Category Archives: Sourcing Innovation

Features ARE NOT Applications; But Applications Require Features!

THE PROPHET recently asked What Procurement Tech Product Categories Were Really Just Features All Along? Which is a great question, except he cheated.

He cheated with the first 5!

  • Supplier performance management
  • Supplier quality management
  • Supplier information management / supplier master data management
  • Supplier diversity
  • Supplier risk management (not supply chain risk!)

We’ve known for years it should be one Supplier 360 solution! (Even though no one offers that when you consider all of the elements that should be there. Heck, none of them even offer the 10 basic CORNED QUIP requirements … in fact, good luck finding a solution that offers 5 of those requirements among the 100+ supplier management solutions).

He you cheated again with the next 3!

  • Should cost / cost modeling (for procurement, not design engineers)
  • RFX and reverse auctions (when not bundled with broader capabilities or services)
  • Sourcing optimization

We’ve also known for yours it should be cost-model and optimization backed sourcing (auction, RFX, hybrid, single source negotiation, etc.) … otherwise, it’s an incomplete solution. But only a fraction of the 80+ sourcing platforms offer true optimization (less than 10) and fewer still do extensive cost modelling. (Note that we are focussed on modelling, not cost estimation — that requires data, and that can, and probably should, be a third party data feed.)

And he was wrong on the last front.

Real Spend Analytics should be standalone. Wrapping restricts it! The modules you use should provide all the specific views you need, but the reason that spend analysis quickly becomes shelfware in most organizations today is the same reason it became shelfware 20 years ago … once you exhaust the limits of the interface its wrapped in, it becomes useless. Go back to the series Eric and I wrote 18 years ago (which you can since Sourcing Innovation didn’t delete everything more than a decade old when it had to change servers in 2024, unlike Spend Matters when it did its site upgrade in 2023).

But Very, Very right in that features are not applications!

And very, very right in that too many start-ups are launching today as features (which will only survive if acquired and rolled up into existing applications and platforms), and not solutions. While apps dominate the consumer world, in business there is not always an app for that, and, frankly, there shouldn’t be. This focus on point-based apps is ridiculous. It’s not features, it’s functions. It’s not apps, it’s platforms. It’s not orchestration (and definitely not spend orchestration), it’s ecosystems!

Recent stats, such as those published by Spendesk put the average number of apps a business uses at 371, with an average of 253 for SMBs and 473 for enterprise firms. WHAT. THE. F6CK? This is insane. How many departments does an average organization have? Less than 10. How many key functional areas? Less than 12. Often less than 10! How many core tasks in each function? Usually less than 6. That says, in the worst case, an enterprise might have 72 distinct critical tasks which might need their own application (but probably not). This says that SMBs have at least 3 times the app they should have, mid-size organizations at least 5 times, and enterprises at least 7 times. That is insane! No wonder there are so many carbon copy SaaS optimizers (as we covered in our piece on sacred cows), because if you have that many SaaS apps, you have features, not applications. And you need to replace sets of these with functional applications that solve your core problems.

(And if you want to know how to prevent app sprawl, before buying yet-another-app, ask yourself “is this supporting a function that should be done on its own, or just a task that should be part of an existing function” … if the latter, it’s a feature, not an application, and if the application it should be part of does not have an upgrade/module that supports the task, then you have the wrong application and it’s time to replace it, not pointlessly extend the ecosystem!)

Follow the Money to Find Future Opportunity — Which Will NOT Be Fully Found With Autonomous Sourcing!

Spend Matters has thrown caution to the wind and followed Gartner’s lead jumping onto the AI Hype Bus (with no steering and no brakes) that is still heading straight for the cliff and are wheeling out webinars on AI faster than a prairie fire with a tailwind. (Needless to say Sourcing Innovation does not think this is a good thing. There are valid uses for AI and automated processing, but fully handing over financial decisions is like wheeling in the Trojan Horse and leaving it unguarded in the server room with unrestricted access to your bank integration.)

Recently, The Maverick advertised yet another Spend Matters webinar on Autonomous and AI Sourcing where he said we should “follow the money”. Which we should, but there are a few things we need to clarify first.

1. No Money Changes Hands In Sourcing

It changes hands in Procurement … and it’s because most companies don’t follow the money after the contract is signed that 30 to 40 cents of negotiated savings never materialize in many companies, which The Maverick should remember from his AMR and Hackett days, as it was laid clear in Mickey North Rizza‘s famous 2009 “Reaching Sourcing Excellence” series, which we know is in his archives.

2. “Speed” is NOT a strategic edge if you don’t get it right!

If you don’t go out with the right strategy, don’t know the current market price, don’t know the reason for the current market price, and don’t have the knowledge to project if the trend is going to continue, stabilize or reverse, going to market is not a good decision … and it’s an even worse decision to automate the sourcing project and secure an award as fast as possible if you don’t know if it’s the best you could have done or the worst you could have done.

3. “Pecunia non olet”, but yet these vendors are asking you to treat it like it does!

They want you to automate spend analysis, sourcing, contracts, purchases, and everything else that involves money by turning over everything to their Agentric AI because, apparently, money stinks and you don’t want to touch it. (But they are quite happy to not only spend yours for you but takes as much of it as they can for their services.)

But here’s what they don’t tell you.

  • AI is NOT Intelligent.
    The level of intelligence in their “AI” is equivalent to the level of intelligence in a carpenter’s hammer. The level of effectiveness is entirely dependent on how skilled the person “training” the system and how skilled the person “using” the system is, just like the effectiveness of a hammer is dependent on how well the carpenter was trained and how experienced he is in it’s use.
  • AI Does Not Know What it Does Not Know.
    If the data is incomplete, the recommendation is very likely incorrect.
  • AI Cannot Do Better than the Best A Human Has Ever Done in Decision Making.
    So, if none of the situations it was trained on led to great results, neither will what it recommends for you.

You need to remember how Gen-AI does its work (or should we say does not work). It is large document search and summarization and chain of compute. Now, the more advanced players are trying to embed knowledge graphs into this, but these are not perfect either. With good training examples, and a very similar situation, the probability it will work well is very good, but it’s still only a probability. As a result, nothing should ever be fully automated where money is concerned. The tools should be used for their recommendations, and if the recommendations are good, and the risk is low, most of the tactical data processing and event management should be automated, but the decisions should ALWAYS be made by a human, who should be involved at every decision point. Even if that decision is verifying the system recommendation. It only takes one miscalculation due to an incomplete data source to project a wrong trend, rush an auction, lock in a price 3X what you are paying now, only for it to fall in a month later when a factory (which went offline temporarily due to a manmade or natural disaster) comes back online and the supply-demand balance returns to normal. And while you may have stocked out for two weeks, those losses will be orders of magnitude less than paying 3X at a contract you have to honour (unless you want to get dragged into court).

Now, if you really want to make money, forget all this Autonomous and Agentric AI BS, look for Augmented Intelligence solutions that make your staff two, three, five, and even ten times more efficient, purchase those, and, remembering that the US infrastructure is crumbling fast (and not going to get renewed under a Republican administration that is more interest in trickle-on economic tax cuts for its billionaires than ensuring you have running water), it’s time to remember how the smart made money in ancient Rome — public bathhouses and latrines. Time to invest in your own desalination facilities and be ready when the public wells run dry. After all, “Pecunia Non Olet“.

With Great Data Comes Great Opportunity!

In fact, it can quadruple your ROI from a major suite.

Not long ago, Stephany Lapierre posted that your team may only be realizing <50% of the ROI from your Ariba or Coupa investment, to which, of course, my response was:

50% of value on average? WOW!

Let’s break some things down.

A suite will typically cost 4X a leaner mid-market offering which is often enough even for an enterprise just starting it’s Best in Class journey (that will take at least 8 years, as per Hackett group research in the 2000s).

Moreover, even if the enterprise can make full use of the suite it buys for 4X, at least 80% of the “opportunity” comes from just having a good process, technology, baseline capability and automation behind it. That says you’re paying 4X to squeeze an additional 20% worth of opportunity in the best case.

On average, it takes 2 to 3 years to implement a suite (on a 3 to 5 year deal). So maybe you’re seeing an average of 66% functionality over the contract duration.

As Stephany pointed out, bad data leads to

  • increased supplier discovery and management times
  • invoice processing delays and errors
  • increased risk and decreased performance insight

As well as an

  • inability to take advantage of advanced (spend) analytics
  • inability to build detailed optimization models
  • decreased accuracy in cost modelling and market prediction

This is even more problematic! Why? These are the only technologies found to deliver year-over-year 10%+ savings! (This is where the extra value a suite can offer comes from, but only with good data. Otherwise, at most half of the opportunity will be realized.)

Thus, one can argue an average organization is only getting 66% of 25% of 80% of its investment against peers (based on 2/3rd functionality, the 4X suite cost, and the baseline savings available from a basic mid-market application that instills good process and cost intelligence) and 50% of 20% (as it is able to take advantage of at most half of the advanced functionality offered by the suite due to poor and incomplete data). In other words, at the end of the day, we’d argue an average company is only realizing 23% of the potential value from an opportunity perspective!

However, as one should rightly point out, the true value of a suite is not the value you get on the base, it’s the ROI on that extra spend that allows for 20% more opportunity than a customer can get from lesser peer ProcureTech solutions.

For example, let’s say you are a company with 1B of spend with a 100M opportunity.

If tackling 20M of that opportunity requires advanced analytics, optimization, and extensive end-to-end data, it’s likely that you’ll never see that with an average mid-market solution with limited analytics, no optimization, and only baseline transactional data. If the company paid an extra 1.5M over 3 years for this enhanced functionality, then the ROI on that is 13X, which is definitely worth it.

Moreover, if the suite supports the creation of enhanced automations, you could get more throughput per employee and realize the base 80M with half or one quarter of the workforce, which would lead to a lowering of the HR budget that more than covers the baseline cost.

However, ALL of this requires great data, advanced capability, and the in-house knowledge to use both. This is only the case in the market leaders. As a result, we’d argue that the majority of clients are only realizing about 25% of the suite’s potential — when sometimes the only thing standing in their way of realizing the rest is good data.

Why Your Standard Sourcing Solution Doesn’t Work For Direct

Too many of you have been there. You sign that seven figure deal for that end-to-end Source-to-Pay suite, spend another seven figures and 18 months integrating with the ERP, PLM, AP, BI and existing Legal CR solutions, and then try to source your first NPI project natively only to … fail. Why is that?

They just weren’t built for direct.

And it’s not just something you can add in later. If the platform wasn’t designed from the ground up for direct sourcing, there’s zero chance it will ever do a decent job at it. (And, FYI, the majority of the S2P suites the big analyst firms are drooling over in their annual quadrants and waves started out as simple indirect Sourcing or Procurement tools.) People who don’t understand the nature of software don’t get this, but software has to be constructed like a building. You might hear vendors and techies throw around MVC model, which stands for Model-View-Controller, when they talk about how new and well architected their solution is, but that just means it’s built in a maintainable web-friendly way for what, and only what, it was initially designed to do.

It all comes down to the data model and the software architecture of the controller, and neither can be a black box. The data model has to be designed from the ground up to support bill of materials and direct sourcing and procurement data requirements. The controller has to provide the infrastructure to support the complexity of the application that is required. For those who don’t understand software, I like to put it this way. If you pour the foundation for a two story house, and buy wooden beams for all of your structure and supports, you can’t build a 10 story apartment building. You need a foundation for an apartment building and steel and concrete supports. (Even though you can theoretically build a 10-story structure on a two-story foundation if you have the right steel and supports, it won’t be stable. The slightest tremor on the Richter scale [which might not even be detectable by a human] or a strong wind will send it crashing down.) You need both. And just like you can’t replace the foundation under a building or replace the entire support structure in real life, you can’t do the same in code. You have to rebuild, usually from scratch.

So why weren’t they built for direct? Well, there are a number of reasons (besides they wanted to get a product to market fast and/or just weren’t smart enough to build a direct sourcing solution). They include:

  1. direct material sourcing is hard
  2. substitution is not guaranteed
  3. demand aggregation is not straight forward
  4. delivery time guarantees and on-time arrival is significantly more important

To understand these, and learn about the rest of the reasons the majority of sourcing solutions were not built for direct, dive into Standard Sourcing Technology Solutions Don’t Work for Direct – Part One and Standard Sourcing Technology Solutions Don’t Work for Direct – Part 2 over on Supply Chain Matters.

Do You Have Continuous Cost Control?

If not, you should, because with tariffs rising, markets falling, inflation out of control, sales dropping (as entire markets are cut off with sanctions and trade wars), we’ve gone beyond the point where every dollar counts to the point where every penny counts on every purchase because those pennies add up as every 100 purchases is a dollar and every 100,000 purchases is $1,000 and when money is as tight as it is now, that is actually value (especially for an organization making millions of purchases a year).

And right now, organizations are wasting a lot of dollars through the entire purchasing process. From poor sourcing strategy and process, to poor sourcing and negotiation, through poor purchasing execution, and poorer logistics management, to poor invoice and payment management. Every step without good cost control adds cost to the process, at a time when you need to be taking cost out just to survive.

And we know organizations are losing across the board because the following is required to keep costs in control:

  • good processes at each step
  • (near) real time market intelligence at each step
  • good systems supporting each step
  • continuous monitoring at each stage

And we’ve never seen an organization, even a best-in-class organization, that has all of this for their Procurement department. In fact, it’s rare to find an organization that has more than half of this. It’s now at the point where your organization may not survive if it does not have:

  • well defined processes for
    • supplier discovery and management
    • sourcing
    • contract award and management
    • procurement, on-and-off contract
    • invoice management and accounts payable
    • logistics and warehousing
    • ongoing analysis
  • (near) real-time market intelligence at each step
    • current, financially stable, accessible suppliers
    • current commodity costs, average overhead costs by region, tariffs, etc.
    • current best practices, standard clauses, and insurable risks
    • market availability, quality, delivery times, remaining contractual commitments
    • current entity information, payment terms, standard processing times, community intelligence on supplier OTD
    • carrier availability, costs, surcharges, etc.
    • changes in spend trends and curves, etc.
  • good systems/modules supporting each step
    • supplier 360 module (not just SIM/SRM/SPM .. all supplier data and interactions)
    • sourcing (RFX) management
    • contract negotiation tracking, signing, and ongoing management
    • e-Procurement that supports ALL purchases through the system
    • I2P with automated invoice processing and workflows
      (85% should be touchless on implementation, 95%+ over time)
    • logistics booking and carrier monitoring
    • best in class spend and performance analysis that updates at least daily
      (and regularly re-runs best-in-class trend and outlier analysis and alerts you to unexpected changes)
  • … with built-in alerting when something unexpected happens or doesn’t happen on schedule / as expected

And you don’t. But you need this now more than ever. So, if you don’t have:

  • processes, define them; they can be basic to start; for example, classic 7-step sourcing is enough to start (even though there are some more refined 11 step processes)
  • market intelligence, get yourself some; in particular, supplier discovery as some of your suppliers will go out of business, be unreachable, or get too expensive in the days to come; cost modelling for major spend categories to understand true costs for better negotiations because even if it only shaves half a percentage point on average, that’s still 500K on a 1M category (and you can get some of these solutions for under 100K a year), and those hundreds of thousands quickly add up to millions; and major news/event monitoring to pinpoint emerging risks as fast as possible
  • modules supporting the entire S2P process, acquire them; note that most of these don’t need to be BiC; for example, all of the major suites will tout the tens or hundreds of millions their big customers have “saved” with their solution, but what they won’t tell you is that at least 90% of that savings simply resulted from the client implementing a good process supported by a tool with a decent workflow solution; in other words, you don’t need the multi-million dollar solution (to start), you’ll see the same benefit from a six figure suite that is better than average in the key modules that matter to you (especially since it will take you years to master the new processes it will support, meaning that for a big suite, it’s usually five years or more before you can see more value than just going with a basic solution given that the journey to Best in Class, as determined by Hackett in the mid moughts, is at least eight years)
  • continuous data modelling and analysis, start now; with your spend analysis and performance tool updated at least daily

you need to make a plan to incrementally acquire what you are missing, most critical need first, until you do. (Remember, don’t try a big bang implementation. No matter what the vendor or Big X will tell you, those always end in big booms.)