Category Archives: rants

Why are We Still Hyping Big Data When We Haven’t Mastered Small Data?

The end of the year is coming, everyone is looking for next year’s tech, and everyone wants cognitive / AI that works on Big Data. But there are two real problems with this:

1. With current hard drive and memory capacities in a single machine, we typically don’t have enough data to fill it (at least with efficient encodings) in our typical back-office functions or unbearable computation times on a quad-core (with efficient algorithm implementations).

2. When it comes to learning, the biggest sets we have for training are typically quite small!

We’ll start with the second point first. Consider spend analytics, where the primary task is to map transactions to a categorization hierarchy. If you want to use a “deep learning” AI, then you need a big data set to train that AI. But how many transactions will the average organization have that have been mapped and verified individually by a human? Maybe 10K or 20K. Even a spend analysis provider will typically have only verified a few hundred thousand or maybe a couple of million compared to the tens or hundreds of millions of transactions its big customers will throw at it a year.

The situation is even worse for contract analytics. A large multi-national might have 20K contracts, but how many have been properly indexed with meta data at the clause and term level? If you have 2K you’ve struck gold. A contracts analytics provider might struggle to cobble together a data set of 20K contracts. This is even smaller data.

And then moving on to the first point. Even though most first (and even second) spend analytics applications are slow, (Opera) BIQ has been able to process and re-categorize a million transactions on a dual-core or better laptop with 8GB of memory in under a minute for almost ten years, and their most recent version on a modern quad-core laptop with 16GB of memory can handle a million transactions in a little over a second! In fact, it can handle ten million transactions in less time than it takes you to enjoy two sips of your coffee. And when you consider that most analytics are only on a set of related categories for a relatively short time period (at most 3 years), the number of transactions is typically only a few hundred thousands for a large company and in the tens of thousands for a mid-size company. That’s not only small data, but data that can, these days, even be processed in your browser (as a new analytics offering will prove next year).

So before you go goo-goo-ga-ga over big data, understand how big your data really is and get the application that works best for your data, which will more likely be at a cost point that works best for Finance as well.

How Many Platforms Do You Need? How Many Platforms Should You Need?

Sourcing is not simple. Finished goods. Made to order goods. MRO products and services. Services. Tail Spend. An average organization, if they need best of breed, might need an indirect, direct, MRO, services, and tail spend solution. Especially since most current platforms only support one such type of (strategic) sourcing project well.

Think about it. Especially in North America, the typical platform was built for indirect. Finished goods and cookie-cutter one-price services. The deep bill of materials (BoM) support required for direct sourcing or Statement of Work (SoW) for services sourcing is not present. And then MRO, which is mainly the re-ordering of parts and maintenance services to keep production lines or distribution humming. And, of course, the management of tactical tail spend as part of an overarching strategic initiative to never pay more than market for anything not worth strategically sourcing.

But an organization that ignores any of this spend is losing. Most organizations are not able to source more than 1/3rd of spend strategically a year, and if the organization only has an indirect or direct sourcing platform, that’s 1/3rd of product spend, or maybe 1/4th down to 1/5th of total spend. Even if you identify a savings of 10%, that’s only 2% to 3%, max, that can go straight to the bottom line.

Considering that, on average 30 to 40 cents of every negotiated dollar of savings does not get realized, that’s only 1% to 2% savings to the bottom line. That’s not enough. Procurement needs to be delivering 5% or more to justify it’s place as the undisputed value king of the organization. To do that, it needs to be identifying 7% to 8% savings, and to do that it needs to be sourcing not 1/5th to 1/3rd of spend, but 2/3rd to 3/4th of spend.

This means that if the solution it has only supports indirect, but it’s product spend is roughly evenly split between indirect and direct, then it also needs a direct platform. And then, depending on what the next biggest category of spend is, it will also need a platform for services spend or tail spend.

But should Sourcing need three different solutions to be the value king? That seems extreme! Especially when the sourcing process in all cases typically revolves around RFX, online bidding, optimization and analysis, award, and contract. It’s not like each type of spend uses completely different processes. The difference is just that direct uses a bill of material and extensive cost models, and services use detailed multi-line statements of work. Couldn’t one build a direct solution that could also do indirect by simply allowing the BoM to be a top line finished product? And couldn’t the grids, models, and analysis adapt to that lack of detail?

And when you think about a statement of work, isn’t that just a bill of materials for a service. Instead of raw materials, it’s individual task components. The grids are similar, it’s really just the realization of the abstraction.

So, we know how many sourcing platforms you typically need (at least three). But how many sourcing platforms should you need?

(Hint: The answer is ONE!)

If You’re Still Negotiating With the Carrot and the Stick …

… you’re not getting anyone’s attention but good ol’ Bugs. And, generally speaking, giving all the hijinks he causes, it’s best if his attentions are focussed on your competition.

So how should you be negotiating? With facts. Preferably binders of facts (but they can be in e-form on your tablet — no need to kill trees unnecessarily.)

Facts that show:

  • you know what the product should cost to make,
  • you know what margin should be healthy for the supplier,
  • you know what value-add services the supplier can offer more economically than you,
  • you know what performance metrics are reasonable, and
  • you know what the market offers are right now (and whether or not the supplier can beat them).

Suppliers don’t respond to sticks if they believe you really need them and they can get away with what they want, nor do they respond to carrot if other customers seem more enticing. Plus, they will wonder what crawl-out shelter you just climbed out of because no one from a modern organization negotiates like that anymore.

Especially if they are a typical sales organization that is all about the relationship (and talking win-win even if their definition of win-win is win for the organization, win for them at bonus time) or a more modern, Gen-X led, millennial-influenced organization that’s all about the synergy.

In the first case there will be value pitches followed by claims no one can do what they do as well and lots of smooth talk to get you off guard for when they indicate that their price (even if it has a margin that is twice industry average) is really as good as it gets and the latter will try to entice a deal from the synergy.

But regardless of organization type, every organization will respond to fact-based negotiations. With fact-based negotiations, they can’t hide fat margin behind claims of high cost, high-value, or synergy as the only way they can dispute your models, metrics, and market insight is to provide their true costs (or own research from third parties if they expect their costs to rise over the expected contract term).

And the above isn’t that hard to gather. It might take some elbow grease and a category expert, but once you’ve built the proper model and identified the proper data sources, it’s quick to update.

All you need for a fairly accurate should cost model is:

  • the bill of material break down
  • the typical energy required to produce one unit (kWh)
  • the typical labour required (labourer hours by labourer type)
  • the average industry margin

If it’s a contract manufactured product, you have this, if not, you can get an industry expert to help you craft a typical bill of materials. Your current supplier, or an industry expert, should be able to roughly estimate the typical energy overhead (based on typical production process). Similarly, your current supplier (or industry expert) should know average labour requirements against the production line.

All that’s left is understanding the acquisition cost of the materials, energy, and labour. Most raw materials are traded on exchanges, so it’s easy to get an average market cost. Most countries either have electrical utilities as state owned organizations or as highly regulated private organizations with standard prices per kWh. And most countries or labour bureaus compile average labour rates. Industry insight gives you standard margins, and you can see it’s not hard to build a reasonably accurate should cost model with expertise and elbow grease. And since the only way for a supplier to challenge it is to provide their costs, you can get even the model more accurate if their costs are actually higher. (And if they don’t challenge your model, then its relatively accurate or their costs are actually lower. In the latter case, they might get a bit more margin than you want to give in negotiations, but chances are you’ve lowered your cost as well with the model.)

This just leaves an identification of what services they can likely offer more economically, which again comes down to good modelling, and performance metrics (along with cost / profit impacts), which you should be gathering across your supply base. Then you can negotiate for better performance metrics (with penalties if they are not met) with an incumbent that isn’t doing as well as they should and wants to keep the business, or baseline metrics with a new supplier that wants the business based on current average performance across the supply base for the metric in question.

So gather your facts, and give yourself a true edge in negotiations.

American Thanksgiving May be Over …

… but there is still much more your Procurement Department could be thankful for. For example, it could be thankful for:

  • a new, better, technology platform …
  • … particularly one that fills a hole …
  • … such as analytics, optimization, SRM, and so on …
  • a services agreement with a niche consultancy …
  • … to help source key categories and transfer knowledge …
  • a services agreement with a technology expert …
  • … to help select the right platforms …
  • a GPO contract …
  • … and tail spend management help

… because, as you should have gathered by now, most Procurement platforms are behind on tech, knowledge, and support and have too much spend not under management. And it will stay that way until such time as your Procurement team gets the software and services support it needs to get to the next level.

So make sure you give your Procurement department enough financial support in their next budget so they can take their operations to the next level. Because, when they get there, you will get many times your initial investment back.

They Terk Er Jerbs! Good for them.

Because, if they were intelligent, if they weren’t already insane, they would be! One definition of insanity is doing the same thing over and over and expecting different results. But an even better one is wanting to do the same mind-numbing task over and over and over again until anyone with a modicum of intelligence would go insane.

Like screwing the same rivet 10,000 times a day. Walking up and down the same 20 aisles looking for sold out products day after day. Or performing well-defined calculations millions and millions and millions of times. This last task is something good accounts payable and procurement folk have to do over their career without AI if they want to realize the savings they should.

I say let the machines do that. And then find ways to do more intelligent actions with the results that the machines can’t do. That’s Procurement Innovation. And if you were on the ball and set up your Google Alert and noticed that the doctor was in L.A. yesterday giving a talk on Procurement, and, more specifically, Procurement Innovation. Procurement Innovation that is going to arise when you let the machines do the tactical drudge work and focus on the more strategic aspects of product acquisition. And give yourself time to get innovative … and creative … instead of just pushing virtual buttons all day. (In some P2P systems, it takes 15 clicks to actually get a product delivered when it should take 0. And how many products do you need? It’s amazing you aren’t insane! Someone should calculate the mental strength and willpower of a Procurement professional. That would be an interesting study.)

One needs to remember that AI is not I, but it is A. It is artificial, and it is extremely well suited to running lots of advanced calculations against expert defined models, well-defined variables, and big data sets to identify opportunities, outliers, and options for pursuit even the smartest of us couldn’t see because our mental calculation powers stop in the ones per second while a typical laptop’s calculation capacity is in the millions per second. Even if the best algorithms we have are, relatively speaking, dumb, the machine will outperform us in evaluating data against models and desired outcomes and identifying the best directions to pursue (which is different than being able to evaluate the perceived best options and actually pick the best ones).

And because of this, it is extremely well suited to checking invoices against POs, goods receipts, and contracts — which is one key to making sure the savings that are negotiated are actually captured. The best I2P systems today with advanced OCR can reach invoice processing accuracy (IPA) levels of 98% with no human intervention, including automatic return to supplier if issues are identified, and the proper configuration of rules can enable up to 100% of these automatically processed, corrected, and confirmed invoices to be automatically queued for payment (and paid). Considering that the average invoice “error” rate in an organization is 10% to 15% and that this typically results in overpayments of 1.5% or more, automatically processing 98% of invoices and eliminating 98% of the errors is huge.

And it’s a key component of two of the innovations — true automation and overspend prevention — that the doctor highlighted in his talk that can be addressed today, and tomorrow, and change your work, and even your life. (When you work smarter, you will get smarter.)