as well as in-depth analyst scores across up to 700 discrete platform capabilities to make sure you start off with the best candidates, among which will be the right solution for your organization.)

But if you’re not one of the lucky supply managers able to convince your boss to let you spend the money on this exercise (which can be carried out by your favorite consulting partner who will help you properly weight the various capabilities given your organizational maturity and need), then you’re going to not only get hit with quite a few sales reps stretching the truth, but a few outright lying (because they know you don’t have any validated data points to go off of), and not feeling a tinge of guilt because they told you up front they were selling you with AI (which you didn’t ask them to define) that really stood for applied indirection, and not the assisted, augmented, or artificial intelligence that you mistakenly assumed it stood for.

So how do you spot it? And get past it?

Here are some tips and tricks to do just that.

1. Ignore their claims, get a demo and ask them to walk through through how it supports your organizational process, which you will lay out the day before

Yes, some vendors have become quite good at combining (robotic process) automation, rules-based workflows, and statistical algorithms to fake AI, to the point that you might think there is actually some machine learning under the hood and, at the very least, they have assisted intelligence technology in the worst case, and probably augmented intelligence that will take your team to the next level. But not very many vendors fall here (and in the grand scheme of things, the reality is that very few vendors fall here), and very few demo masters can pull off a faked end-to-end process demonstration.

2. Have your own data files ready to go!

If they are claiming auto-contract parsing and clause extraction, have some contracts in the correct format (PDF, Word, etc.) ready to go at demo time, that you did NOT give the vendor advanced knowledge about, and ask them to upload and walk you through the process live. Or if it’s a 3-way invoice match process, have matching POs, goods receipt, and invoices in whatever standard they support (cXML, EDI, indexed PDF, etc.) ready to go as well and ask them to suck them in and process them in front of your eyes.

If they survive this, even if it’s not real AI, it’s very advanced automation and an extensive knowledge-base supporting the rules-based workflow, which may be all your organization needs to advance its SUM and get success.  (For example, you don’t need AI for spend categorization – an expert can map your spend to 98% accuracy in 3 days with the right tool even if you are an F500 and then as exceptions come in, you have an expert create overrides, which get fewer and further between over time. Plus, unless we are far, far into the tail, 2% of spend in the category doesn’t even make a dent.)

3. Get a real data scientist / tech expert in on the demos.

Someone who has utilized real AI technology to ask tough questions about algorithms, platform foundations, data stores, and so on. If the provider can’t furnish good answers, there’s probably not too much under the hood.

4. Talk to mature customers.

You want customers who have been with the provider 3+ years, implemented and worked through the full platform offering, executed difficult Sourcing / Procurement projects, had a few failures the provider needed to respond to quickly, and so on. They can give you an idea of how advanced the system has been in practice and how good the provider has been on improving it. And if they give you a good recommendation, even if the system is not as advanced as the vendor claims, there’s probably something there.

It’s easy to not get fooled if you remember that the proof is in the pudding, and if the pudding is good, there are repeat, happy eaters of it.